1. Introduction
To date, the roles of glucose and glutamine in the biology of transformed cells both in vitro and in vivo, have been evaluated in various cellular systems, most often as separate entities. It is well-established that cancer cells need glucose as a source of carbon. They also need glutamine. Not only as an alternative substrate for the Krebs cycle and ATP production, but also as a source of carbon and nitrogen, glutamine is needed for various biosynthetic reactions and glutathione production to support antioxidant defense [
1].Under normoglycemic conditions, the generation of the fundamental metabolite nicotinamide adenine dinucleotide phosphate (NADPH) is secured via the glucose catabolism pathway—the pentose phosphate pathway (PPP). Glucose starvation results in decreased ATP production and could induce oxidative stress by downregulating NADPH production by PPP. Under these conditions, metabolic reprogramming and redox regulation are closely related to the activation of 5′ AMP-activated protein kinase (AMPK) pathway. It is a protective mechanism aimed at prolonging cell survival by preventing excessive NADPH consumption in fatty acid synthesis and increasing NADPH generation in the process of fatty acid oxidation [
2]. When the protective capacity of the AMPK pathway is exceeded, the regeneration rate of glutathione decreases, and there is an increase in ROS, an indicator of the disbalanced cellular redox status. It is well-known that ROS are involved in the nutrient deprivation-induced Warburg effect [
3]. Glutamine has also been shown as a source of NADPH. In pancreatic cancer cells with a K-RAS-regulated metabolic pathway, the glutamine-derived malate can be converted to pyruvate by malic enzyme. This reaction is associated with NADPH generation [
4]. It has also been shown recently that a lack of glutamine may promote rapid and transient activation of AMPK [
5].
During glucose deprivation, activated AMPK phosphorylates the wild type (WT) TP53 at serine 15, leading to G1/S cell cycle arrest and cellular senescence [
6]. WT TP53 has an important role in cellular metabolism. It inhibits the monomeric form of the enzyme glucose-6-phosphate dehydrogenase (G6PD), which is present in the cytoplasm. This event results in one more instance of NADPH depletion. The effect seems to be characteristic of WT TP53, but not its mutant forms. It was proposed as a main function of cytoplasmic WT TP53 in resting cells [
7].
When deprived of oxygen and glucose, the cells activate the AMPK by NAD(P)H:quinone oxidoreductase 1 (NQO1) [
8]. This enzyme was purified and characterized for the first time in 1988 [
9]. It was originally considered only as a flavin adenine dinucleotide (FAD)—dependent, two-electron reductase. There are numerous proofs of its effectiveness, associated with reducing quinones to hydroquinones through a two-electron transfer. The catalytically active form of the enzyme is a homodimeric protein. It has two identical active sites located at the interface between monomers and with one FAD bound per monomer. Each of these two sites is shared by both reduced pyridine nucleotide cofactors, NADH and NADPH [
10]. The model of the NQO1 mode of action (ping-pong-bi-bi kinetic mechanism), proposed in 1974 [
11], is still considered valid. The catalytic cycle is initiated by the binding of reduced pyridine nucleotide in the active site, followed by a hydride transfer to FAD. It leads to a conformational change expelling the oxidized pyridine nucleotide, nicotinamide adenine dinucleotide (NAD+), and creating an environment for quinone binding. The generation of NAD+ makes a strong, functional, yet indirect link between NQO1 and two very important cellular enzymes relevant for metabolism and metabolic reprogramming in cancer. These are NAD+ dependent sirtuin 1 (SIRT1) and PARP-1, a major NAD+-consuming enzyme [
12].
The enzymatic activity of NQO1 can be detected in the cytosol and in the nucleus [
13]. It has an important role in eliminating free radicals [
14] which increase during nutritional stress. According to the most recent data, NQO1 is a central unit of the redox-dependent switch. It depends on NQO1 conformational change, in which NADH has strong protective role against tryptic digestion and loss of the C-terminal NQO1 domain. To a lesser extent, a protective role was also obtained with NADPH [
15].
Altered pyridine nucleotide ratios could induce a switch in protein conformation. This results in binding of NQO1 to a different set of proteins and RNA under oxidative conditions [
16]. Thus, NQO1 action influences the activity of other proteins indirectly, through generating NAD+ (SIRT1, PARP-1) [
17] and through direct binding (hypoxia-inducible factor, alpha subunit, HIF1-α, TP53) [
18,
19].
NQO1 stabilizes both wild-type (WT) [
18] and mutant-types (MT) TP53 protein [
20] by protecting them from the ubiquitin-independent 20S proteasomal degradation. This stabilizing effect is most prominent under oxidative stress. However, the presence of the single nucleotide variation (SNV) rs1800566 that occurs in NQO1 exon 6, strongly decreases the enzymatic activity of NQO1 and abolishes TP53 stability mediated by NQO1 [
21,
22]. This polymorphism, also known as NQO1*2 (heterozygote)/NQO1*2/*2 (homozygote), was shown to be an important factor in a poor clinical response to quinone (mitomycin C, β-lapachone)-based chemotherapy. This is due to a lack of drug bioactivation [
23,
24].
TP53, which is traditionally considered a tumor suppressor, is currently an emerging research topic relating to nutritional stress [
25]. Its connection to NQO1 may be a critical factor for cellular adaptive stress response, especially during nutrient deprivation. The most recent data have shown that the withdrawal of glutamine activates TP53 [
26]. In a glutamine deprived cell, TP53 binds to the promoter of the solute-like carrier family 7, member 3 (SLC7A3). It promotes cancer cell adaptation to glutamine deprivation by upregulating SLC7A3to increase arginine uptake [
27].
