Glioblastoma multiforme (GBM) is the most common and most malignant brain tumor. There are still no curative treatment concepts available [1
]. Rapid growth, despite a lack of energy substrates, and an early migration of the tumor cells, are responsible for its poor prognosis [1
]. To develop these aggressive properties, the tumor cells undergo molecular changes, which can lead to a change in cell metabolism or phenotype. In the early 1920s, Otto Warburg, a German scientist and later Nobel Prize laureate, observed increased lactate levels in tumor cells. He hypothesized that tumor cells generate energy via oxygen-independent glycolysis, despite the availability of oxygen. This process, which has since been named “the Warburg effect”, and which is characteristic of many tumor entities, appears to represent an adaptation mechanism that enables tumor cells to meet the high energy requirements caused by their rapid proliferation [3
]. Proton-coupled monocarboxylate transporters (MCT) have been described in connection with the distribution or membrane passage of lactate produced by oxygen-independent glycolysis. MCT1 and MCT4, in particular, are overexpressed in GBM cells [4
]. Previous studies have shown a positive correlation between the expression of both transporters, with tumor cell proliferation, angiogenesis, and the degree of malignancy of gliomas [5
]. Initial findings from studies focusing on the function of the transporters in animal experiments suggest that MCT4 is responsible for the exportation of lactate. On the other hand, the exported lactate could be absorbed again via MCT1 and then be metabolized for further energy production [4
]. Lactate transporters also seem to play a role in the context of the epithelial to mesenchymal transition (EMT), a process for generating a highly invasive GBM phenotype [5
]. Cells with a stem cell-like character, which are associated with the high invasiveness of GBM cells and their resistance to radio- and chemotherapy [6
], showed an increased expression of MCT1 and MCT4 [5
]. Interestingly, a close relationship between the tumor stem cell theory and the dormancy phenomenon seems to be obvious [7
In the framework of developing new therapy strategies against GBM, inhibitors of MCT1 and MCT4 are promising targets, thus understanding their relation to other tumor progression related processes is of great interest [4
]. Furthermore, the technique of MRSI might take an important position in this context, since it offers the possibility of gaining insights into the metabolism noninvasively. Currently, only a few studies focus on the connection between MCT proteins and other tumor progression related processes in GBM, and detailed in situ work regarding the MCT expression in human GBM is not available. Furthermore, only little is known concerning a possible connection between MCT expression and changes in multi voxel magnetic resonance spectroscopic imaging (MRSI), a method for assessing local metabolites.
Thus, the objective of this study was to compare the expression profiles of MCT1 and MCT4, as well as EMT, stem cell and dormancy markers in human GBM ex vivo tissue samples from the center and edge of the tumor, using quantitative reverse transcription PCR (qRT-PCR) and immunofluorescence staining. Additionally, the results were compared, both with the lactate concentration determined by an enzyme-linked immunosorbent assay (ELISA) in the ex vivo tissues samples, and with a multi voxel MRSI of the areas examined before surgery, in particular with regard to the lactate peak.
