Mitochondrial ROS Induce Partial Dedifferentiation of Human Mesothelioma via Upregulation of NANOG

The expression of pluripotency factors is a key regulator of tumor differentiation status and cancer stem cells. The purpose of this study was to examine the expression of pluripotency factors and differentiation status of human mesothelioma and the role of mitochondria in their regulation. We tested the expression of OCT4/POU5F1, NANOG, SOX2, PI3K-AKT pathway and BCL2 genes and proteins in 65 samples of human mesothelioma and 19 samples of normal mesothelium. Mitochondrial membrane potential, reactive oxygen species (ROS) generation and expression of pluripotency factors were also tested in human mesothelioma cell line. Human mesothelium and mesothelioma expressed SOX2, NANOG, PI3K and AKT genes and proteins and POU5F1 gene, whereby NANOG, SOX2 and phosphorylated (activated) AKT were upregulated in mesothelioma. NANOG protein expression was elevated in less differentiated samples of human mesothelioma. The expression of genes of PI3K-AKT pathway correlated with pluripotency factor genes. Mesothelioma cells had functional, but depolarized mitochondria with large capacity to generate ROS. Mitochondrial ROS upregulated NANOG and mitoTEMPO abrogated it. In conclusion, human mesothelioma displays enhanced expression of NANOG, SOX2 and phosphorylated AKT proteins, while elevated NANOG expression correlates with poor differentiation of human mesothelioma. Mitochondria of mesothelioma cells have a large capacity to form ROS and thereby upregulate NANOG, leading to dedifferentiation of mesothelioma.


Introduction
Malignant pleural mesothelioma originates from pleural mesothelial cells. It is notorious for its resistance to various types of therapy and poor survival. In spite of recent advances in understanding genetic identity of human mesothelioma, it remains unclear which molecular mechanisms drive its malignant behavior.
Human mesothelioma cell line Mero-14 (The European Collection of Authenticated Cell Cultures) was cultured according to the manufacturer recommendation in the Ham F-10 (Merck) culture medium with 15% fetal calf serum (FCS, Merck) at 37 • C, in a humidified atmosphere containing 5% CO 2 . To stimulate mitochondrial ROS production, cells were treated with the mitochondrial electron transport chain complex III inhibitor antimycin A, and the mitoTEMPO was used to scavenge mitochondrial ROS.

Immunocyto(histo) Chemistry
Immunohistochemistry was used to determine the protein expression in human samples or Mero-14 cells. It was performed as we previously published [19]. Briefly, immunohistochemical detection was done using the EnVision Flex System (Dako, Denmark) at room temperature and the positive reaction was stained by the 3, 3-diaminobenzidine tetrachloride (DAB, Dako, Denmark). After fixation in the ice cold methanol, cells were treated with the peroxidase-blocking reagent for 5 min. This was followed by incubation with the diluted primary antibody at room temperature over a period of 1 h. The primary rabbit monoclonal antibodies were: anti-OCT4 (1/500, Abcam, ab200834), anti-NANOG (1/100, Abcam, ab109250), anti-PI3 Kinase p85 alpha (phospho Y607; 1/200, Abcam, ab182651) and anti-AKT1 (phospho S473; 1/200, Abcam, ab81283), while mouse monoclonal antibodies were: anti-SOX2 (1/200, Abcam, ab171380), anti-vimentin, clone Vim 3B4 (1/200, Agilent), anti-cytokeratin 7, clone OV-TL 12/30 (1/100, Agilent). Samples without primary antibodies served as negative controls. Appropriate tissues were used as positive controls for antibodies. Cells or fixed tissues were then incubated for 30 min in a secondary antibody (EnVision HRP, Agilent) and stained with the DAB for 1 min. The reaction was captured with the EVOS imaging system (Thermo Fisher Scientific) for cells or the camera mounted on the light microscope for tissue samples. The intensity of DAB staining in cells was determined using the ImageJ software (NIH), by subtracting light intensity over the cell (nucleus for SOX2 or cytosol for other proteins) from the light intensity of the background. Therefore, a greater value indicates greater expression of a protein. Immunohistochemical (Im) index, as semi-quantitative value, was used to analyze a protein expression in human tissue samples. It was determined by multiplying percentage of positively stained cells (0 = no positive cells; 1 = less than 10%; 2 = 10-50% and 3 = more than 50%) with the staining intensity (0 = no staining; 1 = weak; 2 = moderate and 3 = strong staining).