The state of oxidative stress is of utmost importance for activating
NQO1 transcription, which is mediated by NFE2L2 (Nuclear Factor, Erythroid 2 Like 2: NRF2). When there is an excess of ROS, NRF2 dissociates from its cytoplasmic partner Kelch-Like ECH-Associated Protein 1 (KEAP-1). It enters the nucleus and binds to the
cis-acting elements in an array of NRF2 target genes called antioxidant response elements (AREs) [
28]. These are present in the
NQO1 promoter [
29]. Consequentially, this event leads to an increased transcriptional activity of the
NQO1 gene. This phenomenon has been shown in various models as a part of a strong antioxidative cellular response.
One very interesting molecular-genetic aspect of
NQO1 mRNA is associated with the deposit of four NQO1 transcript variants (TVs) in the GeneBank. The gene itself contains six exons (
Figure 1). All of them are part of the longest transcript (TV1; NM_000903.3, N = 2521 nt). Another three transcripts are characterized, as follows: TV2: NM_001025433.2; exon 5 excluded (N = 2419 nt); TV3: NM_001025434.2, exon 4 excluded (N = 2407 nt); TV4: NM_001286137.2, exons 4 and 5 excluded (N = 2305 nt).
In 1995, Gasdaska et al. described the
NQO1 transcript lacking exon 4 (TV3) in cancer cell lines SW 480 and HT-29. The existence of the corresponding protein was not confirmed [
30]. Seven years later, it was proposed that the polymorphism present at the end of exon 4, rs1131341 (Arg137Trp, also known as NQO1*3*), has a strong influence on NQO1 splicing. As a consequence, the ratio TV1/TV3 (shown by end-point PCR to be around 9.0 in NQO1*1/*1, NQO1*2/*1, NQO1*2/2*), significantly decreases (TV1/TV3 = 2) [
31]. The ratio of TVs may vary depending on stressful conditions [
30]. This was shown only once, in the mononuclear cells obtained from patients before and at various times following treatment with mitomycin [
32]. According to SwissProt, there is only one experimentally verified NQO1 protein variant which is coded by NQO1 TV1. It consists of 274 amino acids (30.868 kDa).
As recently shown in a yeast model, introns negatively regulate growth in a rich medium. They are clearly required for maintaining cellular viability during the deprivation of nutrients (dextrose and phosphates) [
33]. In 2007, Pleiss et al. showed that two different stress-inducers (ethanol exposure and amino acids deprivation) induce unique splicing profiles. This suggests that in yeast at least two independent pathways connect the spliceosome with the cellular environment [
34].
Alternative splicing was shown to take place during nutrient depletion in an organoid model system derived from murine intestinal epithelial cells. This included exon skipping events and events involving full intron retention (IR-S; intron retention simple) and complex intron retention (IR-C; intron retention complex) [
35].
Based on these facts, we wanted to explore selected cellular parameters (cellular viability and proliferation rate, ROS generation) and molecular events included in the axis NRF2-NQO1/TP53, under two different forms of nutritional stress. The transcripts (quantitatively - NRF2, NQO1 and qualitatively - NQO1 splice variants) and proteins (NRF2, NQO1, TP53) in cytoplasmic and nuclear cellular fractions were validated. Three cell lines originating from the head and neck squamous cell carcinomas (HNSSC) were used: FaDu; Cal 27; and Detroit 562. These cells significantly differ with respect to basal NQO1 activity (FaDu > Detroit 562 > Cal 27) [
36]. IMR-90 fibroblasts, which are considered as NQO1 non-expressing cell lines [
37], were used as representative of an untransformed cell line.
2. Materials and Methods
2.1. Cell Lines and Cell Culture Conditions
The cells originating from metastatic pharyngeal cancer (pleural effusion-Detroit 562) and human fetal lung fibroblasts (IMR-90) were purchased from Sigma-Aldrich (St. Louis, MO, USA). The human tongue squamous carcinoma cells (Cal 27) and human hypopharyngeal squamous carcinoma cells (FaDu) were purchased from the American Type Culture Collection (ATCC, LGC Standards GmbH, Wesel, Germany). The cells were cultured in T75 cell culture flasks (Sarstedt AG&Co.KG, Nümbrecht, Germany), in Dulbecco’s Modified Eagle’s Medium (DMEM, D5796; Sigma-Aldrich, St. Louis, MO, USA), supplemented with a 10% fetal bovine serum (FBS, Sigma-Aldrich, St. Louis, MO, USA), without antibiotics, at 37 °C in a humidified atmosphere and in the presence of 5% CO2. Prior to the experiments, the cells were harvested with 0.25% (w/v) Trypsin-0.53 mM EDTA (Ethylenediaminetetraacetic acid) solution and counted with the trypan blue exclusion assay in Bürker-Türk hemocytometer (Brand, Wertheim, Germany). For the experiments performed, the cells were cultured in DMEM, under the following four nutritional conditions (NCs) with respect to glucose and glutamine: NC1-high glucose (4.5 g/L) with l-glutamine (0.584 g/L) (D5796; Sigma-Aldrich, St. Louis, MO, USA); NC2-low glucose (1 g/L) with l-glutamine (0.584 g/L) (D6046, Sigma-Aldrich, St. Louis, MO, USA); NC3-no glucose and no glutamine (A14430, Gibco, Life Technologies Corporation, Grand Island, NY, USA); NC4-no glucose (A14430), but with 0.584 g/L of l-glutamine (Sigma Aldrich, St. Louis, MO, USA).
