Influence of oxygen availability on expression of glutaminolysis genes in human colon cancer cells

Colorectal cancer (CRC) is a malignant epithelial tumor that develops in the colon, rectum, or vermiform appendix and may spread to the liver. It is the third most common type of cancer in Poland, and the most widespread gastrointestinal tract cancer in the world [1]. Due to the aggressiveness of CRC and the lack of targeted therapies, it seems reasonable to investigate the key metabolic pathways involved in cancer development and progression.

Metabolic adaptation to rapid environmental changes is essential for cancer cell survival and proliferation [2]. The tumor milieu is characterized by a high variability of oxygen and nutrient supply, resulting from rapid proliferation and dynamic processes such as necrosis and angiogenesis [3]. Hypoxia leads to an increase of hypoxia-inducible factor 1 (HIF-1) transcription factor activity, which stimulates the transcription of glucose transporters and glycolytic enzyme genes, resulting in elevated lactate synthesis with simultaneous inhibition of mitochondrial pyruvate metabolism [4, 5, 6, 7]. Under normoxic conditions, however, lactate is produced not only by glycolysis but also by a glutaminolysis process. It is known that both processes are stimulated by the c-Myc transcription factor [8, 9, 10]. Most cancer cells produce large amounts of lactate even under sufficient availability of oxygen. This phenomenon is called the Warburg effect [11, 12].There may be differences in mitochondrial metabolism between the cells of primary and metastatic cancer, as metastases exhibit better environmental adaptation and more dynamic growth. In order to increase the rate of proliferation at hypoxia, cancer cell metabolism is adapted to facilitate the uptake of mainly glucose and glutamine for energy production and macromolecule synthesis needed for cell growth. It is known that cancer and normal rapidly proliferating cells exhibit a high glycolytic rate, however, positron emission tomography (PET) using fluorine-labeled deoxyglucose (¹⁸FDG) as a tumor marker showed low effects of targeting glycolysis in cancer treatment [13, 14, 15, 16]. The latest studies indicated that rapidly proliferating cells (neoplastic and normal) depend also on glutaminolysis, which is another metabolic pathway providing energy and substrates for nucleic acids, proteins, and lipids synthesis [17, 18].

It is known that under routine culture cell conditions, i.e. at 21% atmospheric normoxia, glutaminolysis is conducted by both mitochondrial enzymes such as - glutaminase (GLS; the first enzyme in glutaminolysis), aspartate aminotransferase 2 (AST2) and several Krebs cycle enzymes and cytosolic enzymes such as ATP-citrate lyase (ACL) and aspartate aminotransferase 1 (AST1).

Under these conditions, mitochondrial glutaminase converts L-glutamine into L- glutamate, that is transaminated by AST2 to α-ketoglutarate (α-KG) - one of the Krebs cycle intermediates. Next, α-KG is converted to citrate that can be metabolized by the Krebs cycle enzymes or transported into cytosol, where it is splitted by ACL into acetyl-CoA and oxaloacetate, that can be transaminated by AST1 to L-aspartate (L-Asp) (figure 1).

Figure 1

Recently, two alternative pathways of glutaminolysis (cytoplasmic and mitochondrial) have been proposed in cancer cells with reduced oxygen availability [20, 21].During alternative cytoplasmic glutaminolysis, L-glutamate formed by mitochondrial GLS1 can be transported to cytosol by GC (glutamate transporter) and is transaminated by aspartate aminotransferase 1 to α-ketoglutarate that in reductive carboxylation is converted to citrate by cytoplasmic isoforms of isocitrate dehydrogenase and cis-aconitase. In the alternative mitochondrial glutaminolysis the beginning of the pathway is the same as under normal oxygen availability (L-Gln→ L-Glu→ α-KG), however because of hypoxia, α-ketoglutarate is not effectively involved in the Krebs cycle, but it is rather reduced and carboxylated to citrate by mitochondrial isoforms of isocitrate dehydrogenase and cisaconitase (figure 2) [20, 21, 22].

Figure 2

Thus, the reductive carboxylation of α-ketoglutarate to citrate, is a common feature of the two alternative pathways.

The above mentioned alternative glutaminolysis pathways were based only on the study of labeled metabolites [23]. In literature there is no data about the expression or activity of enzymes in volved in alternative glutaminolysis and expression of the mitochondrial L-glutamate transporter. Besides, there is no experimental study concerning its role in the L-glutamate cytoplasmic and mitochondrial balance.

