It is well established that solid tumours, including prostate cancer, exist under fluctuating oxygen tension, during which they are intermittently exposed to hypoxia [1, 2]. Under hypoxic conditions, tumour cells primarily use glycolysis for energy, producing lactate, which is expelled to the tumour microenvironment, allowing tumours to continue their glycolytic activity [3, 4]. Recently, Sonveaux et al. showed that lactate, which is generally considered a waste product, is preferred over glucose by oxidative tumour cells as their primary energy source [5].

Monocarboxylate transporters (MCTs) have been shown to play an important role in various tumours [6]. However, since they facilitate the transport of lactate in and out of cells, their role in this stromal/epithelial cell symbiosis is also attracting interest. MCT1 is a high-affinity transporter and its expression seems to be regulated by multiple signalling pathways, micro-environmental parameters, changes in substrate concentration and pH [7]. MCT4 is a low-affinity transporter, which is abundant in highly glycolytic muscle cells and is one of the many target genes of hypoxia-inducible factor 1 alpha (HIF-1α) [8]. Other targets of HIF-1α include glucose transporter-1 (GLUT-1), the main transporter involved in glucose uptake [9, 10]; lactate dehydrogenase V (LDHV), which is responsible for the conversion of pyruvate into lactate; pyruvate dehydrogenase kinase isozyme 1 (PDK1), which is responsible for the phosphorylation and consequent inactivation of pyruvate dehydrogenase (PDH); and carbonic anhydrase IX (CAIX), a hypoxia-related protein involved in pH regulation [11]. Alpha-methylacyl-CoA racemase (AMACR), pristanoyl-CoA oxidase (ACOX-3) and D-bifunctional protein (DBP), are also important fatty acid oxidation-related proteins in prostate cancer [12, 13], and we also included them in our analysis.

The importance of a lactate shuttle system between cancer cells and surrounding stroma has been described in various tumour types [14–16], but its significance in prostate cancer is not clear [17, 18].

In this study, we aimed to identify a metabolic interaction between CAFs and prostate cancer cells, by analysing the expression of key metabolism-related proteins in CAFs in relation to prostate cancer using prostate tissue samples. We also assessed the clinico-pathological significance of this expression to investigate a possible CAF signature for PCa progression.

We observed obvious differences between the expression of key metabolism-related proteins in CAFs and tumour cells (Figure 1). MCT1, MCT4, LDHV, PDK1, GLUT-1, GLUT-12, CAIX, AMACR, ACOX-3 and DBP were differentially expressed between stromal and epithelial cells, while MCT1, LDHV, GLUT-1, GLUT-12, AMACR, ACOX-3 and DBP were exclusively expressed in prostate cancer cells. MCT4 and CAIX were expressed more strongly in CAFs, and PDK1 stained both malignant glands and CAFs.

Additionally, we assessed whether fibroblasts exhibited differences in protein expression across different stages of malignant transformation by analysing the expression of the same proteins in fibroblasts surrounding benign glands (benign-associated fibroblasts; BAFs), PIN-associated fibroblasts (PAFs) and CAFs. Figure 2 shows stacked bar graphs representing the expression of MCT4, PDK1 and CAIX in more detail, since these proteins were the ones exhibiting a clear expression in fibroblasts surrounding both benign and malignant glands. A statistically significant increase in both MCT4 and PDK1 expression in CAFs compared to BAFs was observed (both p < 0.001). Curiously, CAIX was also observed in fibroblasts surrounding non-neoplastic glands (benign glands and PIN lesions) (see also Figure 3).

Key metabolism-related proteins in fibroblasts and prostate glands across malignant transformation, i.e. from BAFs to PAFs and finally CAFs, were investigated. MCT1 was clearly expressed in the plasma membrane of prostate glands, but not the surrounding stroma. In contrast, the expression of MCT4 increased in fibroblasts with increasing degree of malignant transformation, but not in the prostate glands. PDK1 expression was detected in both glands and stroma, whereas CAIX was only detected in stroma, with no staining in prostate glands (Figure 3).

Several research groups have recently focused on the role of CAFs in the progression and metastasis of prostate cancer, showing that a dynamic interaction between stroma and epithelium might play a critical role in this progression [14, 15, 22–25]. Thus, the essential role played by the cross-talk between stroma and epithelium in carcinogenesis and prostate cancer progression has been increasingly recognised. In this work, we provide evidence for the possible metabolic co-operation between cancer cells and the surrounding fibroblasts by examining the expression of major proteins involved in cellular metabolism. In particular, we focus on differences between cancer cells and tumour-associated fibroblasts, as well as between fibroblasts in different stages of malignant transformation, and examine the possible clinico-pathological significance of the expression of these proteins.

