The objective of this study was to identify changes in glucose metabolism that are associated with cellular transformation and acquisition of tumorigenic and metastatic potential in breast epithelial cells. The MCF10 panel of cell lines are an attractive system for such studies as they are a stable and well-characterized tissue culture model of mammary tumor progression. In order to minimize artifacts due to external components, a minimal common media was designed and used for all four cell lines investigated. The MCF10-A cell line requires growth factors to proliferate in two-dimensional culture, while the other MCF10 cell lines used here respond to such factors but do not require them for growth [5
]. However, the standard MCF10 media formulation contains a number of components which have the potential to alter cellular metabolism, including insulin [17
]. We developed a MEM-based, glucose-centric media supplemented with EGF in which all four cell lines displayed continual proliferation over the 48 h 13
C-glucose labeling period. We believe that growing each of the cell lines on this common background during log-phase growth has minimized any potential metabolic variation due to differential responses to media factors, basal proliferation rate and cell cycle stages.
Thus a comprehensive 2D HSQC NMR and GC-MS profiling strategy was used to characterize the metabolic program of the MCF10 model of mammary carcinoma and to determine if the findings support the Warburg effect or suggest another explanation for the uptake of glucose and increased lactate production in cancer cells. The transformed phenotype is accompanied by (1) increased catabolism of glucose by the pentose phosphate pathway (both oxidative and non-oxidative), (2) an increase in flux through the TCA cycle and a corresponding increase in the cellular pool of succinate, (3) increased flux through the glycine–glutamine hub and a corresponding increase in glutamate, and (4) a substantial increase in fatty acid synthesis and oxidation. These alterations are essentially stable through additional transitions in tumor cell phenotype and are evident in metastatic cells. Perhaps equally as significant though, the study reveals that acquisition of the metastatic phenotype is associated with a massive increase in the biosynthesis of proline. To our knowledge, this is the first study to take a comprehensive and comparative approach toward the links between breast tumor progression and metabolism, and the first study to characterize the metabolome of the metastatic cell.
Our results on glucose uptake and lactate excretion, the two parameters discussed by Warburg, are only partially consistent with his hypothesis. We do observe a doubling of glucose uptake in the metastatic MCF10-CA1a cell line when compared to the non-tumorigenic MCF10-A cells. Increases in glucose uptake were also observed in the less aggressive MCF-10AT and MCF-10AT1 cells, although the magnitude of these increases was small relative to the other tumor cell lines. However, we did not observe any consistent trend toward increased production of lactate by the tumor cells compared to the MCF10-A cells. This is most likely due to two factors: the immortalization of the MCF10-A cell line and the necessities of two-dimensional tissue culture. The MCF10-A cell line has a karyotypic gain of MYC
concurrent with a homozygous loss of CDKN2BA
]. Cytokines have been shown to generally stimulate glycolysis (specifically, IL-3, IL-7 and IL-2) and increase the amount of lactate produced per glucose unit consumed [18
]. Additionally, the MCF10A cell line requires EGF, which has also been shown to increase glycolysis [19
]. Thus the relatively high lactate production in the MCF10-A cell line is most likely a result of the required tissue culture conditions.
However, our findings indicate that the distinctions between the tumorigenic and non-tumorigenic cell lines really center on how glucose is used and how carbon from glucose is shuttled through central carbon metabolism. The comprehensive strategy employed in the present study allowed an assessment of carbon flux through both the oxidative and non-oxidative branches of the PPP, each of which have two outputs—pyruvate and ribose. However, the oxidative branch has the additional role of synthesizing cellular NADPH, which is the primary source of reducing power of the cell. Flux through both branches of the PPP is increased in transformed cells, and the fact that flux through the non-oxidative branch is increased is consistent with the observations in different tumor cell lines [16
]. Perhaps more interestingly, the magnitude of the increase in flux is higher for the oxidative branch. The increase in flux through the oxidative branch is probably a reflection of the amplified need of the transformed cells for the reducing power of NADPH, which is used for many biosynthetic reactions and is involved in protecting the rapidly proliferating cell from reactive oxygen species through glutathione. The increased flux through the oxidative branch also has potential implications for anti-tumor therapy. Many common chemotherapeutics act by inducing oxidative stress [22
], which could be counteracted by the reducing power produced by the oxidative branch of the PPP. Therefore, drugs that independently inhibit the oxidative branch of the PPP may reduce the ability of the tumor cell to protect against oxidative stress brought on by chemotherapy.
The increased activity of the PPP also supports increased synthesis of nucleotides, which are in increased demand in highly proliferative tumor cells. Flux through each branch of the PPP can provide ribose-5-phosphate and then 5-phosphoribosylpyrophosphate, a key nucleotide precursor. The production of nucleotides also requires tetrahydrofolate, which is synthesized from glycine. Thus the reduction in the glycine pool size may relate to the fact that glycine donates methylene units to tetrahydrofolate. Glycine can also be converted to glutathione, which in turn can be converted to glutamate and then to proline. The increased flux through the PPP and the glycine–glutamine hub as cells become more tumorigenic is therefore critical because of the significant demand placed upon nucleotide synthesis during transformation.