2.2. Cell Viability Assay
Cellular viability was measured using EZ4U assay (Biomedica, Vienna, Austria), which assesses cellular viability through reducing tetrazolium salts to colored formazan derivatives in the mitochondria of living cells. The cells were seeded in 96-well plates (TPP, Trasadingen, Switzerland) at a density of 1×104 cells per well and cultivated in 200 μL of previously described media formulations (NC1-NC4), supplemented with 10% FBS. After a cultivation period of 48 hours, 20 μL of the dye substrate (tetrazolium salts) was added to each well. After a 2 h incubation, formazan derivatives were quantified by measuring the absorbance using the microplate reader Multiskan EX (Thermo Electron Corporation, Shanghai, China) at 450 nm, with 620 nm as a reference wavelength. Cellular viability under tested conditions was expressed as a percentage of the viability of the control cells (cells grown in a high glucose + glutamine, NC1, medium).
2.3. Cell Proliferation Assay
The rate of cellular proliferation was estimated through incorporating pyrimidine analogue BrdU (5-bromo-2′-deoxyuridine), in place of thymidine, into the DNA of proliferating cells, using the Cell Proliferation ELISA, BrdU (colorimetric) Kit (Roche Applied Science, Mannheim, Germany). The antibody conjugated anti-BrdU-peroxidase binds incorporated BrdU. The complex BrdU/anti-BrdU-peroxidase was detected by the reaction between peroxidase conjugated to the BrdU antibody and the substrate (3,3',5,5'-tetramethylbenzidine). After reaching a satisfactory color intensity (after incubating between 5 and 30 min), the reaction was stopped with 1 M H2SO4 solution.
The cells were seeded in 96-well plates (TPP, Trasadingen, Switzerland) at a density of 1×104 cells per well and were maintained in 200 μL of previously described media formulations (NC1-NC4), supplemented by 10% FBS. After 48 h of incubation, the assay was performed according to the manufacturer's instructions. The reaction product (3,3',5,5'-tetramethyl-benzidine diimine) was quantified by measuring absorbance using a microplate reader Multiskan EX (Thermo Electron Corporation, Shanghai, China) set at 450 nm (reference wavelength: 620 nm). Cell proliferation was expressed as a percentage of the cells grown under condition NC1 (high glucose + l-glutamine medium).
2.4. ROS Measurement
The intracellular levels of reactive oxygen species (ROS) were detected by DCFH-DA (2ʹ,7ʹ-Dichlorofluorescin Diacetate; Sigma-Aldrich, St. Louis, MO, USA). The cells were seeded in white 96-well plates (Thermo Fisher Scientific, Nunc A/S, Roskilde, Denmark) at a density of 1×104 cells per well and maintained in 200 μL of the previously described media formulations (NC1-NC4), supplemented by 10% FBS. After growing for 48 hours, the cells were incubated with 20 μL of 100 μM DCFH-DA, which was added to the culture media. After 45 min of incubation, the medium containing DCFH-DA was replaced with 200 μL of fresh medium. The fluorescence intensity was measured immediately (zero point) and after one hour, on a plate reader Infinite 200 PRO (Tecan Group Ltd., Männedorf, Switzerland). The excitation/emission wavelengths for DCFH-DA were set at 500/529 nm. The values of the emitted fluorescence were expressed as arbitrary units, which represent the difference between the two points of measurement (one hour and zero point). Additionally, the values were corrected with respect to the cell numbers, which varied in relation to the treatment applied.
2.5. Nucleic Acids Extraction
The cells were cultured for 48 h at a density of 1×106 in T25 flasks (Sarstedt AG&Co.KG, Nümbrecht, Germany), in 5 mL of the previously described media formulations (NC1-NC3), supplemented by 10% FBS.
The total RNA was extracted from the cells cultivated and treated in 25 cm2 flasks (Sarstedt AG&Co.KG, Nümbrecht, Germany). The medium was removed and extraction was performed by TRIzol (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions. The integrity of isolated RNA was determined by electrophoresis, on 1% agarose gel stained with ethidium bromide (EtdBr) (Sigma-Aldrich, St. Louis, MO, USA). As there were no issues relating to the integrity of the extracted RNA, all samples were further purified with gDNA Removal Kit (Jena Bioscience, Jena, Germany), according to the manufacturer’s instructions. The concentration and purity of extracted RNA was determined spectrophotometrically (BioSpec-nano, Shimadzu Biotech, Japan) by measuring the absorbance at the following wavelengths: 230, 260 and 280 nm. The samples were stored at −80 °C.
The genomic DNA was extracted by phenol-chloroform extraction, after an overnight incubation with Proteinase K (QIagen, Holden, Germany), as previously described [
38]. After successful precipitation, the samples of extracted DNA were re-suspended in TE buffer (10 mM Tris, 1 mM EDTA, pH 7.4). The concentration and quality of the extracted DNA was determined spectrophotometrically and electrophoretically, in 1% gel agarose stained with EtdBr. The samples were stored at +4 °C.
2.6. Construction of Primers
For all primers used in this research, with the exception of GAPDH1/GAPDH2 which are commonly used, the modeling through combining the programs Primer-BLAST and Primer3Plus were performed.
Table 1 shows the primer sequences, their exact position on the RefSeq and the expected amplicon sizes.
The composition of the nucleotides of the primers used allowed the authors to perform the polymerase chain reaction under almost identical conditions. The primers for gDNA were selected to anneal to the template at 58 °C, while the primers for cDNA annealed to the template at 59 °C.
The three primers for GAPDH were constructed in a way which allowed combining the primer GAPDH2 with primers GAPDH1—for determining cDNA quality and GAPDH3—for a rigorous check of the potential gDNA contamination. These strict precautionary measures were undertaken because the TaqMan probe used for quantifying NRF2 may bind to the gDNA, at least according a statement provided.