Of all the various brain tumors known to us, GBM is the most malignant and with the worst prognosis. According to the current state of knowledge, this disease cannot be cured. The main reason for this poor prognosis is the rapid tumor growth with the early invasion of individual tumor cells into the surrounding tissue [1
]. In addition, GBM has a high resistance to the currently used therapy strategies, so that tumor cells remaining after therapy represent the source of the rapid tumor recurrences [1
]. The tumor grows rapidly, even though there is an increasing lack of energy substrates, especially in the center of the tumor. To enable this rapid growth, the tumor appears to be able to generate energy from oxygen-independent glycolysis [3
]. The resulting lactate derived from this process is now available for further energy production in tumor parts connected to vessels [11
]. In line with the studies mentioned, we found a higher concentration of lactate in the contrast-enhancing part of the GBM using MRSI. Indeed, lactate accumulates in cancer cells [12
], and as previously shown in animal studies, MRSI seems to be a reliable tool for monitoring therapy [13
]. Instead of the often-used single voxel spectroscopy, we used multi voxel spectroscopy to obtain smaller volumes per region, and to measure different regions of the tumor and its surrounding parenchyma. As expected, the spectral quality in the voxels close to the ventricles and the skull was so poor that the corresponding data was rejected. As described above, representative voxels for each region were chosen, and a TE of 288 ms was chosen to get a better lactate signal at 1.3 ppm [14
]. Nevertheless, contamination of the lactate peak with signals of lipids cannot be completely avoided, and the amount remains unclear. Previous studies described a correlation of the progression-free survival with the lactate/creatine ratio [15
], or the distribution of metabolites with tumor subtypes [17
]. Despite the comparatively low number of patients in this study, the relative changes in the lactate concentrations among the different regions were significant. In accordance with this, we were able to detect a further accumulation in the center of the vital parts of the tumor, compared with the edge as determined by a lactate ELISA. The further distribution of the accumulated lactate is generated by MCTs [18
]. MCT1 and MCT4 were found to be overexpressed in glioma cells [18
]. Our investigations also showed a clear expression of MCT1 and MCT4 in all GBM samples. MCT1 showed a higher expression than MCT4 in the center and at edge of the tumor, which falls in line with the ubiquitous expression of MCT1 described in the literature [20
]. Interestingly, we observed a statistically significant decreased gene expression of MCT1 at the edge of the tumor compared with the center. According to the ratios of mRNA expressions of mainly MCT1 between center and edge of the tumor, three main GBM groups could be distinguished. Whereas the first group was characterized by a lower MCT1 mRNA expression at the edge of the tumor combined with a higher expression of cell-type markers connected to tumor progression at the edge of the tumor, the second group was characterized by a lower expression of MCT1 and MCT4 at the edge of the tumor, and a heterogeneous expression of the cell-type specific markers in the different regions. The last group revealed a heterogenous pattern of all examined markers. Interestingly, the relative differences in the lactate concentrations seem to vary specifically in the center and at the edge of the tumor in these groups, as well. Since GBM is known for its distinct inter- and intra-tumor heterogeneity, the observation of a wide range of different patterns is not surprising [21
]. Even considering the subgroups classified by The Cancer Genome Atlas (TCGA), a single tumor could be shown to consist of a heterogeneous mixture of cells representing all of the different subgroups [21
]. Since one of the hallmarks of GBM is central hypoxia, a higher degree of heterogeneity can be expected, especially at the edge of the tumor. Here, we find regions showing different levels of vascularization and oxygen conditions. Moreover, the function of MCT1 and MCT4 is still under discussion, since MCT1 has been shown to be involved in lactate uptake [4
] and also lactate efflux [4
]. Since MCTs allow passive transport, their functions also rely on intra- and extracellular lactate levels and pH gradient [20
]. In general, previous studies have shown a complex influence of the microenvironment on the gene expression of MCT1 and MCT4. For example, hypoxia was shown to cause the up- or downregulation of MCT1 [4
]. Furthermore, the source of the marker expression needs to be regarded, since previous studies have postulated a reverse Warburg effect. In the reverse Warburg effect, stroma cells from the microenvironment produce lactate via aerobic glycolysis, so that the lactate can be used for further energy production in tumor cells. In this context, MCT4 has been shown to serve as a lactate exporter, whereas MCT1 serves as a lactate importer [25