Mitochondrial Membrane Potential Measurements
Mitochondrial membrane potential (∆Ψm) was analyzed, as we previously published [21]. Briefly, Mero-14 and human aortic endothelial cells (The European Collection of Authenticated Cell Cultures) were grown in 96-well plates. Endothelial cells were used as the control, non-cancerous cell line. A 30 min prior to recording, cells were transferred to the phenol red-free Hank's Balanced Salt Solution medium and loaded with the 100 nM tetramethylrhodamine ethyl ester (TMRE, Invitrogen, Eugene, OR, USA) for 15 min. After the TMRE washout, the fluorescence intensity was detected at room temperature using the EVOS imaging system and the data were analyzed with the ImageJ software.

Reactive Oxygen Species Measurements
ROS generation was investigated, as we previously published [22]. Briefly, cells were grown in the 96-well plates. After being treated with the antimycin A (2.5 µM, 5 µM and 10 µM) for 3 days, cells were incubated for 30 min in 2 µM CM-H 2 DCFDA (Invitrogen, Eugene, OR, USA). The fluorescence of CM-DCF was acquired by EVOS imaging system at room temperature and analyzed by ImageJ software.

Statistical Analyses
Statistical comparisons were performed using the t-test for two groups or the ANOVA with Tukey post hoc test for multiple group measurements. Differences at p < 0.05 were considered significant. Correlation in the gene expression was analyzed using the Pearson correlation coefficient.

OCT4/POU5F1, NANOG, and SOX2 in Mesothelioma
Samples of human normal mesothelium and mesothelioma were analyzed pathohistologically and immunohistochemically to determine the expression of the pluripotency factors, OCT4, NANOG and SOX2 ( Figure 1). In mesothelium and mesothelioma, OCT4 protein was not expressed, but NANOG and SOX2 proteins were expressed in portion of cells. SOX2 expression was dominantly nuclear, as previously described [17], while NANOG was expressed in both cytosol and nucleus. Comparison of Im index showed a greater expression of NANOG and SOX2 proteins in mesothelioma than in mesothelium, indicting a dedifferentiation of mesothelioma compared to mesothelial cells from which mesothelioma originates. The expression of SOX2 protein showed no difference among different histological subtypes, while NANOG protein was significantly more expressed in the pleomorphic than in the tubulopapillary mesothelioma. Since the pleomorphic mesothelioma is less differentiated than the tubulopapillary, this indicates that NANOG expression increases with the loss of differentiation of mesothelioma. Unlike OCT4 protein, POU5F1 gene was expressed in both human mesothelium and mesothelioma, suggesting a posttranscriptional block in its expression in both tissues. Similar to their proteins, NANOG and SOX2 genes were expressed in both mesothelium and mesothelioma. Unlike protein expression, the genes of pluripotency factors, POU5F1, NANOG and SOX2, were less expressed in mesothelioma than in mesothelium. As expected, mesothelioma cancer markers KRT5 and WT1 were upregulated in mesothelioma compared to mesothelium.

PI3K, AKT, and BCL2 in Mesothelioma
PI3K-AKT pathway is important for the malignancy of mesothelioma [23], and it can induce anti-apoptotic BCL2 protein [14]. The mesothelioma samples used for testing OCT4, NANOG and SOX2 expression were also used for investigating the expression of phosphorylated (activated) PI3K (Tyr607; p-PI3K), phosphorylated/activated AKT (Ser473; p-AKT) and BCL2 ( Figure 2). p-PI3K protein was expressed in mesothelium and mesothelioma to the similar extent according to the Im index. p-AKT was expressed in portion of cells in mesothelium and mesothelioma, but overall, it was more extensively expressed in mesothelioma. However, BCL2 protein was expressed only in several mesothelium and mesothelioma samples, with lower expression in mesothelioma. Analysis of the expression of p-PI3K and p-AKT showed no differences among different histological subtypes of mesothelioma. Both PI3K genes, PIK3CA and PIK3CD, all 3 AKT genes, AKT1, AKT2 and AKT3, as well as BCL2 gene were detected in mesothelium and mesothelioma samples. Unlike PI3K and AKT proteins, all their genes were downregulated in mesothelioma. Only BLC2 exhibited downregulated gene and protein in mesothelioma.   genes KRT5 and WT1. Data are means ± SEM; *p < 0.05 vs Mesothelium or Pleo.