The selection of primers which would allow for the amplification of all four NQO1 TVs in one reaction was based on the primary structure of the NQO1 TV1 mRNA (
Figure 1). The primers were complementary to the stretch of nucleotides positioned in the 3′ region of the exon 2 (NQO1F) and 5′ region of the exon 6 (NQO1R), respectively. All primers which allowed for examining the exon/intron boundaries sequences were used only on genomic DNA.
2.7. Reverse Transcription, RT – PCR, and PCR
The reverse transcription was performed with a High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA), with anchored Oligo(dT)23 primers (Sigma-Aldrich, St. Louis, MO, USA) and 1 µg of total RNA in a 20 μL volume, according to the manufacturer’s instructions. The reaction conditions were: 25 °C/10 min; 37 °C/120 min; 85 °C/5 min; 4 °C/indefinite. After finalization of the reverse transcription, 80 μL of sterile, deionized water was added to the tubes to achieve a total volume of 100 μL of cDNA, which was used for subsequent reactions.
The efficacy of reverse transcription was assessed with the end-point polymerase chain reaction (PCR) using the primer pair GAPDH 1/GAPDH 2 and 1 μL of diluted cDNA. This template volume was used as a standard in all end-point PCR reactions. For discovering the potentially present traces of contaminating DNA, the primer pair GAPDH2/GAPDH3 was used, as the sequence of the GAPDH3 primer is complementary to the nucleotides in intron 5. The polymerase chain reaction was carried out in GeneAmp PCR System 2400 (Applied Biosystems, Foster City, CA, USA). The reaction mixture (12.5 μL) contained AmpliTaq 360 Gold Master Mix and GC Enhancer (Thermo Fisher Scientific, Waltham, MA, USA), home-made nuclease free-water and primers (final concentration: 400 nM).
The genomic DNA was amplified with the same sets of chemicals and in the same volume, with 200 ng of gDNA. The reaction conditions were: Predenaturation 95 °C/5 min, followed by 35 cycles: 9 °C/30 s; 58 °C and 59 °C for gDNA and cDNA, respectively/30 s, 72 °C/30 s. The final elongation was at 72 °C, for 7 min.
2.8. Densitometry, Purification of PCR Products from Agarose Gel and Sequencing
The PCR products were separated by electrophoresis in 2% agarose gel and photographed for future densitometric analysis, which was performed with ImageJ [
39]. Bands of interest were cut out of the gel and purified using a GenElute Gel Extraction Kit (Sigma-Aldrich, St. Louis, MO, USA), according to the manufacturer’s instructions. After being eluted from the column, 10 µL of eluate was loaded in 2% agarose gel in order to determine the purity and amount of eluate that was going to be used in the sequencing reaction. Only the amplicon corresponding to TV4 needed to be purified, re-amplified one more time with NQO1F/NQO1R and purified again. A sufficient amount of the template was obtained for the sequencing reaction only after this additional procedure.
The sequencing reaction contained 12 µL mixture of the purified PCR product (approximately 50 ng per 100 bps), nuclease-free water and 1 µL of the forward and reverse primer used in PCR, respectively (primer concentration 3.2 pmol/µL), for a total volume of 13 µL. The samples were sequenced at the DNA Sequencing Core Facility of the Rudjer Boskovic Institute.
2.9. Real-Time RT-PCR and Rationale for Specific TaqMan’s Probe Selection
For quantification of NQO1 and NRF2, TaqMan chemistry was used. The reactions were performed in a 7300 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). We used always 1.5 μL of cDNA template, 10 μL of TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific, Waltham, MA, USA), 1 μL of the probe and 7.5 μL of sterile, deionized water. The reactions were performed in triplicate for each template and for each probe and in at least three biological replicates, which we tried to associate with three consecutive passages. Microamp 96-well rxn plates (Thermo Fisher Scientific, Waltham, MA, USA) were used, and amplification was performed under the following conditions: Incubation 50 °C/2 min + 95 °C/10 min, followed by 40 cycles; 95 °C/15 s, 60 °C/60 s. The following probes were used for house-keepers: GAPDH—Hs99999905_m1 (as GAPDH was used in the end-point PCR reactions); TBP—Hs00920495_m1 (as the TBP was used in the WBs); HPRT1—Hs02800695_m1 (as it was shown to be relatively stable in some published studies) [
40].
For relative quantification, delta-delta Ct 2–∆∆Ct developed by Applied Biosystems was used, which presumes identical amplification efficiencies of the target and reference gene and the Pfaffl method [
41], which served as a correctional factor in subsequent calculations.
