In this work we have compared the exo-metabolome of three different UBCcls: 5637, RT112, RT4. Under the conditions of our experiments these three cell lines showed different growth rates: 5637 and RT112 cells show a faster proliferation (=0.03 h-1) than RT4 cells (=0.02 h-1) (Additional file 3). In a typical experiment, cells were grown in RPMI culture medium for 1, 2 or 3 days in separated containers and the experiment was repeated 4 to 5 times to have enough replicates for statistical analysis.
Using 1H-NMR spectroscopy, we have followed the daily variation of the concentration in the extracellular medium of 35 metabolites. They included amino acids (66%), organic acids and derivatives (17%), carbohydrates (6%) and others (11%) (Additional file 4). The assignment of the metabolite signals in the 1H-NMR spectra was confirmed using two-dimensional experiments, like 1H-1H TOCSY and 1H-13C HSQC. The list of chemical shifts used to uniquely identify all metabolites can be found in the supplementary material (Additional file 4).
Before the data analysis, two types of normalizations were applied (see Additional file 2). In the first one we have divided the rate of variation of a metabolite by the rate of cell growth, obtaining the specific rate of variation, qM, in pmol/cell. The second normalization takes into account the relative molar weight of the variation of a metabolite with respect to the appropriate sum of all other compounds that are consumed (wS) or excreted (wP). These two normalization schemes help to understand the role of a compound in the cell metabolism from two different points of view. qM measures the degree of upregulation of the exchange between the compound and the extracellular medium or of the metabolic pathway in which it is involved. The wS value can be interpreted as the weight of a nutrient on the total "diet" of a cell, although it does not necessarily reflect the degree of activation but the relative participation of the related pathway in the total metabolism of the cell. The analysis of the normalized variations shows that 40% of the metabolites were only consumed, 11% were only excreted and 49% showed a complex pattern (Figure 1). Among the nutrients, glucose shows the highest variation, followed by glutamine, serine, branched-chain amino acids (BCAAs) and arginine. The excretion profile is mainly formed by lactate and alanine, along with small amounts of pyruvate and formate.
A multivariate model was then calculated to determine the degree of differentiation of the three UBCcls exo-metabolomes including data from all daily variations. The corresponding supervised model (OPLS-DA) showed a significant separation among the classes (Additional file 5). Variables that mostly contributed to the separation included metabolites belonging to the pyruvate and serine metabolisms, plus arginine, glutamine and BCAAs. These particular pathways will be examined in detail in the following chapters.
Glycolysis is the most active metabolic pathway in 5637 cells
The calculated qM values for glucose and lactate of the three lines were very similar (Figure 2A and 2B, left panels). Those of glucose were all around 17 pmol/cel, in agreement with the values of 14 and 17 pml/cel measured for RT4 and TCSSUP cells, respectively [8]. Given that the growth rates of 5637 and RT112 cells are 34% higher than those of RT4 cells, and that the calculated qM value is the same for all lines, the former consume 34% more glucose per hour than the latter, but the higher consumption is explained exclusively by the faster cellular replication and not by an upregulation of glucose uptake. A similar reasoning can be made for lactate, which is excreted 34% more per hour by lines 5637 and RT112 than by RT4.
The weight of glucose in the diet of the three cell lines is different: it represents 83%, 70% and 68% of the total diet of cells 5637, RT112 and RT4, respectively (Figure 2A, right panel). As a consequence, also the weight of lactate in the total excretion is different, and represents 90% for 5637 and RT112 cells against only 82% of RT4 cells (Figure 2B, right panels). An analysis of these data shows that glucose consumption and lactate excretion are the most important events in the exchange of compounds between 5637 cells and the extracellular medium, while RT cells show a more varied consumption with a lower glucose weight.
While all cell lines excreted small amounts of pyruvate, cells 5637 did so to a lesser extent. (Figure 2C). On the other hand, alanine excretion is higher for RT4 cells (Figure 2D), and shows a very clear trend among the UBCcls, in the order RT4>RT112>5637. These values show that within the 5637 cells most of the pyruvate finishes its fate converted into lactate. So, it is possible to conclude that glycolysis is the major active pathway for the production of ATP and biomass in 5637 cells, is of intermediate importance for RT112 cells and is of lower importance for RT4 cells.