Correlation among Expression of Pluripotency Genes and PI3K, AKT, and BCL2 Genes
Pluripotency factors and PI3K-AKT pathway can regulate each other [15][16][17][18]. The only approach that can test these associations in human mesothelioma samples is the analysis of correlation among the expression of their genes. These results are presented in Table 1 and Figure 3.
Although PI3K-AKT pathway is posttranscriptionally regulated, we also found strong correlations among PIK3CA and all 3 AKT genes (AKT1, AKT2 and AKT3), and PIK3CD and AKT1. This suggests that the PI3K-AKT pathway is also regulated at the level of gene expression. Pluripotency factor genes were in correlation with at least one PI3K or AKT gene, but not with anti-apoptotic BCL2. This suggests that the expression of all tested pluripotency genes is regulated in association with expression of PI3K and AKT genes. Inhibition of PI3K/AKT pathway with wortmannin in Mero-14 cell did not significantly reduce NANOG or SOX2 protein expression, indicating that pluripotency factors act upstream of PI3K/AKT pathway. BCL2 was associated with both PIK3CA and PIK3CD genes.  Although PI3K-AKT pathway is posttranscriptionally regulated, we also found strong correlations among PIK3CA and all 3 AKT genes (AKT1, AKT2 and AKT3), and PIK3CD and AKT1. This suggests that the PI3K-AKT pathway is also regulated at the level of gene expression. Pluripotency factor genes were in correlation with at least one PI3K or AKT gene, but not with anti-apoptotic BCL2. This suggests that the expression of all tested pluripotency genes is regulated in

∆Ψm and ROS Generation in Mitochondria of Mero-14 Cells
Since ROS may modulate the extent of cell differentiation and the expression of pluripotency genes [24,25], and mitochondria-derived ROS depend on the ∆Ψm [26], we performed the following experiments to characterize Mero-14 cell mitochondria and their capacity to generate ∆Ψm and ROS.
Individual Mero-14 cells contained a large number of mitochondria that accumulate TMRE ( Figure 4A), indicating that they are functional and capable of generating ∆Ψm. However, the average ∆Ψm of Mero-14 cell mitochondria was lower than that of non-cancerous human endothelial cell line ( Figure 4B). In accordance with this, the analysis of frequency distribution of ∆Ψm in 250 individual cells of both cell lines revealed that the Mero-14 cell population was shifted toward lower ∆Ψm values than the non-cancerous cell population. However, both populations overlapped considerably, and a good portion of Mero-14 cells contained mitochondria polarized similarly or even more than the median of non-cancerous cell population. Altogether, this indicates that the individual Mero-14 cells contain large number of functional mitochondria, that Mero-14 cells are on average less polarized than the non-cancerous cells, and that a substantial number of Mero-14 cells contain mitochondria with normal ∆Ψm.
The baseline CM-DCF signal, an indicator of ROS bioavailability (the difference between formation and degradation of ROS) was low in Mero-14 cells ( Figure 4C, D), indicating low baseline ROS bioavailability. However, upon addition of antimycin A, a potent stimulator of ROS production at the respiratory chain complex III, Mero-14 cells responded with a dose-dependent raise in ROS generation that increased eight fold at 10 µM antimycin A. This indicates that in spite of low baseline ROS bioavailability, the mitochondria of Mero-14 cells have a large capacity for ROS generation.
toward lower ΔΨm values than the non-cancerous cell population. However, both populations overlapped considerably, and a good portion of Mero-14 cells contained mitochondria polarized similarly or even more than the median of non-cancerous cell population. Altogether, this indicates that the individual Mero-14 cells contain large number of functional mitochondria, that Mero-14 cells are on average less polarized than the non-cancerous cells, and that a substantial number of Mero-14 cells contain mitochondria with normal ΔΨm.