By categorising the protein expression of stromal cells associated with prostate cancer, we describe a compartment that is not well studied and will contribute to an improved understanding of prostate cancer. We observed significant differences between CAFs and tumour glands with respect to the expression of key metabolic proteins. In particular, CAIX and MCT4 selectively labelled cancer associated fibroblasts in contrast to malignant glands, where CAIX and MCT4 were only present in very few cases. On the other hand, a distinct, strong membranous expression of MCT1 was consistently observed in cancer cells, suggesting a role for MCT1 in the transport of lactate into tumour cells from the acidic extracellular matrix, suggesting that lactate might be used as a fuel by oxidative cancer cells. We also observed that proteins involved in fatty acid oxidation, such as AMACR, ACOX-3 and DBP, were restricted to the tumour cells, which is consistent with the presence of a metabolic pathway different from glycolysis, and compatible with oxidative phosphorylation in prostate cancer cells. It is important to note that fatty acid oxidation is already considered a major source of acetyl-CoA for the Krebs cycle [13], which further supports our hypothesis.

The expression levels of GLUT1, a key glucose transporter, define the rates of glucose influx into the cells. In the present study, CAFs did not show GLUT1 or GLUT-12 expression, and LDHV was also difficult to detect. However, this possibly reflects the limits of the immunohistochemical technique to detect these proteins at the baseline concentrations present in CAFs. Indeed, we have previously found very few cases positive for GLUT-1 and GLUT-12, and this expression was not present at the plasma membrane, suggesting a low level of activity of these proteins in prostate cancer cells (unpublished results). Thus, assessment of other GLUT isoforms may be worthwhile.

Interestingly, we also observed that protein expression of MCT4, PDK1 and CAIX in prostate fibroblasts changes during malignant transformation, suggesting that the existing stroma might also suffer alterations and play a role in this metabolic adaptation of cancer cells beyond the well-studied role of newly formed stroma.

From the above immunohistochemical findings, it seems that well-organised metabolic regions composed of tumour cells and CAFs may contribute to the ability of the tumour to overcome the adverse microenvironment.

Our hypothesis is in agreement with those of Fiaschi et al.[17], who describe the metabolic reprogramming of CAFs towards the Warburg phenotype as a result of contact with prostate cancer cells. Using in vitro studies, they showed lactate production and efflux by de novo expressed MCT4 in CAFs and also demonstrated that, upon contact with CAFs, prostate cancer cells were reprogrammed towards aerobic metabolism, with an increase in lactate uptake via the lactate transporter MCT1. Furthermore, pharmacological inhibition of MCT1-mediated lactate uptake dramatically affected PCa cell survival and tumour outgrowth. However, in this study, no data regarding clinico-pathological associations were shown, and few cases were assessed. These findings are in contrast with others ([18], which describe an energy recycling path between the aerobic stroma and the anaerobic cancer cells within the framework of the Warburg effect. These conclusions are based mainly on the observation that LDH1 is evidently expressed in CAFs, and the presence of MCT1 in prostate cancer cells was attributed to its role in lactate efflux and not its uptake. We recognise the importance of assessing LDH1; however, in our study we assessed for the first time MCT4 and CAIX as important markers of hypoxia in a larger cohort. Our findings corroborate the work of Whitaker-Menezes et al.[16], who described a “reverse Warburg effect,” where CAFs undergo aerobic glycolysis to produce lactate, which is subsequently used as a metabolic substrate by adjacent cancer cells. In this model, “energy transfer” or “metabolic coupling” between the tumour stroma and epithelial cancer cells fuels tumour growth and metastasis via oxidative mitochondrial metabolism in anabolic cancer cells. We believe that this is also the case in prostate cancer, although more studies are needed to demonstrate this.

Also, we assessed important clinico-pathological parameters and found significant associations with poor prognosis, raising once more the possible role of CAFs in disease management. We believe that these changes are likely to be a by-product of tumour biology with further influence on patient outcomes that need to be explored more deeply.

In summary, we found differences between prostate cancer cells and CAFs using tissues from 480 patients, showing elevated expression of MCT4 and CAIX in CAFs and demonstrating for the first time that the concomitant expression of MCT1 in tumour cells and MCT4 in fibroblasts in the same tissue is clinically significant, and associated with poor prognosis. Indeed, the stromal expression of hypoxia-regulated proteins appears to be prognostic of poor outcome in prostate carcinomas, suggesting that tumour hypoxia may influence tumour-associated stromal cells in a way that ultimately contributes to patient outcome. Figure 4 shows a schematic representation of the lactate shuttle between CAFs and PCa cells to illustrate the hypothesis presented in our work.

In summary, we show for the first time that there is a clinico-pathological significance for the MCT1/MCT4 lactate shuttle in prostate cancer. In fact, it seems that the stromal expression of hypoxia-regulated proteins is an adverse prognostic factor in prostate carcinomas, suggesting that tumour hypoxia may influence tumour-associated stromal cells in a way that ultimately contributes to patient prognosis.