A metabolomic approach to identifying platinum resistance in ovarian cancer

Our metabolomics analysis demonstrated that distinct metabolic profiles can be detected for platinum sensitive A2780 and resistant C200 ovarian cancer cells (Figure 1). The altered metabolites encompassed all major molecular categories and metabolic pathways (Figure 2) and appeared to be inter-connected. Arranging the significantly altered metabolites from the KEGG metabolic pathways and the top canonical pathways from IPA into a single map revealed a high level of connectivity and interdependence among the pathways (Figure 5). For example, the glycolysis-to-citrate cycle not only generates adenosine triphosphate, which is required for all biosynthetic pathways to occur, but also synthesizes essential metabolites such as glyceraldehyde 3-phosphate, a precursor for the generation of amino acid serine. Serine then enters the methionine pathway to produce cysteine, leading to GSH production. GSH, apart from protecting cells from oxidative stress, also maintains the NADPH/NADP+ ratio, which is required by the folate cycle and nucleotide biosynthesis, among others [29]. Cysteine can be converted back to pyruvate by transamination [29]. The glycolysis-to–pentose phosphate pathway metabolites are also the sources for precursors of nucleotides and modifiers of lipids, purine and protein modifications. Thus the metabolism of a cell runs its own ‘cycle of life’ in unison that sustains the cell function and growth. Most of the altered metabolites participate in more than 1 pathway in significant ways and the change in that 1 metabolite could have a resonating effect for other pathways.

Collectively the largely significant altered metabolites belonged to the methionine-cysteine metabolism and one of its downstream metabolic tributaries of the GSH pathway (Figure 6). The methionine pathway along with folate and transulfuration pathways constitutes the one-carbon metabolism [29,30]. The one-carbon metabolism is known as a metabolic integrator that assimilates products from the glucose pathway and amino acids to fulfill integral and essential cellular biosynthesis for biological functions, such as cellular turnover and proliferation. The one-carbon metabolism pathway has also been investigated as a major contributor to the process of oncogenesis [30]. Studies have shown that methionine levels are directly involved in promoting proliferation of cancer cells [31] and in protecting the cancer cells against chemotherapeutic drugs like Fluorouracil [32]. Methionine deprivation has been shown to inhibit tumor cell growth in various cancers both in in vitro and in vivo models along with chemo-sensitizing cancer cells [33-36]. We observed decreased levels of methionine in resistant C200 cells, which could indicate its increased utilization (Figure 3B), and could suggest the presence of ‘methionine dependency’ in the resistant cells. Methionine-dependency is defined as inability or reduced ability of cancer cells to proliferate even when methionine is replaced by its precursor homocysteine, this phenotype has been demonstrated in several cancers like prostate and lymphoma [37,38]. Since methionine is an integral participant of various metabolic pathways, the exact mechanism underlying methionine-dependence of various cancer cells has been difficult to elucidate [37]. Homocysteine is an important metabolite of the methionine pathway that is coupled with the folate cycle and on gaining a methyl group from methyl THF converts back to methionine [29,39]. Methionine can also be produced by the salvage pathway via the crucial enzyme methylthioadenosine phosphorylase (MTAP), which has been shown to be deleted in various cancers [40]. Methylthioadenosine, the substrate for MTAP, was observed to be present in lower quantities in C200 cells compared to A2780 (Figure 3B), which could again suggest use and diversion of methionine to other pathway end-product metabolites like GSH or putrescine (discussed below), rather than the normal metabolic cycle. Homocysteine through transsulfuration reactions also gives rise to cysteine. One of the most important roles for cysteine is to act as the limiting factor for synthesis of the antioxidant GSH [39,41,42]. GSH is the main endogenous antioxidant and protects the cells from metabolic stresses by nonenzymatically reducing substances like peroxides and free radicals and maintaining an intracellular reducing environment. It also activates GSH s-transferase and detoxifies xenobiotics and other cell damaging compounds [42]. GSH has been established as a protective mechanism against the increased oxidative stress encountered by the cancer cells [42,43], which has been implicated as a contributing factor to chemoresistance development in cancer cells [41]. GSH is present in 2 forms: the reduced and the oxidized GSH disulfide forms. C200 cells displayed higher levels of both forms of GSH with the oxidized form being higher than the reduced form (Figure 3G). This could indicate a higher level of oxidative stress in the C200 cells and a very active GSH mediated antioxidant system that is offering added protection to the cells. Thus is our system, the lowered methionine levels could also indicate its eventual utilization to maintain the high levels of GSH observed in the C200, as a means to encounter the assault of chemotherapeutic drugs.

