Several studies suggested that the expression and actions of the endocannabinoid/vanilloid receptor TRPV1 are altered during the development of tumours and of several cancers [30,31,32,33,34,35,36]. Additionally, it is reported that EC cells in culture enter apoptosis in response to AEA and other ligands through a TRPV1-dependent pathway [16]. Following on from these observations, it was reasoned that similar changes might exist in EC tissues in vivo, but up to the time of writing, this hypothesis had not yet been tested. Hence, the transcript and the protein expression levels of TRPV1 were compared between human atrophic endometria and the two histological types of EC [6]. That TRPV1 expression may also be related to functional components of the ECS and those of cell growth, survival and apoptosis was likewise evaluated.

Surprisingly, TRPV1 transcript (mRNA) levels were found to be highest in the normal atrophic endometrium and greatly reduced in the two types of EC tissue (Figure 1). The level of TRPV1 transcript in normal tissues was not converted efficiently to protein even though significant amounts of TRPV1 protein could be detected in the atrophic control and in normal endometrium and skin (Figure 2 and Supplemental Figure S1). These data suggested that either the antibodies used were not effective in IHC or that in EC there is disconnect between mRNA (transcript) levels and protein expression levels. Accordingly, we performed additional experiments to show the validity of the antibodies in IHC (Supplemental Figure S1) and confirmed TRPV1 staining in the proliferating keratinocytes of the stratum basale of human skin and in the proliferating glandular cells of the endometrium, indicating the antibodies to be useful in IHC and in actively dividing cells. Therefore, our conclusion is that in EC, changes in gene expression at the transcriptome level does not always lead to changes in gene expression at the proteome level (i.e., the remaining 50% of TRPV1 transcript is not efficiently translated into TRPV1 protein). Indeed, this paradigm has recently been discussed as a key point in general carcinogenesis [37] with a disconnect between mRNA levels (transcripts) and protein expression for numerous genes. However, although the relative levels of TRPV1 protein were significantly lower in both the oestrogen-dependent Type 1 EC samples and in the oestrogen-independent Type 2 EC samples, there was enough DAB staining to observe a clear difference in staining patterns in the different grades of Type 1 EC samples. Furthermore, TRPV1 staining in grade 1 EC (the lowest grade of Type 1 EC) was observed to be more in the glands than in the stroma, with the stroma being nearly devoid of staining. By contrast, in grade 2 EC samples (the intermediate grade of Type 1 EC), the staining was observed to be the opposite, with more intense staining being observed in the stromal region than in the glands. Nevertheless, TRPV1 protein levels were significantly lower in all the various grades of Type 1 (1, 2 and 3) and both forms of Type 2 EC (sarcoma and carcinosarcoma). This could be a major cause for concern in our study because our mRNA samples may have been contaminated with stromal elements that could have caused the lack of correlation between transcript and protein levels. However, we were very careful to ensure that only cancer tissue was micro-dissected free during the processing of the biopsies, and so the differential staining in stroma and gland would not be a contributory factor in our analyses. To ensure that was the case, both stromal and glandular staining levels in the histomorphometric analyses were combined to provide a more holistic score akin to that obtained from the RT-qPCR studies, and reasonable correlations between transcript and protein levels were obtained (Figure 7). The alternative hypothesis is that different tumours have different expression levels of transcript and protein and so provide a more variable observation. What is clear, however, from these studies is that TRPV1 protein expression is reduced in both Type 1 and Type 2 EC.