The most in vitro studies on glutaminolysis performed on various tumor cells such as lung, breast and colon, were done at atmospheric oxygen level (21%) or hypoxia (≤ 1%). Whereas oxygen concentration in tumor environment is highly variable, ranging from 10% to 1% and even transient anoxia. Thus, the data obtained so far do not reflect tumor metabolism under different oxygen conditions in vivo.

In the current work, for the first time, we study the gene expression of enzymatic and transporter proteins involved in the alternative glutaminolysis pathways at tissue normoxia (10% oxygen) in comparison to tissue hypoxia and atmospheric normoxia.

The statistical analysis was performed using Statistica 13.0 (StatSoft, Inc, Oklahoma, USA). Statistically significant differences in the expression of the examined genes were performed using Student’s t-test. The analysis of ANOVA variance was used to evaluate the dependence of the mRNA level of the tested genes and the oxygen concentration. Correlation between the expression of individual genes and SW480, SW620 cells proliferation was determined using the Pearson test. Results calculated from 6 separate experiments, were expressed as means ± SD, and considered statistically significant at P<0.05.

The results of quantitative RT-PCR analysis showed that mRNA expression of GLUT1 and HIF-1α for both studied lines was significantly higher at hypoxia than at tissue and atmospheric normoxia (P<0.001) (fig. 4).

Figure 4

The level of GLS1 expression at hypoxia was significantly lower (P<0.01) than at tissue normoxia for both cancer cell lines. At 21% oxygen concentration, GLS1 expression for SW480 cells was the lowest (P<0.0001), whereas in SW620 cells the expression was similar to the results obtained at 1% oxygen (fig. 5 A, B).

Figure 5

In SW480 cells the level of AST2 expression significantly decreased with increasing oxygen concentration and was the highest at 1% (P<0.001) (fig. 5 A). In contrast, no significant differences in AST2 expression for SW620 cells were observed (fig. 5 B).

AST1 expression was much lower than AST2 for both studied cancer cell lines, furthermore in all oxygen concentration itsexpression was at the same level (fig. 5 A, B).

ACL expression in SW480 cells at 1% oxygen was significantly higher (P<0.001) than at 10% and 21%, whereas the expression of the same gene in SW620 cells was significantly lower (P<0.01) in comparison to 10% and 21% oxygen pressure (fig. 5 A, B).

The level of PC expression at 1% and 21% oxygen was comparable for both studied cancer cell lines, while at 10% oxygen concentration its expression was 1.5 - fold lower in SW480 cells than in SW620 cells (P<0.01) (fig. 6).

Figure 6

The PC expression in SW480 cells decreased with increasing oxygen concentration, but significant differences were observed only between 1% and 21% oxygen pressure (P<0.01) (fig. 6). In SW620 cells PC expression was similar at 1% and 21% oxygen, but significantly lower in comparison to 10% (P<0.001) (fig. 6).

At all studied oxygen concentrations the profile of GC1 isoform expression was similar for both cancer cell lines and its expression was significantly higher in SW480 than SW620 cells. In both studied cell lines GC1 expression was the lowest at 10% oxygen, whereas significant differences was observed between 21% and 1% (P<0.01) or 10% (P<0.001) for SW480 cells and 21% and 10 % (P<0.05) for SW620 cells (fig. 7).

Figure 7

The GC2 expression was higher in comparison to GC1 for both studied cell lines at all oxygen concentration. In SW480 cells GC2 expression was significantly higher only at 1% in comparison to 10% oxygen (P<0.05) (fig. 7). In contrast, in SW620 cells GC2 expression significantly raised with increasing oxygen concentration (P<0.01) (fig. 7).

ANOVA analysis showed that in SW480 cells the expression of GLS1, AST2, ACL and GC2 gene was dependent on oxygen concentration (F (12.105) = 108.48; P = 0.00002). While, in SW620 cells oxygen concentration influenced the expression of GLS1, ACL, PC and GC2 gene (F (12.105) = 10.875; P = 0.000001).

Variable oxygen availability resulting from the disproportions in the dynamics of cancer cell proliferation and endothelial cells responsible for angiogenesis may result in local hypoxia in the insufficiently vascularized area of the tissue. Therefore, cancer cells have developed mechanisms that allow them to adapt to differing oxygen and nutrient availability [24, 25].