Mitochondria-Derived ROS Induce NANOG Expression
Next we tested whether Mero-14 cells recapitulate the expression pattern of pluripotency factors in human mesothelioma, and examined whether mitochondria-derived ROS affect it, potentially modifying differentiation status of mesothelioma.
Immunocytochemical analysis showed that human mesothelioma line exhibited similar pattern of OCT4, NANOG and SOX2 protein expression, and that subcellular localization of NANOG and SOX2 corresponded to human mesothelioma samples ( Figure 5). Namely, OCT4 expression was undetectable, while NANOG and SOX2 were expressed almost in all cells. NANOG was expressed in the entire cell and SOX2 predominantly in the nucleus. Application of 2.5 and 5 µM antimycin A, which in previous experiments enhanced mitochondrial ROS formation, dose-dependently increased the expression of NANOG. However, 10 µM antimycin A had no effect on the NANOG expression. Interestingly, cells with extremely high expression of NANOG, which can represent cancer stem cells, were enriched upon ROS-inducing antimycin A treatment ( Figure 5A). Mitochondria-targeted ROS scavenger mitoTEMPO abolished the effect of 5 µM antimycin A on NANOG, verifying that mitochondrial ROS induced NANOG expression and thereby partial dedifferentiation of mesothelioma cells. Antimycin A and mitoTEMPO had no effect on the expression of OCT4 and SOX2, nor the expression of control proteins vimentin and cytokeratin 7.