2.10. Protein Extraction and Western Blot Analyses
The cells were cultured for 48 hours at a density of 1×10
6 in T25 flasks (Sarstedt AG&Co.KG, Nümbrecht, Germany), in 5 mL of the previously described media formulations (NC1-NC3), supplemented by 10% FBS. The proteins were extracted with NE-PER nuclear and cytoplasmic extraction reagents (Thermo Scientific -Pierce Biotechnology, Rockford, IL, USA), supplemented by the protease inhibitor (Complete Mini Protease Inhibitor Cocktail Tablets; Roche Applied Science, Mannheim, Germany). The protein amount was estimated using the Bradford method [
42]. The absorbance was measured at 595 nm using the microplate reader Multiskan EX (Thermo Electron Corporation, Shanghai, China). The protein samples were mixed with Laemmli buffer and boiled for 5 min at 95 °C. The equal amounts of protein (10 μg) were loaded on the gel (9% resolving and 5% stacking), separated electrophoretically and transferred to nitrocellulose membranes (Roti®-NC, Carl Roth, Karlsruhe, Germany). The transfer efficacy was evaluated by staining the membranes with Ponceau S solution (Sigma Aldrich, St. Louis, MO, USA). The stained membranes were scanned. After incubating the membranes with 5% nonfat milk (Cell Signaling Technology, Danvers, MA, USA) in Tris-buffered saline (TBS; 50 mM Tris-Cl, 150 mM NaCl, pH 7.6) containing 0.1% Tween-20 for 1 h, the membranes were probed overnight with the following primary antibodies: (all rabbit monoclonal, Cell Signaling Technology, Danvers, MA, USA): anti-NQO1 (1:1000; CST: #62262); anti-NRF2 (1:1000; CST:#12721); anti-TP53 (1:1000; CST:#2527); anti-TBP (1:1000; CST:#44059); anti-β-actin (1:1000, CST:#8457). The last two antibodies were used as the loading controls for nuclear and cytoplasmic fractions, respectively. The expected molecular weights of the detected proteins were: NQO1—29 kDa; NRF2—97–100 kDa; p53—53 kDa; TBP 35–45 kDa; β-actin 45 kDa. After three washings of the membranes with TBST (0.1% Tween 20 in 1× TBS), the immunoreactive bands were detected with an HRP- linked anti-rabbit IgG secondary antibody (1:2000; CST: #7074). The immunological complexes were visualized using SuperSignal
TM West Pico PLUS Chemiluminescent Substrate (Thermo Scientific, Rockford, IL, USA) and Alliance 4.7 (UVITEC, Cambridge, UK). The protein expression levels were quantified using ImageJ and/or Image Studio Lite (LI-COR, Lincoln, NE, USA) analysis software. The relative change of signals obtained was calculated after normalization according to the loading controls and Ponceau S signals.
2.11. Statistical Analyses
Each experiment related to cellular biology (viability, proliferation, ROS generation) was performed in technical triplicates or quadriplicates and repeated three times (as specified in the Figure legend). The data obtained was analyzed with 1-way ANOVA and Tukey post-hoc test, as indicated in the figure legends. The same principle was applied for producing and analyzing the data obtained with molecular biology methods in biological triplicates. For both analyses and visualization, GraphPad 6.0 was used. The statistical significance of the differences obtained for all data analyzed was considered significant at p < 0.05.
4. Discussion
The lack of nutrients leads to cellular oxidative responses. We wanted to explore the NRF2-NQO1 axis by inducing stress through incubating cancer cell lines in the following: A medium with a decreased concentration of glucose (1 g/L) (NC2); a medium with a complete lack of glucose (NC4) but supplemented with glutamine; and, in order to induce the strongest level of ROS, by creating an artificial situation, using a medium deprived of glutamine and glucose (NC3). However, due to trace amounts of glucose and glutamine in FBS, these cells were still exposed to a minimal concentration of glucose (~0.125 g/L ≈ 0.69 mM) [
46] and glutamine (0.05 mM) [
47].
The cellular viability, ROS generation and rate of cellular proliferation, were measured and compared under these four conditions, in three cancer cell lines, originating from HNSCC: FaDu, Detroit 562 and Cal 27. According to COSMIC and published papers, these three cell lines can be classified according to their TP53 mutation status [
48]: 1. FaDu—heterozygous mutation leading to substitution of arginine with leucine at aa248 (R→L); 2. Cal 27—mutation (unknown zigosity status) leading to substitution of histidine with leucine at aa193 (H→L); 3. Detroit 562 mutation (unknown zigosity status) leading to the substitution of arginine with histidine at aa 175 (R→H). All these mutations are in the DNA-binding domain of the TP53 protein. Based on the cellular effects observed in the cancer cells, molecular-genetic experiments were chosen to be performed under conditions NC2 and NC3, to which fetal lung fibroblasts IMR-90 were also exposed. This cell line was used to show how MYC, after being transduced in IMR-90, makes IMR-90 addicted to glutamine [
49]. Terashima et al. were able to stimulate NRF2 entering in the nucleus of these cells under the low glucose condition [
50]. Thus, it was assumed that IMR-90 is likely to be sensitive to NC2 [
49]. Many research groups were not able to show the expressed NQO1 protein in these cells in their native state [
36]. However, there are also data showing the expressed, mature NQO1 in IMR-90 [
51].
4.1. Cancer Cells and Fibroblasts IMR-90 Differ in Their Sensitivity to Nutrient Deprivation
We were aware that the chosen conditions were likely to reduce cellular viability and increase the level of ROS, but we could not predict to what extent that would happen. What was visible immediately was that the viability of the cancer cells under NC2 decreased significantly, but far less than under NC3. With respect to NC2 and the cancer cell lines, our results were in the strongest agreement with results published by Terashima et al., who presented a similar decrease of HepG2 viability cultured in a medium with low glucose (1 g/L) [
50] In IMR-90 only, the NC2 condition had extremely strong effects with respect to the examined cellular parameters. In IMR-90 only, the effect of the NC2 condition was much stronger than in the cancer cell lines, confirming that IMR-90 are primarily dependent on glucose. The result obtained on IMR-90 is in agreement with data presented by Yunova et al., who also convincingly showed the sensitivity of IMR-90 to glucose deprivation [
49]. With respect to glutamine, van den Heuvel et al. showed the insensitivity to glutamine depletion of normal human lung fibroblasts (NHLF)) cells originating from lung fibroblasts [
52].