A central metabolite S-adenosylmethionine (SAM) from the methionine pathway, acts as a donor for methylation reactions involving methylation of histones, DNA, RNA and all general protein methylations [39,44], and also participates in biosynthesis of phosphatidylcholine, the major component of cell membranes, by forming the polar head group with choline [30]. Methylation processes have been widely implicated in cancer progression, including ovarian [45,46]. Aberrant methylation has also been proposed as a contributing factor for acquisition of chemoresistance, especially in resistance against DNA-damaging platinum drugs [47,48]. Recently, a cisplatin-resistant cell line derived from the sensitive A2780 ovarian cancer cell lines were shown to preferentially select for DNA-hypermethylation and the obtained methylated gene signature was found to partially hold validation in a small subset of patient relapsed ovarian tumors [49]. Thus it is possible that the chemoresistance in C200 ovarian cancer cells could be a result of increased methylation of selective genes, which could be reflected in its altered metabolism. SAM also provides methyl groups for biosynthesis of polyamines, a vital class of products involved in cell proliferation, and have been shown to be increased during malignancy [50]. Polyamines include putrescine and its derivatives spermidine and spermine, which are synthesized from ornithine and shown to be required for proliferation [50-52]. On one hand, inhibition of polyamine synthesis has been shown to inhibit cancer cells, but on the other hand recent studies have reported decreased levels of spermine as a metabolic biomarker for cancer cells [50,53]. We observed an increased level of putrescine and decreased levels of spermine and spermidine (Figure 3G). This could either indicate a block after putrescine formation or increased utilization of these metabolites. Ornithine decarboxylase (ODC) is the first enzyme required for polyamine synthesis and is regulated by Myc oncogene [50,54]. Induction of spermidine/spermine N1-acetyltransferase 1(SAT1), the key enzyme regulating catabolism of polyamines has been reported to result in decreased spermine and spermidine and increased putrescine levels [55], similar to our observation in C200 cells (Figure 3G). ODC and SAT1 have been shown to be overexpressed in neoplastic prostate tissue [56]. Platinum drugs have also been shown to regulate polyamine metabolism enzymes including SAT1 [18]. Thus acquisition of platinum resistance in C200 cells could be related to perturbations in the enzymatic makeup of polyamine metabolism.

Our observations are similar to other studies where the altered methionine pathway has been advocated to play a role in ovarian and other cancers and its metabolites presented as biomarkers. Alteration of methionine pathway was suggested early in ovarian cancer [57], where accumulated homocysteine in ascites was an indicator of malignancy. A recent study has shown cisplatin treatment of embryonic mouse cells to induce significant changes in the methionine degradation pathway, including GSH and polyamine metabolism, along with other methionine associated pathways [18]. Sarcosine (n-methylglycine), a product of the methionine degradation pathway was reported as a biomarker in urine of metastatic prostate cancer patients [19]. Comparison of metabolites from early recurrence (within 2 years of surgery) and recurrence free (more than 5 years) prostate cancer patients found elevated products of methionine catabolism in serum, which included sarcosine, cysteine, cystathionine and homocysteine [58]. Targeting of the methionine pathway is being actively investigated in cancer therapeutics. Targeting the epigenetic status by inhibiting methylation of various genes in tumors is one of the earliest and most pursued chemotherapeutic approaches [59-61]. SAM-mediated methylation and polyamine synthesis is being actively investigated in preclinical studies of various cancers [53,62]. Inhibition of GSH activity is also a major area under consideration for specific targeting of chemoresistance [42]. Thus the methionine pathway metabolites appear to have the potential to act as biomarkers specifically for metastatic or recurrent tumors. Together, with the observation that most recurrent tumors are also chemoresistant and our findings, surfacing of the methionine pathway as an indicator of chemoresistance in ovarian cancer is significant. Thus, an in-depth analysis of the methionine metabolism encompassing and integrating the levels of input, intermediate and outcome metabolites flux, along with the expression and activation status of the enzymes catalyzing these reactions will provide a complete picture in understanding the functional significance of this pathway in platinum chemoresistance of ovarian cancer.

While lipids were the most altered class of compounds (36.87%) observed between the resistant and sensitive cell lines, they were not well represented in the KEGG pathways. However, 78% of the lipids had at least 1 identifier recognizable by IPA where lipid metabolism was the fourth altered biofunction. Additionally, all of the top 5 networks had substantial participation of lipid metabolites, related enzymes and downstream effectors. Approximately 70% of the lipids were decreased in C200 cells, which could be suggestive of a decreased turnover or a decrease in the metabolism of membrane lipids. Decreased circulating phospholipids (plasmenylphosphoethanolamine and lysophosphadylcholine) and increased levels of lysophosphatidic acid (LPA) have been proposed as biomarkers for ovarian cancer [63,64]. LPA has been shown to be a potent mitogen for ovarian cancer cells and are found in increased amount within ascites [63]. Plasmenylphosphoethanolamine can act as an antioxidant and its absence may be a consequence of a higher oxidative environment associated with chemoresistant cells. Lysophosphadylcholine is a main membrane component and could be utilized as a building block for increased proliferation of cancer cells [63]. The largest group affected was the long chain fatty acids, which included palmitate, oleate, stearate, etc. (Figure 7). Alternatively, since fatty acids are the major building blocks for the synthesis of triacylglycerides, which are mainly used for energy storage, this could also indicate an increased breakdown of the fatty acids by the mitochondria for energy production. The IPA network analysis projected energy metabolism and lipid metabolism metabolites as contributors in 3 of the 5 top networks (Figures 4B and Additional file 7: Figure S4; Additional file 3: Table S3). A closer look at the involved molecules supports an altered energy metabolism between the resistant C200 and sensitive A2780 cells. The C200 cells have higher levels of metabolites from the tricarboxylic acidcycle (pyruvate, citric acid and succinic acid) and lower fatty acids (palmitic, oleic and linoleic acid) and links to the enzymes from the respective pathways (tricarboxylic acid cycle: aconitase, succinate dehydrogenase, lactate dehydrogenase; fatty acid oxidation: Carnitine palmitoyltransferase-1, Acetyl CoA Carboxylase) (Figure 6). Therefore the lower levels of overall lipids may be indicative of either decreased biosynthesis or increased utilization of lipids. This could suggest a further alteration in the energy metabolism of the chemoresistant cells, along with the already altered metabolism established in cancer cells [65].