Among such adaptations to hypoxia is the activation of HIF-1α factor, that leads to the expression of genes involved in things such as energy metabolism, glucose transport and cell proliferation (GLUT1, vascular endothelial growth factor (VEGF), and others) [26, 27]. It is known that in various tumors GLUT1 expression correlated with chronic tissue hypoxia and limited glucose availability [28, 29]. Chung et al. [30] showed that in SW480 and SW620 cell lines, in normal and cancer colon tissue both HIF-1α and GLUT1 mRNA expression were higher under hypoxia than in atmospheric normoxia (21% oxygen), which is in agreement with our findings. Moreover, we have shown that the expression of both factors was higher in 1% hypoxia than in 10% tissue normoxia. Therefore, both HIF-1α and GLUT1 levels vary under a physiologically relevant range of oxygen tension [31].

In this study, we found that the number of both SW480 and SW620 cells was higher in 1% hypoxia than in 10% oxygen (physiological conditions) and it indicates that oxygen supply is a limiting factor in the growth of colon cancer cells in vivo. Moreover, we showed that the proliferative potential of SW620 cells was higher than of SW480 cells. We found differences between cell numbers at 1% and 10% measured by the TB method. We did not observe any discrepancies in cell mitochondrial activity estimated by MTT assay, however. This may be explained by a higher count of mitochondria at 10% normoxia. It is known that c-Myc is activated at normoxia, resulting in mitochondrial biogenesis [32, 33].

The cancerous process is considered to be a metabolic disease. Changes in cellular metabolism are induced by various levels of oxygen availability, which influence genes’ expression of proteins involved in glycolysis and glutaminolysis pathways [34].

According to the literature, the activity and expression of glutaminase (the first glutaminolysis enzyme) is increased in liver, prostate, or breast cancer [35, 36, 37, 38]. We have shown previously that GLS1 expression was higher in patients with CRC [39], which is in agreement with results obtained by Huang et al. [40]. Our results are concordant with those obtained by Xiang et al. [41] for GLS 1 expression at 1% and 20% oxygen. Interestingly, we found that GLS1 expression in SW480 and SW620 cells was significantly lower in hypoxia compared to tissue normoxia (10% oxygen). Moreover, an increase in GLS1 expression at 10% compared to 1% was higher in SW620 than SW480.

It is known that GLS1 expression depends on the c-Myc transcription factor and that GLS1 expression is elevated in many types of cancer [18]. Thus, higher GLS1 expression can be caused by higher c-Myc mitochondrial biogenesis induced in metastatic cells [10, 32, 33, 42]. The lower of GLS1 expression observed in the present study for both cancer lines at 1% oxygen may be associated with the substitution of c-Myc by the HIF-1α, which is a less effective GLS1 expression activator [41, 43, 44, 45].

The next step of glutaminolysis is conversion of L-glutamate to α-ketoglutarate, catalyzed by mitochondrial aspartate aminotransferase (AST2; transamination) or glutamate dehydrogenase (GLUD1; oxidative deamination) [46, 47].

We found that AST2 expression in SW480 cells was significantly higher in hypoxia than in tissue normoxia, while in SW620 cells it was at a similar level, with a non-significant tendency to increase at 10%. In contrary, the increase of AST2 expression in SW480 cells in hypoxia may be associated with a decrease of glutamate dehydrogenase activity (GLUD1). Son et al. [23] showed that pancreas cancer cells with a K-ras mutation exhibited a decreased level of GLUD1 expression, whereas the AST2 gene knockdown in these cells in 21% normoxia significantly reduced L-Asp level, which resulted in the inhibition of cell proliferation. Since both studied cancer cell lines have a K-ras mutation, an increase of AST2 expression may be associated with a decrease of GLUD1 activity. It can be assumed that AST2 substitutes for GLUD1 providing α-KG, which in hypoxia is an indispensable substrate in both alternative cytosolic and mitochondrial glutaminolysis pathways. In addition, higher AST2 expression observed in hypoxia may be due to the reduced supply of L-Glu within the mitochondria resulting from lower GLS1 expression. Moreover, AST2 regulates the synthesis of aspartate that is transported to cytoplasm by the aralar transporter and is used for nucleic acid synthesis. In contrast, cytoplasmic AST1 did not change under the influence of various oxygen concentrations in both colon cancer cell lines because it may be c-Myc independent. In our previous study we found that addition of L-Asp to SW480 and SW620 cell cultures resulted in enhanced lactate synthesis [44]. It may indicate that the role of AST1 is the conversion of an excess of mitochondrial L-Asp to oxaloacetate and then lactate (oxaloacetate – malate – pyruvate – lactate axis) with concomitant synthesis of L-Glu for cytoplasmic alanine aminotransferase (ALT) and phosphoserine aminotransferase 1 (PSAT1), which catalyze the conversion of L-Glu into α-KG [48, 49].