As shown on
Figure 2A,D, among all four tested cell lines exposed to NC2, the viability of fibroblasts decreased below 50% (app. 80% in cancer cell lines). It did not decrease further when there was a lack of both glucose and glutamine (NC3). One possible explanation of this effect may be the high expression of glutamate/cystine antiporter solute carrier family 7 member 11 (SLC7A11, also called xCT) in IMR-90 cells [
53]. In cancer cells, SLC7A11 mediates the efflux of intracellular glutamate, thereby rendering them metabolically less adaptable and more reliant on glucose for survival [
54]. The drop in cellular viability of all cancer cell lines, contrary to IMR-90, was less pronounced under NC2 than under NC3. The proliferation rate of FaDu under NC4 was far below the proliferation rate in Cal 27 and Detroit 562. This indicates that there is a differential sensitivity to glutamine between FaDu and the other two cancer cell lines, when there is a lack of glucose.
Both FaDu and Cal 27 cells have been reported to express a high level of SLC7A11 [
55]. The relative unresponsiveness of FaDu observed under NC4 may be explained, at least in part, by the observed high increase of nuclear NRF2 (
p = 0.0414) (
Figure 10F), joined with increased
NRF2 transcription and the significantly increased generation of ROS (
p < 0.0001) (
Figure 2C), in response to NC2. Considering that NRF2 has been reported to induce
SLC7A11 expression in response to glucose starvation [
51], it is conceivable that FaDu cells, in which exposure to NC2 upregulated NRF2, would under NC4 (complete lack of glucose, albeit supplemented with glutamine) have a more pronounced expression of
SLC7A11 and efflux of glutamate that would prevent the recovery of cellular proliferation. Cal 27 cells, which were also reported to express a high level of
SLC7A11 [
55] but did not respond to NC2 by statistically significant upregulation of nuclear NRF2, significantly recovered cellular proliferation under NC4NC4), as contrasted with FaDu cells.
Under NC3, FaDu was the only cancer cell line which had a significant increase of NQO1 in the nucleus (NC2 vs. NC3 p = 0.0178). However, at the transcriptional level, significant changes relating to NQO1 transcriptional activity in FaDu were not detected, neither under NC2 (accurately measured), nor under NC3 (approximated).
Cal 27 was the cell line with the lowest basal level of NRF2 (
Figure 10A,B, NC1—nuclear fraction). Probst et al. were able to show that the cell lines with high basal NRF2 activity exhibited little or no increase in
NQO1 mRNA levels following NRF2 activation with the compound RTA 405 [
56]. We think this fact can explain part of our results obtained on NQO1 transcripts.
4.2. Importance of Defining the SNVs, in a Given Experimental Model
When
NQO1 transcripts were quantified, a significant increase of the
NQO1 transcript was shown only in IMR-90 and Cal 27 but not in FaDu (under NC2) and Detroit 562. Unfortunately, however high was the expression in IMR-90, there was a minimal amount of NQO1 protein. The level of the
NQO1 transcript in IMR-90 was strong and it did not differ from the mRNA NQO1 signal in the cancer cell lines. A majority of research groups had not detected NQO1 protein in IMR-90 (at least not under restful conditions). However, it seems that nobody reported the data on the IMR-90
NQO1 genotype. After performing the DNA sequencing, it was concluded that this cell line is a NQO1*2/*2 homozygote (rs180566). Some inconclusive data obtained on another human fibroblast cell line, WI-38, encouraged us to analyze it in the same way. It was shown that WI-38 also bears rs180566 but is a heterozygote (NQO1*1/*2) cell line. Thus, at least with respect to NQO1, IMR-90 and WI-38 are not normal fibroblasts. Accordingly, any conclusion obtained on one cell line cannot be automatically translated to another (or any other) cell line. Regarding rs180566, detailed modeling performed by Lienhart et al. [
43] did not provide a structure-based explanation for the lower enzymatic activity of NQO1 P187S. A plausible explanation given in an in vivo model by Tsvetkov et al., should be considered in light of discovering that E3 ligase STUB1/CHIP (C terminus of Hsc70-interacting protein) regulates the NQO1 protein level through ubiquitination and degradation [
57]. The heterozygote P187S (rs180566) was shown to be a stronger STUB1 interactor with an increased susceptibility to ubiquitination by the E3 ligase STUB. Thus, we concluded that the homozygote, S187S, may have even stronger affinity for STUB1. This finally resulted in an almost undetectable NQO1 in IMR-90, notwithstanding the increase of
NQO1 transcriptional activity (
Figure 3;
Figure 4) under NC3 which was associated with a physiological increase of nuclear NRF2 (
Figure 10B). An increased sensitivity of IMR-90 to NC2 may be a consequence of lacking a strong mechanism for influencing the NAD(P)+/NAD(P)H redox balance during the stress-related events, due to catalytically insufficient NQO1.
4.3. FaDu Proliferative Potential with Respect to GLS1 and Decreased Sensitivity to Glutamine When Deprived of Glucose
It was very interesting to see that that the effect of glutamine on FaDu, with respect to proliferative potential when deprived of glucose, exerts a less positive effect than with Cal 27 and Detroit 562 (
Figure 2B). The enzyme, Glutaminase 1 (GLS1), enables malignant cells to undergo increased glutaminolysis and utilization of glutamine as an alternative nutrient. A recently published study has clearly shown that the expression of
GLS1 in FaDu is far less prominent than in Detroit 562 cells [
58]. Thus, the capacity of FaDu for glutamine utilization seems to be constitutively decreased. Although we did not measure
GLS1, we hypothesized that the less prominent rescuing effect of glutamine on FaDu than on Detroit 562 and Cal 27, in the absence of glucose, occurs as a result of low basal expression of
GLS1 In IMR-90, GLS1 is also present at a low level and its pharmacological inhibition does not change the level of intracellular ATP [
59]. Sandulache et al. discovered that a majority of HNSCC cancer cell lines show a dependence on glucose and not glutamine [
60]. This study also referred to FaDu. However, FaDu, as contrasted with all of the other 14 tested HNSCC (Detroit 562 and Cal 27 were not included in the panel), did not exert a similar rate of sensitivity to non-metabolizable D-glucose analogues. The data was: IC
50 10.90 (FaDu) versus IC
50 0.79 with UMSCC22B—the most sensitive cell line.