Discussion
We demonstrated that human mesothelium and mesothelioma express NANOG and SOX2 proteins and POU5F1, NANOG and SOX2 genes. Compared to normal mesothelium, mesothelioma exhibited higher expression of NANOG and SOX2 proteins and lower expression of POU5F1, NANOG and SOX2 genes. Poorly differentiated histological subtype of mesothelioma had higher expression of NANOG than the subtype with the greater degree of differentiation. While mesothelioma had downregulated all tested PI3K, AKT and BCL2 genes, p-PI3K was unaltered, p-AKT was upregulated and BLC2 protein was downregulated. The expression of PI3K-AKT pathway genes was in association with POU5F1, NANOG and SOX2, and PI3K and BCL2 gene expressions were also in association. Mero-14, human mesothelioma cell line contained functional, but mostly depolarized mitochondria that exhibited large capacity to generate ROS. Individual Mero-14 cells had large number of functional mitochondria, while smaller proportion of Mero-14 cells contained mitochondria with ∆Ψm equal to non-cancerous human cells. Mesothelioma cell line recapitulated the expression pattern of OCT4, NANOG and SOX2 proteins in human mesothelioma. Mitochondrial ROS upregulated NANOG, an effect abrogated by the mitochondrial ROS scavenger mitoTEMPO. This is the first study that extensively tested the expression of OCT4/POU5F1, NANOG and SOX2 genes and proteins in human mesothelium and mesothelioma. For the first time, we showed that genes and proteins of pluripotency factors NANOG and SOX2 are expressed in normal mesothelium and malignant mesothelioma. Most of the studies showed the expression of pluripotency genes and proteins in mesothelioma cell lines, but not in human mesothelium or mesothelioma. This includes OCT4 protein expression [27], POU5F1, NANOG and SOX2 gene expression in cell lines [28,29] or cancer-initiating cells [30]. Conversely, it has been also shown that POU5F1, NANOG and SOX2 are not expressed in mesothelioma cell lines [31]. We found that POU5F1 was expressed, but OCT4 was not, suggesting its posttranscriptional suppression. Images from Figure 1 show that NANOG and SOX2 staining clearly delineates mesothelioma tissue, which may encourage future studies that would test NANOG and SOX2 as a clinical diagnostic markers in immunohistopathological assays. Although NANOG and SOX2 genes were downregulated in mesothelioma compared to mesothelium, their proteins were upregulated in the cancer. Such opposing results between genes and proteins likely reflect posttranscriptional, translational, and degradation regulation, which ultimately determines protein levels [32]. Since protein expression is directly related to cellular phenotype and therefore more relevant than gene expression, our results indicate that mesothelioma has upregulated NANOG and SOX2 pluripotency factors compared to mesothelium, its tissue of origin, indicating dedifferentiation of mesothelioma. Moreover, poorly differentiated histological subtype, pleomorphic mesothelioma showed greater expression of NANOG than the better-differentiated tubulopapillary subtype. This further demonstrates that increased NANOG expression correlates with poor differentiation of mesothelioma. Others have shown that the high expression pluripotency factors correlates with poor differentiation of different cancer types and poor clinical outcome [33].
The subcellular localization analysis of pluripotency factor protein expression showed that NANOG and SOX2 were expressed in the nucleus and cytosol of human mesothelioma and Mero-14 cells. Topal et al. showed that increased flux of OCT4 and NANOG, but not SOX2, from the nucleus into cytosol is associated with their degradation and differentiation of stem cells [34]. Thus, our observation that NANOG is present in the nucleus of mesothelioma, further corroborates that differentiation processes are inhibited in mesothelioma, rendering mesothelioma a highly undifferentiated cancer.
PI3K-AKT pathway is dysregulated in human mesothelioma [23] and it promotes mesothelioma cell proliferation [35]. Here we showed that PI3K and AKT genes and their phosphorylated forms of proteins are expressed in mesothelium and mesothelioma. Although PI3K and AKT genes were suppressed in mesothelioma, p-AKT protein was upregulated, indicating its increased activity. Thus, our results demonstrate excessive activity of AKT that contributes to the malignant phenotype of mesothelioma. Majority of studies indicate that PI3K-AKT pathway activates OCT4 [15], NANOG [16] and SOX2 [17]. However, OCT4 and NANOG could act upstream of PI3K-AKT pathway [18]. Similarly, a study also showed that upon phosphorylation of OCT4 by AKT, OCT4 dissociated from AKT1 promotor, stimulating its expression [36]. We showed for the first time in human mesothelioma that PI3K and AKT genes were in positive correlation with the expression of POU5F1, NANOG and SOX2, suggesting positive regulation at the level of gene expression. Thus, our results indicate that increased activity of PI3K-AKT pathway is likely associated with observed upregulation of NANOG and SOX2 in human mesothelioma. Since we showed that wortmannin did not significantly affect the expression of NANOG or SOX2, this suggest that PI3K/AKT pathway does not regulate pluripotency genes in mesothelioma and that pluripotency factors act upstream of it.
BLC2 expression was in correlation with PIK3CA and PIK3CD, suggesting association among the expression of these genes. However, BCL2 gene and protein were downregulated in mesothelioma. Although it would be intuitive that, due to its anti-apoptotic activity, low BCL2 expression would suppress cancer, studies show controversial results. Some correlate high BCL2 expression with poor survival [37], while other indicate good prognosis of mesothelioma patients with high BLC2 expression [38]. However, the later study did not determine the BCL2 expression in normal mesothelium, which is important for complete evaluation of those findings. Thus, a downregulation of BCL2 in mesothelioma could promote cancer. However, since BCL2 is poorly expressed in normal mesothelium in our study, it is more likely that it is not important for the mesothelioma.
Immunocytochemical quantification of protein expression in cell culture is highly sensitive and accurate, due to the ability to identify individual cells and normalize the signal to background [39]. The advantage over the time-consuming western blot [40], is that immunocytochemical quantification is not prone to error due to protein crosslinking [41], it allows cell visualization and analysis of cell subpopulations, protein quantification in subcellular compartments and high-throughput experimentation when working with cell culture. This is important when studying SOX2 and NANOG expression, because their subcellular localization can determine the cell fate [34]. However, such quantification of protein expression in tissue slides does not always allow such approach due to the existence of multiple cell layers. Immunohistochemical analysis yields semi-quantitative Im index, which enables simple comparisons [42,43].
