Introduction

Pleural mesothelioma (PM) is a rare, asbestos-related, relatively chemo-resistant tumor that arises from the mesothelial pleural surface. PM is more common in males and the highest incidence is reported in the sixth and seventh decade of life. The most common PM histological subtype is the epithelioid one, followed by biphasic and sarcomatoid subtypes. Patients generally have poor prognosis, strongly related to histology and to the stage of the disease at diagnosis, with median survival for treated patients from 6 to 18 months for epithelioid and much lower survival rates for the sarcomatoid histotype [1]. Chemotherapy with pemetrexed (PEM) and platinum is the standard of care with improvement in overall survival of roughly three months, compared to single cisplatin agent [2,3,4].

PEM as a multi-targeted anti-metabolite inhibits multiple molecules of the folate metabolic pathway, especially thymidylate synthase (TS), an enzyme essential for DNA synthesis and repair [5]. TS is overexpressed in various cancer types and is associated with metastatic spread and reduced overall survival. Moreover, TS expression levels are upregulated following treatment with chemotherapy, including agents that inhibit its activity. In several studies, high levels of this enzyme seem to be correlated to reduced PEM efficacy in different tumor types, including mesothelioma, colon, lung, and breast carcinomas [6,7,8,9,10,11].

Genetic studies on PM reported a low prevalence of oncogene driver mutations and low tumor mutational burden, and the vast majority of recurrent mutations predominantly result in loss-of-function of tumor suppressors, including BAP1, TP53, CDKN2A, NF2, and LATS2 [12, 13]. The low frequency of genomic events in PM has moved the interest on epigenetic regulation of PM growth and progression with possible integration among different mechanisms [14].

Micro-RNAs (miRNAs) are a class of small (about 18–22 nucleotide long) non-coding RNAs that function in post-transcriptional regulation of gene expression [15, 16]. Micro-RNAs are expressed in physiological conditions in a cell- and tissue-specific-manner, but their aberrant expression in tumor tissues is associated with tumor-specific characteristics, thus supporting their potential role as diagnostic, prognostic or predictive biomarkers [17, 18]. Recently, a set of significantly downregulated (miR-874, miR-31, miR-203, miR-200a, miR-143, miR-200c, and miR200b) and upregulated (miR-139-5p, miR-210, miR-944, and miR-320) miRNAs were found in pleural effusion from PM patients [19] compared to other diseases (adenocarcinoma and benign pleural disease) and more recently several studies proposed microRNA as potential therapeutic target for resistant PMs [20,21,22]. In this context, miR-215 and miR-375 have been already reported as strong modulators of TS in different cancer cells [23], but no data are available about their role on targeting TS in PM. Only miR-215-5p was reported to interact with MDM2-p53 cell signaling and to be associated with poor prognosis in PM [24]. Furthermore, miR-215 and miR-375 are also known to be involved in epithelial-to-mesenchymal transition (EMT) [25,26,27] and their action was connected, among the others molecules, to ZEB1 (Zinc Finger E-Box Binding Homeobox 1) activity [28, 29]. EMT is both a physiological and pathological process related to embryonic developments [38]. In our series, among the 51 epithelioid PMs available for reticulin stain, 22/51 (43%) had a pattern characterized by the reticular fiber framework lost or with loose meshes surrounding large tumor nests (Loose, L-pattern) (Fig. 5a, b). Interestingly, in 29/51 (57%) PM cases with the same epithelioid morphology, the L-pattern was associated to variable extension of tumor areas with denser, fine-to-tight reticulin fibers surrounding individual or small group of tumor cells (Fine/Tight pattern) (Fig. 5c, d). Moreover, considering the epithelioid PM subgroup only, we found that those cases with a component of Fine/Tight reticulin pattern, as compared to the epithelioid PM with a pure L-pattern, were characterized by significantly either higher ZEB1 protein expression levels (Mann–Whitney p = 0.002) (Fig. 5e), and significantly higher TS protein expression levels (Mann–Whitney p = 0.01) (Fig. 5f), and lower miR-215 and miR-375 expression levels, even if not significant (Mann–Whitney p = 0.25 and p = 0.08, respectively. Data not shown).