The dysregulation of both ‘classical’ (CB1 and CB2) [13,15,17] and ‘non-classical’ (TRPV1 and GPR55) [21,28] endocannabinoid receptors in EC may provide insights into the variable effects of endocannabinoids observed in different tumour models. For example, in certain tumours like gliomas, that abundantly express cannabinoid receptors, plant-based cannabinoids (phytocannabinoids) have been reported to be more effective in inducing tumour cell death. In other tumour types where cannabinoid receptor expression is absent or low, such as in breast cancer, phytocannabinoids may promote tumour growth and metastasis, or inhibit cytotoxicity, possibly by suppression of the anti-tumour immune response [38]. These observations would align with the reduced expression of CB1, CB2 and TRPV1 receptors in EC, suggesting a potential suppression of the anti-tumour immune response occurs in such individuals [38]. This could be a general phenomenon that extends to other types of cancer, as evidenced by studies in human colorectal cancer where the cells have lower CB1 mRNA and exhibit increased tumour growth [39]. Similarly, reduced cannabinoid receptor expression is associated with an increase in apoptosis in astrocytoma, whereas no apoptosis occurs in astrocytoma cells with high receptor expression [40]. Furthermore, CB1 knockout, but not CB2 knockout APCmin mice (a model of colorectal cancer) have an increased risk of intestinal polyp development, a phenomenon that can be replicated by treatment with CB1 antagonists [39]. These receptor reductions are similar to the findings shown here, where the TRPV1 expression was reduced in the malignant tissues when compared to the benign tissues, suggesting that increased apoptosis may be present in EC.

Conversely, elevated cannabinoid receptor levels have been observed in certain malignancies including prostate cancer, pancreatic cancer and non-Hodgkin lymphoma [31,41], where higher CB1 expression correlates with increased disease severity and poor prognosis in prostate and pancreatic cancer [42,43]. Similarly, TRPV1 is overexpressed in both human prostate cancer [31] and squamous cell carcinoma of the tongue [30]. These observations highlight the variability in cannabinoid or TRPV1 receptor expression patterns across different cancer types, suggesting that receptor expression is likely to be tumour-dependent.

In addition to the effects mediated through cannabinoid receptors, evidence suggests that AEA also can induce cytotoxic effects via a cannabinoid-receptor-independent mechanism [44], with AEA acting as an agonist on TRPV1 receptors. These observations lend support to the hypothesis that loss of CB1 and TRPV1 receptors would result in a loss of the cytotoxic effects of AEA in such tissues. These findings suggest that reduced expression of CB1 and CB2 receptors and the total loss of TRPV1 may play a key role in the pathogenesis of EC. In such cases, endometrial tissues may become less responsive to the repressive effects of circulating and locally produced ligands, particularly AEA [32].

In EC cell lines, the amounts of several proteins also did not correlate well with transcript levels [44]. For example, the Hec50co cells (that represent Type 2 EC tissue) were reported to have higher levels of CB1, CB2 and TRPV1 transcript than that of Ishikawa cells (that represent Type 1 EC tissue), yet in Western blot only the CB2 protein level appeared to be greater [44]. Whether these differences were significant or not is difficult to determine since the authors did not perform those comparisons. Nevertheless, these data suggest that the disconnect between transcript levels and protein in EC tissues continues through into cell lines, supporting the idea that protein expression does not always follow transcript changes in EC (Figure 7) and in other tumours [37].

Because the Ishikawa cell line appears to be a good model of Type 1 EC to study the effects of endo-cannabinoids/vanilloids on cell survival [16], we performed experiments to determine what effects AEA and capsaicin (a TRPV1-specific agonist) have on Type 1 EC cell survival. The data indicated that AEA only inhibited Ishikawa cell survival at the highest concentrations, with the effect significantly maintained for up to 24 h (Figure 8). These data are qualitatively similar to those previously published [16]. When treated with capsaicin for equivalent times and with similar concentrations, Ishikawa cells seemed unaffected and only the highest doses at the longest time-point had any effect on the cells, which showed increased cell numbers (Figure 8). These data are contrary to those published by Fonseca et al. [16], where they concluded that AEA and other ligands were acting through the same receptor (TRPV1) to induce apoptosis. If that were the case, then it is reasonable to assume that if AEA is inducing apoptosis through the TRPV1 receptor, then another TRPV1 agonist would have the same effect. The reason that capsaicin did not induce cell death in our study is not known, but points to a possibility that AEA and capsaicin bind to different parts of the TRPV1 receptor, one on the intracellular surface near the T1-TM region [45] and one on the extracellular surface very close to the plasma membrane lipids in the S-region [46,47], respectively. Alternatively, the p53 mutation that Ishikawa cells possess [48] prevented activation of BAX-mediated apoptosis [49]. In other cell types, AEA needs to be transported into the cell in order to have its effect [45], and then it needs to take a different route to its final binding position, which is unaffected by 5′iodoresiniferatoxin (iRTX), which was used in the previous study [16], so the conclusion that different parts of the TRPV1 protein are being targeted by AEA and capsaicin is not unreasonable [50]. Nevertheless, the question remained concerning the cell viability effect of AEA in EC: is it due to increased apoptosis as suggested in a previous study [16] or due to reduced cellular proliferation?