Pluripotency factors OCT4 and SOX2 [9] and NANOG [10] are markers of cancer stem cells. Here, we showed that antimycin A treatment enriched cells with very high expression of NANOG ( Figure 5A), which can correspond to cancer stem cells. This indicates that mesothelioma reprograming with mitochondrial ROS likely upregulates cancer stem cells. Moreover, by utilizing advantages of immunocytochemical approach, we demonstrated that SOX2 and especially NANOG are expressed in majority of Mero-14 and mesothelioma cells, not just these putative cancer stem cells. This suggests that NANOG and SOX2 are expressed almost in all cell subpopulations of mesothelioma.
In spite of Warburg effect that describes suppressed mitochondrial function in cancer, an increasing number of studies indicate that cancer cell mitochondria function normally and that they have an important role in cancer cell survival [11]. Suppressed mitochondrial metabolism in cancer cells may originate from downregulation of pyruvate dehydrogenase that converts pyruvate to acetyl-CoA that readily enters the citric acid cycle. As a result, mitochondrial would generate lower ∆Ψm, less ATP and less ROS [44]. Thus, it was important to determine the capacity of mesothelioma cell mitochondria to generate ∆Ψm, as this may dramatically affect their ability to generate ROS. Here we showed that the majority of mesothelioma cells contain functional mitochondria. The frequency distribution of ∆Ψm is shifted to lower membrane potentials in the population of mesothelioma cell line than in non-cancerous human cell line, reflecting mitochondrial depolarization in Mero-14 cells, and suggesting Warburg effect in mesothelioma. However, a considerable number of Mero-14 cells contained mitochondria that exhibited equal ∆Ψm to non-cancerous human cells, indicating normal mitochondrial metabolism. Such dispersion in the distribution of ∆Ψm in human mesothelioma cell line corresponds to variable suppression of mitochondrial metabolisms in subpopulations of cells. Our study is first to characterize the mitochondria of human mesothelioma cells and show that, although depolarized, they are functional and capable of generating ∆Ψm.
We demonstrated that, upon stimulation, mesothelioma cell mitochondria generated ROS that upregulated NANOG protein. Our results also indicate that human mesothelioma cells exhibit low baseline ROS bioavailability. This could arise either from low ROS formation or enhanced ROS scavenging capacity, or both. Since Mero-14 cells contain depolarized mitochondria, and depolarized mitochondria generate less ROS [22,26,45], it is plausible that low baseline ROS bioavailability stems from its reduced formation by mitochondria. Proton pumping by respiratory chain complexes is less opposed by lower ∆Ψm, which improves coupled electron flux across respiratory chain. As a consequence, electron leak out the respiratory chain and formation of superoxide would be reduced. On the other hand, numerous studies showed that cancer cells can exhibit high levels of ROS scavengers and antioxidants, such as catalase, which can also contribute to observed low baseline ROS formation in mesothelioma cell mitochondria [46]. Importantly, we demonstrated for the first time that mitochondria of human mesothelioma cells have large capacity (at least an eight fold increase upon stimulation) to form ROS. One of our key findings is that mitochondrial ROS induces NANOG protein expression in mesothelioma, which emphasizes the key role of mitochondria in the maintenance of undifferentiated state of human mesothelioma cells. Since upregulation of several pluripotency factors induces cellular reprogramming and dedifferentiation in Yamanaka's experiment [1], such induction of NANOG in our experiment indicates that partial reprogramming and dedifferentiation of mesothelioma cells can be driven by mitochondrial ROS. Our observation that NANOG is more expressed in human mesothelioma subtype with poor differentiation status further supports the role of NANOG expression in regulation of differentiation status of human mesothelioma. Others have shown that H 2 O 2 can upregulate POU5F1, NANOG and SOX2 in mesothelioma line [28]. However, their experiment relied on non-physiological, exogenous application of H 2 O 2 , while we stimulated endogenous mitochondrial ROS formation, which corresponds better to conditions occurring in the tumor. The upregulation of NANOG by mitochondrial ROS could result from the activation of redox-sensitive transcription factor Nrf2, which binds upstream of NANOG gene promoting its expression, and also controls NANOG protein degradation [47]. It has been shown that Nrf2 stimulates proliferation of cancer cells via synthesis of GSH and AKT signaling [48]. Another pathway for NANOG induction by ROS involves HIF-1α. It has been shown in various types of cancer cell lines that ROS can activate HIF-1α, which induced the expression of pluripotency factors [49]. Somewhat surprisingly, we observed that higher (10 µM) antimycin A concentration did not upregulate NANOG. This possibly reflects inhibitory effect on pluripotency factors when ROS act at very high concentrations, which describes a non-linear dose-response curve observed in many physiological systems [50]. Overall, our results indicate an important role of mitochondria and mitochondrial ROS in regulation of differentiation status of human mesothelioma and thereby aggressive behavior of this cancer. In accordance with our results, studies in other cancer and normal cells, showed a crucial role of mitochondrial metabolisms in driving the differentiation process [51,52]. Moreover, cancer cells without mitochondria seem to be unable to form tumors and spread locally and distally [53]. Thus, targeting mitochondrial may represent novel and efficient approach in treating mesothelioma [54].

Conclusions
Our study demonstrates that NANOG and SOX2 genes and proteins are expressed in human mesothelium and mesothelioma and they may have diagnostic potential. The poor differentiation of human mesothelioma is associated with enhanced expression of NANOG, which can be induced by mitochondrial ROS. Although mostly depolarized, mitochondria of human mesothelioma cells have a large capacity to generate ROS, and thereby trigger partial dedifferentiation and drive reprogramming of mesothelioma. The positive association in the expression of genes of pluripotency factors (POU5F1, NANOG and SOX2) and of PI3K-AKT pathway (AKT1, AKT2, AKT3, PIK3CA and PIK3CD) indicates positive regulation in their expression, while pluripotency factors act upstream of the PI3K-AKT pathway.