According to literature data, α-KG produced with the participation of AST2, depending on the prevailing aerobic conditions, is converted to citrate in the classical or alternative glutaminolysis pathway [50].

In the classical glutaminolysis under normoxia conditions, α-KG is included in the Krebs cycle, where it is converted to oxaloacetate, which after condensation with acetyl-CoA forms citrate. The citrate is transported to the cytoplasm, where it is split into acetyl-CoA and oxaloacetate in an ACL-catalyzed reaction.

In the current paper, ACL expression in SW480 was significantly higher at 1% than 10% oxygen. It can be related to reduced citrate supply from both alternative pathways. Studies carried out under hypoxia conditions have shown that active HIF-1α inhibits α-ketoglutarate dehydrogenase, slows down the Krebs cycle, and reduces oxaloacetate synthesis [51, 52]. Moreover, HIF-1α inhibits pyruvate dehydrogenase and leads to decreased acetyl-CoA synthesis [5, 53, 54]. Therefore, citrate sources are both alternative glutaminolysis pathways. In metastatic cells ACL and PC expressions were significantly higher at 10% than 1% oxygen, which may indicate that SW620 cells are more responsive to c-Myc influence. Confirming this thesis is a study conducted by Lin et al. [55], which showed higher mitochondria numbers in SW620 than in SW480 cells. As mentioned above, c-Myc activates mitochondria biogenesis, resulting in the increase of mitochondrial enzyme activity and higher citrate availability for ACL, which can explain its elevated level in SW620 cells.

According to the literature, the expression of the glutamate transporter (GC) has never been studied before under hypoxia and tissue normoxia conditions.

The presence of two mitochondrial isoforms of the GC transporter (GC1, GC2) in normal tissues was confirmed in many studies, but there is no data about their existence in colon cancer cells [36, 56, 57, 58]. We showed for the first time the presence of both GC isoforms in human colorectal cancer cells and the effect of oxygen on their expression. The expression of the GC1 isoform was lower than of GC2 in both hypoxia and normoxia for SW480 and SW620 cells, which indicates that GC2 is the major isoform responsible for L-Glu transport. Glutamate is pumped out from mitochondria to cytosol by GC2 transporter and this consequently causes a decrease in its mitochondrial concentration. In turn, transport of L-Glu to the mitochondria requires the participation of an aralar transporter, which simultaneously carries the mitochondrial L-Asp formed by AST2 (exchange L-Glu/L-Asp).

We also confirmed our previous findings that studies routinely conducted under non-physiological atmospheric normoxia may not reflect the real situation in vivo, where different and changeable oxygen conditions exist. Therefore, results obtained at variable oxygen concentrations may better correspond to the in vivo conditions [48].

In summary, the above-mentioned findings suggest a difference in metabolic adaptation to oxygen availability in vivo between primary and metastatic colon cancer cells.

Higher expression of glutaminolytic genes at 1% oxygen in SW480 cells indicates adaptation to a deficiency of nutrients in the tumor microenvironment at chronic hypoxia, whereas higher expression in SW620 cells at 10% may result from increased oxygen-dependent mitochondrial biogenesis. It is known that metastatic SW620 cancer cells exhibit higher mitochondrial metabolism (c-Myc, α-PGC 1) and show features similar to cancer stem cells (CSC) [59].

To the best of our knowledge, the present study comparing the influence of physiological oxygen levels on glutaminolytic gene expression in cancer cells is pioneering research. Since the oxygen level in the tumor milieu never exceeds 10%, we suggest carrying out future studies at tissue normoxia rather than at atmospheric normoxia. In the present work we focused on expression of glutaminolysis genes under differing physiological oxygen levels. Our results give support for further comprehensive studies involving enzymatic protein expression and factorsi.e HIF-1α, c-Myc affecting their activity.