Fig. 5
figure 5

Reticulin pattern, ZEB1, and TS protein expression in epithelioid PM subgroup. (a) PM with clear epithelioid morphology (hematoxylin–eosin stain, 100 × magnification) and corresponding (b) reticulin fiber stain of the same case showing a loss of reticulin fibers surrounding very large tumor cell nests (loose-pattern) (silver reticulin stain, 100 × magnification); (c) PM with epithelioid morphology (hematoxylin–eosin stain, 100 × magnification) and corresponding (d) reticulin fiber stain of the same case showing a loss pattern of reticulin fibers surrounding large tumor cell nests (loose-pattern, upper left angle) associated to a fine, more filled reticulin pattern surrounding individual or small group of tumor cells (fine/tight pattern, lower right angle) (silver reticulin stain, 100 × magnification). Graphic representations of (e) ZEB1 and (f) TS protein expression according to loose or fine/tight reticulin pattern

Full size image

miRNA and TS expression and morphology according to genetic subgroups.

Immunohistochemistry for BAP1 and p53 was performed in 62 cases (51 epithelioid, 4 biphasic and 7 sarcomatoid PM) with available material.

BAP1 nuclear expression was lost in 32/51 (63%) epithelioid, 1/4 (25%) biphasic and 1/7 (14%) sarcomatoid PM. No correlation was found between BAP1 alteration and miRNA expression, nor TS expression nor different epithelioid morphology (data not shown).

p53 complete loss or overexpression (altered status) was found in 7/51 (14%) epithelioid, 0/4 biphasic and 4/7 (57%) sarcomatoid (3 completely lost in tumor cells with positive internal control and 1 overexpressed). No correlation was found between altered p53 and miRNA expression, nor TS expression nor different epithelioid morphology (data not shown).

Survival analyses

Follow-up was complete in 57/71 (80%) of patients from 1 to 52 months. At the time of the study, 60/71 (85%) patients were died. Median OS was 10 months.

A little significant difference was found in survival curves between cases with low and high TS protein expression: lower levels of TS protein give a slightly better survival as compared to higher levels (12 vs. 8.5 months, log-rank p = 0.05). No significance for survival was found between cases with low or high expression of miRNA-215 and miRNA-375 or between cases with epithelial PMs with loose or fine-to-tight reticulin stain (Supplementary Fig. 3).

As regard genetic profile, BAP1 status showed a scarce correlation (log-rank p = 0.09, data not shown) with survival while no significance was found for p53 and survival. However, considering BAP1 negative cases only, no significance for survival was found between the high or low expression of miRNA-215 and -375 or between epithelial PMs with or without EMT morphology (data not shown); on the contrary, a significant better survival was found for those cases with BAP1 loss and TS low expression (17 vs. 10 months, log-rank p = 0.04) (Fig. 6), thus confirming the central role of TS protein in regulating PM progression.

Fig. 6
figure 6

Survival curves of cases according to BAP1 status and TS protein levels. A significant better survival was found for those cases with BAP1 loss and TS low protein levels (median survival 17 vs. 10 months, log-rank p = 0.03)

Full size image

Discussion

In this study, we demonstrated that (1) miR-215 and miR-375 are expressed in PM tissues in a significant histotype-dependent manner and inversely correlated with TS protein levels; (2) miR-215 and miR-375 induce a specific modulation on TS protein expression in PM cell lines, although this not implied changes in sensitivity to PEM in vitro; and (3) the regulatory activity of miR-215 and miR-375 could be involved into EMT in PM tissues via TS protein expression modulation.

Despite the recent improvements in targeted oncology therapy, no efficient option is still available for PM treatment, a part of recent advances in immunotherapy [39, 40]. One of the causes of therapeutic failure in treating PM is due to the development of resistance to standard chemotherapy, namely PEM-based, largely attributed to the increased expression of the TS enzyme, the main target of PEM [41]. Our group demonstrated that immunohistochemical levels of TS could be predictive of response to therapy in a selected series of PEM-treated PM patients. According to our study, patients with higher levels of TS tumor protein demonstrated poorer overall survival, while a low-to moderate TS level showed a better correlation with PEM-based response to therapy [42]. However, few evidence demonstrated a possible modulation strategy of TS both on PM and other cancers.