In contrast to MCT1, GFAP showed a statistically significantly increased expression at the tumor edge compared with the center. Since GFAP is the principal component of intermediate filaments in astroglial cells and high-grade gliomas seem to lose GFAP expression [28
], the increased expression could be caused by the lower density of tumor cells at the edge of the tumor as a zone of tumor infiltration. Our observation of a significantly increased expression of the EMT marker β-catenin at the edge of the tumor is in line with previous studies, which have shown a higher expression of β-catenin, particularly at the invasive front of GBMs [29
]. In addition, an increased expression of the stem-like cell markers OCT4 and KLF4 was detectable at the edge of the tumor compared with the center. Most studies postulate the perivascular niche, particularly in the subventricular zone and the hippocampus, and the hypoxic/perinecrotic niche to be distinct regions where stem cells are enriched. However, recent studies have shown tumor stem-like cells to be located in the invasion niche, found at the tumor periphery of GBM [28
]. In particular, GBM cells located at the resection margin were shown to proliferate rapidly, and to be more invasive than GBM cells at the center of the tumor [31
]. With regards to the dormancy phenomenon and the slightly increased expression of the dormancy markers EPHA5 and IGFBP5 at the edge of the tumor observed in our study, the only connection published to date is an induction of MCT4 and dormancy markers by hypoxia [5
]. The current understanding of GBM suggests that up to 50% of the tumor mass is generated by the tumor stroma. These include endothelial cells, pericytes/mesenchymal stem cells, immune cells, and glial cells [34
]. Since cellular entities other than tumor cells can also be the source of the MCT expression, the cellular composition of the tumor at the center and edge also influences their gene expression levels. For example, immune cells have been shown to express MCT1 and MCT4 [20
The plasma membrane location and the activity of MCT1 and MCT4 were both shown to be regulated by co-expression with the same chaperone CD147 (basigin) [36
]. Thus, an expression of both markers in the same cell seems to be explainable. A co-expression of MCT1 and MCT4 has also been shown to be present in breast cancer cells [37
]. In our previously published study, we observed an expression of β-catenin and vimentin in GBM cells [38
]. A co-staining of both markers with MCT1 and MCT4, which we found in this study, was thus expected. Additionally, in the case of the stem-like cell marker KLF4, co-expressions were found with MCT1 and MCT4. This finding is supported by a study in which an increased lactate transport via monocarboxylate transporters (MCT1-4) was found in glioma stem-like cells [39
]. Moreover, previous studies showed an upregulation of MCT4 in GBM neurospheres [5
Our observation of an expression of MCT4, but not MCT1, in dormant cells could imply that this cell type also contributes to the reverse Warburg effect. Since dormant cells are in cell cycle arrest, they themselves have no high energy demand.
As expected, we were able to show an expression of MCT1 and MCT4 in GFAP-positive cells as a marker of astroglial origin. Moreover, only MCT1 was expressed in many vWF-positive endothelial cells. In support of our results, an expression of MCT1 was already described in previous studies [4
]. Lim et al. postulated a primary expression of MCT1 in endothelial cells [5
Since MCT1 and MCT4 have an important contribution to the maintenance of glycolytic metabolism and consequently tumor cell survival, they depict an interesting aim for targeted therapy. The effectiveness of inhibiting the activity and expression of MCT1 was already shown using in vitro and in vivo GBM models [4
]. Besides a reduction of tumor mass, a significant decrease in the number of blood vessels around the tumors was found in the treated group. In addition, a synergistic effect between the best described MCT inhibitor in literature, primary inhibiting MCT1, α-cyano-4-hydroxy-cinnamic acid (CHC) and temozolomide (TMZ) was observed, hence tumor cells could be sensitized to TMZ by this pretreatment. It has to be emphasized that the effect of the used inhibitor was dependent on the metabolic state of the cell line used, showing most efficiency particularly in more glycolytic cells [4
]. In addition, MCT inhibition yielded promising results in the case of glioma stem cells (GCS) [6
]. Since MCT1 was shown to be upregulated in GCS, the selective inhibition of MCT1 by AR-C117977 especially decreased the viability of GSCs, compared with that of non-GSCs. In fact, the treatment reduced cell viability at lower concentrations in GSCs than in non-GSCs [6
]. A phase I clinical trial of the MCT1 inhibitor in patients with advanced cancer (NCT01791595) is already ongoing in the United Kingdom. Expected side effects, due to the ubiquitous expression of the MCTs, have not been detected to date [41
Depending on the metabolic and molecular profile of the tumor (metabolic state, stemness and dormant properties), the administration of an antimetabolic drug targeting MCTs could be verified.