4.4. NQO1 and Its Splice Variants—TaqMan Probes Validation
Regardless of the conditions applied, three
NQO1 transcript variants were detected. Based on WB analyses, only the longest one, TV1, as compared with the other two, TV3 and TV4, seemed to translate in the mature protein. The change of the ratio TV1/TV3 was measured. We have shown that only in Cal 27, under NC3, did this ratio increase. It decreased in the other three cell lines. It is unfortunate that the
NQO1 transcript fold change under NC3 in FaDu could not be accurately measured (
Figure 3F), where there was a significant decrease of the TV1/TV3 ratio (
Figure 6). Based on the Ct values, we estimated that the
NQO1 transcription, under NC3, does not significantly vary from the control condition (NC1) (
Figure 3). The change of TV1/TV3 ratio was not influenced by rs1131341 [
31], as all cell lines had the same genotype with respect to this SNV. Thus, we concluded that the change TV1/TV3 depends on both conditions applied and the cell-type specific response, which clearly differentiated Cal 27 from other cell lines [
61]. It is possible that, through this mechanism, this cell line compensates NQO1 basal activity, which is significantly lower (90 U) than the NQO1 activity in Detroit 562 (790 U) and FaDu (1400 U) [
36]. It also raises a question with respect to compensatory mechanisms associated not only with the level of transcription, but also with splicing, when there is a high ROS generation. Cal 27, the cell line with the lowest NQO1 activity under basal conditions, had the strongest increase of NQO1 transcript and adjusted the TV1/TV3 ratio in favor of TV1. FaDu, the cell line with the highest NQO1 activity under basal conditions, had no increase of NQO1 transcript and reduced the TV1/TV3 ratio.
4.5. Activation of NRF2 and Sensitivity to Glutamine
In the Results section, we presented how some cellular parameters are indicative of a different sensitivity to glutamine, which is higher in Cal 27 and Detroit 562, and lower in FaDu. How is that to be explained, in addition to the previously given explanation relating to the cellular transporters? First, it should be noted that the intensity of the nuclear and cytoplasmic NQO1 signal was far less prominent in Cal 27 under NC1, than in the other two cancer cell lines (
Figure 10A,B). The same phenomenon was shown in Li’s paper [
36]. The reason for less NQO1 protein is not the presence of rs18066, as Cal 27 are NQO1*1/*1. One possibility may be that these cells have a genuinely less active NRF2-KEAP1 pathway. Indeed, Romero et al. have suggested NQO1 as a suitable biomarker for NRF2 activation, when researching a human KRAS-mutant lung adenocarcinoma (LUAD) [
62]. In our experimental model, this was not shown. However, one can speculate that the NRF2-KEAP1 axis, weaker in Cal 27 than in FaDu and Detroit 562, can be strongly activated. Based on the results presented, Cal 27 cells indeed have a stronger potential for activating the NRF2-NQO1 axis than the other two cancer cell lines. The activation of the NRF2-KEAP1 axis, which was shown to be, according to the majority of the parameters measured, highly dependent on glutamine [
63], needs to be explored further in dynamic and not end-point experiments.
4.6. TP53 and Its Potential to Influence Phenomena Observed
Regarding the influence of MT TP53 on NRF2, there are many data, but there are no conclusions. Lisek et al. showed that mutant TP53 increases NRF2 localization to the nucleus of cancer cells, where it redirects NRF2 to ARE elements of specific genes to activate their transcription. Conversely, it sequesters NRF2 from other targets, leading to their downregulation [
64]. Kalo et al. have shown that induction of stress in HCT116 cells bearing TP53 mutant R273H results in NRF2 nuclear accumulation. However, the transcription of target genes was induced to a much lesser extent than in HCT116 without TP53 activity (TP53−/−). They also showed that the down-regulation of endogenous mutant TP53, results in increased mRNA levels of NQO1 and Hem Oxygenase (HMOX-1). Thus, they proposed that MT TP53 promotes the survival of cells with high level ROS [
65]. Under NC3, a decrease of nuclear TP53 in Cal 27 is related to an increased transcription rate of
NQO1. At the same time, the strong increase of TP53 in the nuclei of FaDu under NC3 may be associated with its silencing effect on
NQO1 transcription (as estimated). This would be in accord with Kalo’s results, at least at the level of transcriptional activity of only
NQO1. When we measured the transcriptional activity of
HMOX-1, a tremendous transcriptional activity of
HMOX-1 in IMR-90 and FaDu under NC3 was recorded (data not shown). This does not necessarily mean that this increased transcription relies on NRF2, as we need to perform chromatin immunoprecipitations and selective silencing in order to understand the molecular mechanisms involved.