From the studies of Fonseca et al. [16], it was clear that in vitro the effect of AEA was induction of apoptosis. To extend those observations, we examined the expression of cell proliferation and apoptosis markers in the same biopsies used for the TRPV1 protein expression reported herein. These samples were also taken from women who also had tissue and plasma levels of AEA measurements [14], corresponding FAAH and NAPE-PLD H-scores [32], CB1/CB2 H-scores [13] and GPR55 H-scores [28]. Since tissue and plasma AEA levels were increased in these EC patients when compared to the atrophic endometria patients, then this would represent a continuous increased AEA level in in vitro Ishikawa cell experiments.

As expected, all of the EC samples showed increased expression of the Ki-67 proliferation marker (Figure 5). Indeed, Ki-67 immunohistochemical staining is often used to classify the molecular subtype of EC present [6,51]. These data were supported by the cell culture experiments performed herein and by Fonseca et al., where Ishikawa cells continue to proliferate in low serum conditions [16] or even in the absence of serum (this study). This suggests that even in the presence of AEA or other TRPV1 agonists, the overarching cell cycle signal remains ‘proliferate’. The demonstration that levels of the pro-apoptotic marker BAX increased in all parts of EC tissue (Figure 4), especially in the glandular epithelial cells, supports the study of Fonseca et al. [16]. Although BAX expression also increased in the stroma but not significantly in the Type 2 EC samples is probably a reflection of the low sample numbers. By contrast, the expression of the anti-apoptotic marker Bcl-2 was statistically unaffected by EC, but sub-analysis of the different types of EC revealed the reason for this was the elevated expression of Bcl-2 in the Type 1 EC glands and stroma (Figure 5). There was a significant reduction in Bcl-2 expression in the Type 2 EC samples, supporting the idea that loss of anti-apoptosis results in the generalised apoptotic effect described in EC [16]. This conclusion was supported by measuring one of the ‘hallmarks of apoptosis’, the BAX:Bcl-2 ratio [52], where it was increased in EC tissue, but only in the glandular epithelial cells and is most acute in the Type 2 EC samples (Figure 6), which is also the most aggressive form of EC that often forms metastatic lesions [6,51]. The BAX:Bcl2 ratio is important because these two proteins work together to control apoptosis, with a homodimer of BAX resulting in pore formation in the outer membranes of mitochondria [53,54], whilst the BAX-Bcl2 heterodimer prevents apoptosis by preventing BAX homodimer binding to the mitochondrial membrane and p53 [54,55], an anti-tumour protein important in preventing cell division [56]. These data suggest that although EC tissue is proliferating, even in the presence of high AEA tissue levels and plasma levels [14], this effect is only partly due to dysfunctional apoptotic signalling and only in the Type 2 EC patient. In patients with Type 1 EC, the elevation of NAEs, such as AEA, is working through a non-intrinsic apoptotic mechanism, possibly utilising a different receptor (Figure 9). Indeed, we recently suggested that in Type 1 EC patients, the key regulator of glandular epithelial cell proliferation is the increased expression of the orphan receptor GPR55, which also binds these ligands [28], possibly through dimerisation with CB1 [57]. The expression of the TRPV1 receptor in EC is evidently significant for understanding the biology of EC tumours, yet further investigation remains imperative. This conclusion is based on the known functions of TRPV1 in cellular homeostasis through the intrinsic apoptotic pathway that involves Ca²⁺ ion influx (Figure 9) and the lack of positive correlation between TRPV1 expression and the BAX:Bcl2 ratio in our patient samples (Supplemental Tables S2 and S3). In the normal situation, activated TRPV1 would result in Ca²⁺ ion influx to affect the intrinsic apoptosis pathway and damage mitochondrial membrane function, resulting from an increase of the BAX:Bcl2 ratio, going on to activate the caspase-dependent apoptosis cascades [58]. In our clinical Type 2 EC samples, the BAX:Bcl2 ratio is inversely correlated with TRPV1 expression, suggesting the intrinsic apoptotic pathway is not active here (Figure 9: [58]). This means that the normal function of BAX (i.e., to bind to mitochondrial membranes to induce apoptosis and to p53 thus prevent anti-tumour events), is no longer functional, thereby allowing cellular proliferation. This was confirmed by the increase in Ki-67 expression (Figure 3), since Ki-67 protein is only produced by cells undergoing active cell division. The observed loss of TRPV1 protein and function in both types of EC directly results in a loss of proliferation control. In the normal proliferative phase endometrium and Type 1 EC, the main cellular proliferation signal is 17β oestradiol, where BAX:Bcl2 ratios are reduced, suggesting that the main regulator of the BAX:Bcl2 ratio in Type 1 EC is 17β oestradiol and not activation of the TRPV1 through an intrinsic apoptosis signal [59]. While there is new evidence supporting the involvement of TRPV1 activation by the vanilloid capsaicin and by the endocannabinoid AEA in regulating cancer cell growth and disease progression in other gynaecological cancers [60], similar studies in EC models suggest that although these agonists might induce apoptosis, an in vivo impact that is unlikely to be substantial in EC [59]. This is primarily due to the considerably lower expression of TRPV1 in EC compared to tissues such as skin, cervix or atrophic or normal proliferating endometrium (see Supplemental Figure S1). Similarly, because the classical cannabinoid receptors CB1 and CB2 also show decreased expression in EC, then AEA is probably activating GPR55 receptors directly instead [28]. Therefore, the elevated levels of plasma and tissue AEA found in patients with EC [14] must affect cell numbers through a non-TRPV1 receptor mediated pathway, perhaps one that involves the TRPV2 receptor, as suggested by Marinelli et al. [61]. In this situation, a mixed Type 1/Type 2 EC cell phenotype would have to be active for AEA to act as an endovanilloid.