In this regard, our observation of a subgroup, showing a lower MCT1 expression at the edge of the tumor compared with the center combined with an induction of EMT, stemness and dormancy markers at the invasive front and a high regional concentration gradient of lactate becomes interesting. In this group, the response to a MCT1 inhibitor might be limited by the aggressive properties of the invasive front of the tumor, leading to tumor recurrence. On the other hand, tumors showing a more homogenous distribution of the markers might profit more from this targeted therapy. In the previous work of our group, we found the chemotherapeutic agent AT101 combined with TMZ to be an effective treatment strategy against dormant glioma cells in vitro [7
]. Depending on the molecular profile, a combination of this therapy with metabolic targeting could be another treatment option. Our observation of an expression of MCT4, but not MCT1, in dormant cells could also imply that patients with a higher percentage of dormant tumor cells also profit from inhibiting this transporter.
Furthermore, our findings indicate a possible connection between the distinct regional gene expression patterns of MCT1, EMT, stemness and dormancy markers with the regional lactate concentration. Recent examinations using hyperpolarized carbon-13 MRSI could also show a significant correlation between the lactate to pyruvate ratio and MCT1 expression in breast cancer [43
]. Definitely, an exclusive contemplation of the regional lactate concentration will not allow any conclusion to be drawn concerning the molecular features of the tumor. Nevertheless, the development of the MRSI technique delivers more and more insights into tumor metabolism. In future, it might be possible to use the information gained by MRSI for the evaluation of the use of an antimetabolic treatment.
In addition, MRSI can monitor the response to treatment, as was already demonstrated for the efficiency of TMZ in GBM in vitro and in vivo [44
]. Concerning the MCT inhibitor CHC, MRSI analysis indicated distinct changes in the brain metabolite profiles, due to the application of the metabolic inhibitor, offering the option to monitor the therapeutic effects of this agent as well [42
Of course, the number of patients included in the study was relatively small and surely limits the transferability of our results into a clinical context. The already known strong heterogeneity of the disease makes it even more difficult to detect patterns in the gene expression profile of the different markers.
Despite some limitations, this work provides a detailed look at the location of MCT1 and MCT4 in GBM, revealing clear connections with EMT, stemness and dormancy processes. In order to better understand the exact function and role of MCT1 and MCT4, further research is required. In particular, a further investigation regarding a connection with dormant cells as a further examination of the microenvironment would be of great interest.
Despite all research efforts, glioblastoma multiforme still remains an uncurbable disease. Especially, metabolic reprogramming to an aerobic glycolysis, as well as EMT, stemness and dormancy phenomenon play crucial roles in tumor progression and therapy resistance.
A consideration of the local distribution of associated markers revealed a significant downregulation of the lactate transporter MCT1 at the edge of the tumor, whereas the gene expression of the EMT marker β-catenin and the stemness markers KLF4 and OCT4 was induced at the edge of the tumor. A further analysis concerning the gradient of the marker expression from center to edge of the tumor revealed different subgroups with distinct expression patterns. A strong decrease in MCT1 expression and increase in EMT, stemness and dormancy marker expression at the edge of the tumor compared with the center correlated with a high lactate gradient between the different regions. The higher concentration of lactate in the center of the tumor compared with the edge was measured in vitro (lactate ELISA) and in vivo (MRSI). Beside revealing different cell types to be the source of MCT expression, a so far unknown relation between MCT4 and the dormancy phenomenon was found by immunofluorescence double-staining.
Since MCTs have become a promising target of therapy in GBM, our results point out that attention should be paid to the connection between metabolic reprogramming with EMT, stemness and dormancy phenomenon, which might contribute to the resistance against antimetabolic drugs.
The found subgroups with distinct patterns of gene expressions imply, that beside a molecular screening to evaluate a possible response to targeted therapy, a combination with agents against the described tumor promoting processes might be reasonable. In addition to the monitoring of therapy response, a further development of the MRSI technique might contribute to select patients with an assumable greater benefit from antimetabolic treatment in the future.
Altogether, understanding the mechanisms involved in generating the aggressive properties of glioblastoma multiforme is imperative to improve the disastrous prognosis of this disease. More research and innovation are urgently needed.