Micro-RNA-215 and -375 were recently found to target TS in other tumors [23], so, due to the important predictive role of TS in PM, we explored their potential role in targeting and modulating TS in PM. Micro-RNAs were already studied as modulator of TS expression in several cancer models [43], and specific miRNAs were described in PM diagnosis and prognosis [44, 45], but scanty evidences about therapeutic role of miRNAs were described in PM [46]. In particular, only mir-215-5p was recently reported to exerted significant cell killing by activating p53 function and inducing apoptosis [24]. On the other hand, at the best of our knowledge, no functional evidence on mir-375 were reported in PM.

In our consecutive series of PM tissues, the gene expression of miR-215 and miR-375 revealed a significant specific distribution of both miRNAs according to PM main histotypes: both miRNAs were more highly expressed in the better epithelioid than in the poorer biphasic and sarcomatoid histotypes, thus suggesting a correlation of the miRNA expression levels with a different program of neoplastic mesothelial cells differentiation. This result confirmed literature data by Siddiqui and coworkers who correlated miR-215 and miR-375 lower expression with less aggressive form of tumor and described them as negative modulators of EMT phenomenon [23, 47]. In the present series, TS protein levels were found significantly higher in biphasic and sarcomatoid than in epithelioid PM (Mann–Whitney p = 0.0013). This observation is not surprising given the role of TS in tumor cell cycle, the poorer histotype-related prognosis in PM and the high chemotherapy resistance of mesenchymal histotypes. Interestingly, TS protein and both miR-215 (p = 0.009) and miR-375 (p < 0.0001) expressions were overall significantly inversely correlated in the present series and a strong correlation between both miR-215 and miR-375 expression (Spearman test p < 0.0001, r = 0.71) was also found, in line with the hypothesis of a same target for these two miRNAs to TS.

Based on these results, we decide to induce a transient overexpression of miR-215 and miR-375 in PM cell lines for evaluating their activity on TS modulation in vitro: both transfected cell lines showed a direct decrease of TS protein expression, thus confirming that TS is a target of miR-215 and miR-375 and supporting the significant inverse correlation between miRNAs and TS protein expression found in PM tissues. Unfortunately, viability assay failed to reveal a significant growth inhibition in PEM-treated miRNAs overexpressing cells, respect to untreated control. In a previous study, Abu Lila et al. obtained an improvement of chemosensitivity to PEM in vitro and in vivo reducing TS expression by shRNA transfection in PM models [11]. This data was not confirmed by our study, but a different experimental model was used. Moreover, TS gene silencing in a tumor cell could reduce the tumor cell proliferation per se, for the cell cycle role of TS in tumor cells. Our aim was to modulate TS, as we obtained with the transient transfection, but, besides TS modulation, other complex signaling pathways could take part on regulating cell growth response to PEM.

Different reasons could be at the base of this event. Firstly, it is known that PEM has multiple targets: other than TS, it could also inhibit dihydrofolate reductase (DHFR), and glycinamide ribonucleotide transformylase (GART) enzyme activity [5], so its inhibitory effect on PM patients could be due also to other secondary mechanisms. Secondly, the capability of cancer cells to escape the drug toxicity could depend on several complex mechanisms other than TS protein action [48, 49].