One important aspect of the events which seems connected to accumulating phenomena related to nutritional stress and TP53, may be on an entirely different level. The ubiquitin ligase Mdm-2, which mediates TP53 degradation in the proteasome [
66], is a transcriptional target of TP53 [
67]. Qie et al. [
68] reported the upregulated transcription of MDM2 in Hep3B cells cultured in a glutamine free medium, despite the homozygous deletion of TP53, pointing to the existence of alternative regulatory mechanisms. Considering the role of MDM2 in TP53 degradation, if the reported increase in MDM2 transcripts translates to an increased protein level, one would expect a decrease in the TP53 upon glutamine deprivation in TP53 expressing cells. Although it is seemingly surprising, a redistribution of TP53 among cellular compartments (cytoplasm and nucleus) was observed, rather than its decrease in the cells deprived of glutamine (NC3;
Figure 10B). A possible explanation for this effect may be the nuclear retention of TP53 due to its poly(ADP-ribosyl)ation that prevents TP53 interaction with the nuclear export receptor CRM1 [
69]. Chiodi et al. recently reported that glucose and/or glutamine deprivation causes very rapid PARP-1 activation and protein poly(ADP-ribosyl)ation [
70]. This is consistent with the intracellular distribution of TP53 that was observed in all the tested cells grown in a low glucose medium (NC2) or glucose- and glutamine-free medium (NC3) (
Figure 10C). Namely, in all cell lines, TP53 was less abundant in the cytoplasm of the cells grown under NC2 and almost completely absent in the cytoplasm of the cells simultaneously deprived of glucose and glutamine (NC3). Contrary to cytoplasm, a significant amount of TP53 was present in the nuclear fraction. The retention of TP53 in the nuclei of all cells exposed to NC2 and NC3, regardless of their TP53 mutational status, is consistent with the report that three amino acids (E258, D259 and E271) are the targets of poly(ADP-ribosyl)ation, and that TP53 failed to get poly(ADP-ribosyl)ated only when all three of them were replaced with alanine [
69]. Therefore, poly(ADP-ribosyl)ation can retain both mutant and WT TP53 in the nucleus.
The importance of NQO1 activity for TP53 accumulation in the cell strongly argues in favor of the involvement of poly(ADP-ribosy)lation in TP53 stabilization. This is because it was shown that the inhibition of NQO1 activity by dicumarol induces proteasomal degradation of WT and MT p53 [
18]. The enzymatic activity of NQO1 is needed for generating NAD
+, which is a co-substrate for PARP-1 that transfers ADP-ribose moieties from NAD
+ to proteins including TP53 [
71]. Experimentally induced PARP-1 or NAD
+ deficiency has been reported to result in a significantly reduced level and activity of TP53 [
72]. Therefore, the accumulation of TP53 in the nuclei of the cells exposed to glucose and glucose/glutamine deprivation (NC2 and NC3), may be partly mediated by NQO1 (and other oxidoreductases like WOX1 [
73]. Asher et al. [
18] suggested that, considering the presence of several putative TP53-binding elements in
NQO1 promoter,
NQO1 may belong to TP53-inducible genes involved in a positive autoregulatory loop that regulates the level of TP53. Therefore, the mutational status of TP53 may have a profound influence on NQO1 expression in cells that are exposed to nutritional stress.
5. Conclusions
After modulating FaDu, Cal 27 and Detroit 562 for the vital cellular parameters of viability, proliferation and generation of ROS while cultivating them under four different nutritional conditions (NC1-NC4), some general conclusions can be drawn: (a) In relation to all three parameters analyzed, these cell lines showed sensitivity to glucose deprivation; (b) when having available minimal amounts of glucose and glutamine (NC2) FaDu, Cal 27, Detroit 562 responded strongly with respect to all three parameters; (c) only FaDu cells showed an increased need for glucose, not glutamine (NC3 versus NC4), for sustaining replication activity. A strong increase of ROS influences the NRF2-NQO1 axis in these cells in a fashion which is apparently cell-type specific. When considering the activation of the axis through NRF2 nuclear accumulation, the strongest response under a milder condition (NC2) was recorded in FaDu, associated with a decrease of the nuclear TP53 signal. Under harsh conditions (NC3), Cal 27 and Detroit 562 responded with NRF2 nuclear accumulation, associated again, with a decrease of the nuclear TP53 signal. Obtaining the same phenomena under different conditions (Cal 27 and Detroit 562 versus FaDu) pointed out the differences in response to identical stress, which correlates with the fact that only FaDu cells did not recover their replicative potential when deprived of glucose, in the presence of glutamine.
When considering the activation of the axis through an increase in NQO1 transcription, only Cal 27 responded adequately, through an increase in the NQO1 transcription rate and a modulation of alternative splicing, in favor of TV1. FaDu responded in an entirely different fashion, with a decrease in NQO1 transcript and a modulation of alternative splicing, in favor of TV3. These responses may be consequential with respect to NQO1 enzymatic activity in Cal 27 and FaDu, which was previously shown to be 15 times higher in FaDu. Thus, the whole response of the NRF2-NQO1 axis to stress should be considered in the broader context of the cellular background.
Detroit 562 is the cell line which moderately activated its NRF2-NQO1 axis, on both the transcriptional and protein level. It is the only cancer cell line which had no significant increase of ROS, under NC2.
Fibroblasts IMR-90 were entirely dependent on glucose. These cells exhibited a physiological cellular response relating to the activation of NRF2-NQO1 axis during nutritional stress, which resulted with hardly detectable NQO1 signals when compared to the cancer cell lines. IMR-90 are homozygous – NQO1*2/*2, with respect to rs1800566.
Without making the genotyping in respect to rs1800566, we would still be confident that IMR-90 indeed are NQO1 non-expressing cells. According to all available data, their rs1800566 genotype directs an extremely high rate of the NQO1 protein degradation, although the NRF2-NQO1 axis in these cells activates during nutritional stress.
Thus, when making important conclusions on the strength of the NRF2-NQO1 axis through NQO1 protein level/enzymatic activity, the status of rs1800566, as well as specifics of the cellular background, are always relevant and must be considered.