In this case, AEA could work co-ordinately with other endocannabinoids (such as 2-AG) through the combination of low TRPV1 activation (to initiate a cell death or apoptosis pathway), whilst 2-AG (or the phytocannabinoid CBD) activates a different part of the apoptosis or autophagy pathways [61]. Clearly, as TRPV2 receptors are expressed in both normal and cancerous tissues, more research in this area is justified. Since TRPV2 has also been shown to be a prognostic marker in Type 2 EC [52], then its expression in Type 1 EC needs to be established. Additionally, since other members of the vanilloid receptor family have not been characterised in EC, then future studies should establish their expression levels and functions especially in EC development and potential in EC patient treatment with TRPV2 receptor-specific antagonists.

In that respect, it is hoped that innovative targeting of TRPV1 receptors [46] or TRPV2 receptors [61] with sub-receptor specific agonists or antagonists (depending on the tissue under investigation) may contribute to a novel tumour-suppressive treatment strategy for EC and other gynaecological cancers. The knowledge that TRPV1 is a ligand-gated Ca²⁺-permeable ion channel, which is usually activated by noxious stimuli such as acid and heat, and is involved in transmission and modulation of pain [69], and may explain why uterine pain is not detected in patients with early-stage EC, when TRPV1 expression is decreasing. Moreover, both AEA and acyl-dopamine are now considered to be endovanilloids because they bind to this receptor [69], and the concomitant increase in tissue AEA production (and other NAEs) we measured [14] may be a reflection of the loss of the receptor in the diseased EC tissue.

A limitation of the findings presented in this study is that we used only one EC cell line. It would be useful to use other cell lines to determine whether our data would be replicated. It is anticipated that our findings with other cell lines would provide robust supportive evidence on the hypothesised mechanism presented herein (Figure 9) or elsewhere [16].