Finally, to better investigate the role of miRNA in PM and clarify their different expression distribution in PM histotypes, particularly in epithelioid PM, we investigated a possible involvement of miR-215 and miR-375 into PM EMT process. Published data demonstrated that ZEB1 is one of the hallmark of EMT [29] and one of its activity is related to repress miR-375 [23]. We investigated ZEB1 immunohistochemical expression in PM tissues and we found that ZEB1 was significantly (Mann Whitney, p = 0.0012) overexpressed in the sarcomatoid histotype (H-score mean value: 245.9) compared to the epithelioid PM (H-score mean value 138.11) confirming already published data [33]. However, in our tissue series, no correlation between miRNA and ZEB1 protein expression was found, thus suggesting that epithelial or mesenchymal differentiation could be regulated not by ZEB1 activity on these miRNAs in PM. Interestingly, we found a variable distribution of the ZEB1 levels in epithelioid PM subgroup ranging from undetectable to high levels. To test if this distribution could represent a different propensity of higher ZEB1-expressing epithelioid PM cases to progress to a mesenchymal histotype, we classified PM morphological features by means of reticulin stain, to better define different histological component. Recently, specific reticulin patterns were described in different PM histotypes [50], with a loose pattern associated to epithelioid and a tight pattern to sarcomatoid histotype. In our series, the pure epithelioid subgroup was characterized by a completely loose pattern of reticulin fibers, but in a half of epithelioid PM cases, reticulin stain highlighted the presence of a tumor component with fine-to-tight reticulin pattern associated to the pure epithelioid. We hypothesized that this feature could represent an initial transformation to a more mesenchymal type, probably representing cases with initial “transformed” patterns [51]. Interestingly, in this group, both higher protein levels of ZEB1 and TS expression were found. Recently, Siddiqui et al. [23, 47] described a role of TS, other than ZEB1, to drive EMT in non-small cell lung cancer, with direct correlation between the two molecules. In our PM series nor direct correlation between TS and ZEB1 was found, neither inverse correlation between miRNAs and ZEB1. On the contrary, those epithelioid PM with a fine-to-tight reticulin pattern associated component were characterized by a significantly higher TS levels and miR-375 and miR-215 lower expression levels, confirming these miRNAs as strong modulator of TS in EMT phenomenon [23].

Furthermore, we segregate our cases from a genetic point of view into groups according to BAP1 and TP53 altered or wild-type status, since these two genes are the most mutated in PM [14] and their mutational status could be proved by immunohistochemistry with reliable results [36, 37]. According to these groups, we found no significant correlation neither in miRNA or TS expression, nor in PM with or without EMT morphology. These events lead us to conclude that neither BAP1 nor TP53 genetic profile influence the expression and function of miRNAs in targeting TS. Furthermore, in our series, BAP1 loss was not associated with EMT morphology of epithelioid PM, as already reported by de Reynies and coworkers that found the mesenchymal phenotype more associate with a BAP1 retained status [52].

Taken together, our results allowed us to hypothesize that TS could take part in the epithelial to mesenchymal transition in PM, providing an additional reason for the poorer outcome of high TS expressing cases, linked not only to an intrinsic chemoresistance [42], but also to a propensity towards a more aggressive histotype. Nonetheless, it must be noticed that PM is per se a unique biological model of complex tumor characterized by the epithelial and mesenchymal type but also by the coexistence of both epithelial and mesenchymal features in the same tumor sample, suggesting an intrinsic epithelial-to-mesenchimal transition [53].

Finally, we found a borderline significance (p = 0.05) for survival analysis in those cases with different TS protein levels, while we failed to found significant survival differences either for miRNA expression, or for EMT morphology. Nevertheless, the curve trends seem to suggest that the state of miRNAs high or TS low or EMT absent confers a better patient outcome.

Furthermore, we found a significant better survival for those cases with both BAP1 loss and TS low protein levels, thus confirming the central role of TS modulation in PM patients’ outcome (Fig. 6).

A limit of these analyses is the consecutive and retrospective nature of our series that implies the heterogeneity of clinical patient stages at diagnosis and the different therapeutic treatments received that could have influenced the survival. Nevertheless, even if we are aware that this data do not demonstrate a direct effect of miRNAs in patient better survival, our study suggests that the modulation/inhibition of TS protein levels by miRNAs could be a good strategy to prevent PM growth.

Conclusions

In conclusion, in the present study, we demonstrated that TS could be a target of both miR-215 and miR-375 in PM tumor model; TS modulation by these miRNAs is not associated with response to PEM nor to EMT-marker ZEB1; however, the strong association of both miRNAs and TS expression with PM histotypes and — in epithelioid group — with mesenchymal-like component, suggests that this pathway might drive EMT processes in epithelioid PM. Furthermore, at least in BAP1 mutated patients, low TS protein levels are associated with better patients’ survival.