RT cells show an active oxidative metabolism
RT4 cells show increased consumption of arginine, glutamine, BCAAs and serine, along with increased excretion of formate with respect to 5637 cells (Figure 3). All these compounds are related to metabolic pathways that occur inside the mitochondria, require an active oxidative metabolism, and were already found altered in different tumors and cancer cell lines [9-11]. RT4 cells show also a higher serine consumption than the other two cell lines, along with a higher excretion of a product of its catabolism, formate, a further confirmation of an active mitochondrial activity and active oxidative phosphorylation metabolism (OxPhos) [12]. RT112 cells, on the other hand, show a similar consumption of glutamine and BCAAs with respect to RT4 cells, but serine consumption and glycine and formate excretions are lower. This implies that the activation of the OxPhos metabolism is intermediate between those of RT4 and 5637 cells.
The risk of progression associated with cell lines correlates with the balance between glycolysis and OxPhos
The diet and excretion profiles of the three lines are summarized in Figure 4. It is clear from these graphs the different weight that glucose shows in the three diets, decreasing its importance in the 5637-RT112-RT4 series. In contrast, glutamine, BCAAs and arginine show an opposite trend and are more important in the diet of RT cells. Regarding the excretion profile, lactate shows the same high weight for the two fast growing lines, 5637 and RT112. The excretion of alanine follows an opposite behavior, in line with the shift of pyruvate metabolism, which is the cause of lower lactate excretion and higher alanine excretion in RT4 cells.
Figure 5 shows the different pathways described so far, with the indication of the degree of activity suggested for the different cell lines studied. In conclusion, 5637 cells, associated to tumors with high risk of progression, consume almost exclusively glucose and excrete lactate, index of a high weight of glycolysis in their metabolism. At the other extreme, RT4 cells, associated with tumors at low risk of progression, show mitochondrial activity and consequently active OxPhos metabolism. RT112 cells are metabolically located between the other two lines, in accordance with their mixed genetic signature.
The lactate/alanine ratio is a sensitive marker of the equilibrium between glycolysis and OxPhos both in vitro and in vivo
The pyruvate produced by glycolysis can either be transformed into lactate in the cytosol, or enter the mitochondria where, among other reactions, can be converted to alanine through a transamination reaction using the amino group of glutamate and producing alpha-ketoglutarate [13,14]. This means that the value of lactate excretion is directly proportional to the degree of glycolysis activity, whereas the degree of alanine excretion can be used as a measure of mitochondrial and OxPhos activities. For this reason, the lactate/alanine ratio is a metabolic measurement of the glycolysis/OxPhos equilibrium.
We have noticed a high and significant correlation between the concentrations of lactate and alanine excreted during the three days by the UBCcls (Figure 6A). The fact that there is a linear relationship indicates that the relative amount of pyruvate used for the synthesis of one or the other metabolite remains constant during the 72 hours. From the analysis of the slope values, it is possible to calculate that cells 5637 excrete 62 lactate molecules per alanine molecule, while this value is reduced to 25 and 14 for cells RT112 and RT4, respectively (Figure 6B). These numbers translate very sensitively the balance between non-oxidative and oxidative metabolic states observed so far, correlating a high value of the lactate/alanine index with the preponderant use of glycolysis and a low one with OxPhos. This postulates it as a good candidate to become an in vivo marker to measure the metabolic balance, and thus the risk of progression. To verify this possibility, we have used NMR spectroscopy to measure lactate and alanine concentrations in urine samples from 49 NMIBC and 42 MIBC patients. As in the case of the measurement of metabolite concentrations in the extracellular medium, NMR spectroscopy represents a reproducible and very accurate method to determine these two concentrations also in this biological fluid. Although the profile of urine shows a complex overlap of signals, lactate and alanine show two signals in resolved areas of the spectrum that make their quantification by deconvolution straightforward (Additional file 6). Our results show a significant two-fold increase in the value of the lactate/alanine ratio in MIBC patients with respect to NMIBC (Figure 6C), reflecting a higher abundance of cell lines with active glycolysis in high-grade tumors. On the other hand, 70% of NMIBC patients present a low value (< 1) (Figure 6D), suggesting that in these tumors the metabolic balance is displaced towards the OxPhos.