Two key alterations in the TCA cycle were also observed in the H-ras
transformed MCF10-AT cells; these alterations persisted in cells with increased hyperplasticity and metastatic potential. The changes include a near doubling of flux through the TCA cycle, and a corresponding increase in the size of the succinate pool. The exact size of the increase in the succinate pool could not be determined because the pool size is extremely low in the non-transformed MCF10-A cells. Nevertheless, we have determined the change to be at least 5-fold. This finding should be interpreted in the context of recently published work showing that high levels of cellular succinate can inhibit HIF-1α prolyl hydroxylase, and thereby cause a pseudo-hypoxic cellular response [27
]. Such a response is expected to activate genes under the control of HIF-1α and thereby promote a tumorigenic phenotype. In fact, germline mutations in either fumarate hydratase or succinate dehydrogenase can drive this event and lead to a predisposition to tumors [28
]. Therefore, the increased levels of succinate in the transformed MCF10-A cells may actually contribute to their tumorigenicity. The mechanisms underlying the increase in the succinate pool are not readily apparent. It is unlikely that inhibition of succinate dehydrogenase is involved in the transformed MCF10 cells because the total flux through the TCA cycle was increased. Thus, there is the distinct possibility that an increase in TCA cycle flux independently drives an increase in the succinate pool, which in turn could increase the level of HIF-1α. In other systems, such a connection between glycolysis and HIF-1α levels has been proposed, although succinate was not involved [29
In this study two major alterations were observed in the glycine–glutamate hub. First, in the transformed cell lines displayed a 2- to 4-fold increase in the flux of carbons through GSH (indicating greater GSH turnover). This increase in flux through GSH is observed even as the size of the GSH pool decreases, indicating that increased demand for GSH in the metastatic cell is greater that the cell is able to replenish via de novo synthesis. Much like the results concerning the pentose phosphate pathway (which provides the NADPH needed for glutathione reduction), these findings demonstrate the significant demand placed upon cellular redox buffering during tumorigenic transformation. Second, we observed a substantial increase in the flux from glucose to proline in the MCF10-CA1a cell line. To our knowledge this is the first report of increased proline biosynthesis in metastatic cancer cells, which could be connected to increased turnover of the extracellular matrix in metastatic cells. Since tumor cell invasion is linked to increased degradation of collagen, one might also presume that as tumor cells invade they must replenish the digested extracellular matrix in order to adhere. This process would require increased synthesis of collagen and thus, since collagen is proline rich, a greater need for proline. There are few if any studies that examine the collagen synthesis by tumor cells, but the finding presented here indicate that this process should be the topic of future study.
Finally, these findings are consistent with our own observations for another pathway, the synthesis of fatty acids in breast tumor cells. There is a large body of work showing that fatty acid synthase (FAS) is up-regulated in a wide range of cancers [30
] and that inhibition of FAS can halt tumor cell proliferation and induce tumor cell death (for review see [31
]). In fact, we have reported that Orlistat, a drug approved for obesity, is an inhibitor of fatty acid synthase and that this drug elicits cell death in a number of breast tumor cell lines [32
]. Not only are the findings of the present study consistent with these prior observations but they also add another dimension. Each of the fatty acids is derived from either elongation or desaturation (or both) of palmitate by the action of stearoyl-CoA desaturases and long chain fatty acid elongase. Both stearoyl-CoA desaturase [33
] and long chain fatty acid elongase [34
] are part of the SREBP-regulated lipogenic program that is often activated in solid tumor progression [35
]. Increased stearoyl-CoA desaturase expression and activity results in increased mono-unsaturated fatty acid (MUFA) levels and this increase in MUFAs is associated with cancer risk [36
] and mechanistically may help modulate membrane fluidity [37
], and anchorage-independent growth [38
] in tumor cells. The fact that tumor cells convert the newly synthesized palmitate to other fatty acids also suggests that stearoyl-CoA desaturases and fatty acid elongase should be explored as potential drug targets.
To our knowledge, this is the first study to take a comprehensive and comparative approach toward the links between breast tumor progression and metabolism, as well as the first study to characterize the metabolome of the metastatic cell. The results provide an essential foundation for subsequent interrogation of individual metabolic steps, and for gauging the requirement of each step in tumor cell progression. The analysis provides important information that will help guide the interpretation of studies of breast tumor metabolism under hypoxic conditions and studies of the metabolism of whole tumors in vivo, which are comprised of multiple cell types. It also provides a rationale for using metabolomics for cancer detection and diagnosis, as well as for the development of novel therapeutics that specifically target the tumor cell metabolome.