Malignant pleural mesothelioma (MPM) is a rare type of cancer that originates from the mesothelial cells of the pleura. Its occurrence is strongly associated with asbestos exposure following a 20 to 30 years-long latency period. The median overall survival of MPM patients is around 8–14 months and currently no targeted therapy is approved as a treatment for this type of cancer. This can be explained by the fact that oncogenic mutations are rarely present in MPM tumors and the most frequently mutated genes are tumor suppressors such as BAP1, NF2 and CDKN2A (Nicolini et al., 2019). Accordingly, there is no established molecular classification of MPM that could serve as a predictive marker for current therapeutic options.

Macroautophagy (here thereafter referred to as autophagy) is a degradation pathway whereby intracellular components are engulfed into double membraned structures (autophagosomes) for their degradation into the lysosome. Autophagy plays an essential role in both health and disease and it is a highly conserved pathway across eukaryotes. It is essential in maintaining cellular homeostasis to provide energy and to eliminate damaged organelles, protein aggregates or pathogens. Autophagy has also been involved in multiple human pathologies, including cancer (Levine and Kroemer, 2019). It is becoming clear that autophagy plays a dual role in cancer, and whereas it has a protective function against tumorigenic insults at the early stages of tumor development, it also has a pro-tumorigenic role in established tumors (Vega-Rubin-de-Celis, 2021).

Autophagy is regulated at multiple levels. For instance, the so-called autophagy-related (ATG) genes are evolutionary conserved and regulate multiple steps in the pathway, some of them controlling autophagosome formation and both selective and non-selective autophagy (Vega-Rubin-de-Celis, 2021). Autophagy is initiated upon multiple signals, including nutrient and stress clues that are mediated by mTORC1 (mechanistic target of Rapamycin complex 1) and AMPK (AMP activated protein kinase), or by autophagy cargoes from multiple sources such as damaged organelles or unfolded proteins that are mediated by ATG11. Active mTORC1 phosphorylates and inactivates ULK1 (UNC-51-like kinase 1), the protein kinase unit of the initiation complex, that also contains ATG13, ATG11 and ATG101. Upon autophagy-inducing conditions, this initiation complex translocates to the ER and initiates autophagosome formation. AKT1 (Protein kinase B) is an upstream regulator of mTORC1 that modulates its activity by phosphorylating and inactivating the TSC1/TSC2 complex as well as by direct phosphorylation of the mTORC1 component PRAS40 (Proline-rich AKT substrate 40), leading to mTORC1 activation and autophagy inhibition (Laplante and Sabatini, 2012). An mTORC1-independent autophagy regulation function of AKT1 was also described through direct phosphorylation of Beclin 1 (Wang et al., 2012), and through regulation of XIAP (X-linked inhibitor of apoptosis), inhibiting basal levels of autophagy (Huang et al., 2013). In low-energy conditions, AMPK phosphorylates ULK1, activating the complex. Once activated, ULK1 phosphorylates components of the PI3KC3 (class III phosphatidylinositol 3-kinase) complex that includes ATG14, Beclin 1, VPS34 and VPS15, generating PI3P on the autophagosome-initiating membranes. mTORC1 and AKT also regulate autophagy and lysosomal biogenesis at the transcriptional level through modulating the localization of the transcription factor EB (TFEB), a master regulator of lysosomal biogenesis (Sardiello et al., 2009) and some autophagy genes (Settembre et al., 2011). TFEB is regulated by mTORC1 (Peña-Llopis et al., 2011), and phosphorylation by mTORC1 at residues S122 (Vega-Rubin-de-Celis et al., 2017), S142 (Settembre et al., 2012; Nezich et al., 2015) and S211 (Martina et al., 2012; Roczniak-Ferguson et al., 2012; Settembre et al., 2012; Vega-Rubin-de-Celis et al., 2017) has been shown. In addition, AKT phosphorylates TFEB at S467 independently of mTORC1 to promote its cytosolic retention (Palmieri et al., 2017). TFEB is also phosphorylated by ERK1/2 at S142 and this modification retains TFEB in the cytosol (Settembre et al., 2011).

A number of studies have shown that autophagy plays an important role in the regulation of cell death and therapy resistance in MPM tumors. Autophagy induction might contribute to the malignant transformation upon asbestos exposure, when DNA damage and inflammation are induced through the release of the high mobility group box 1 (HMGB1) protein by mesothelial cells and macrophages. HMGB1 also initiates autophagy by interacting with RAGE receptor regulating the mTORC1-ULK1 pathway, and through Beclin one phosphorylation, promoting the survival of the transformed mesothelial cells (Xue et al., 2020). It was shown that both mesothelioma cell lines and tumor spheroids can strongly differ in their basal autophagy level and this influences their sensitivity to therapy. Both PI3K/mTOR dual inhibitors and BCL-XL-selective BH3 mimetic treatments induced protective autophagy in MPM cells, although combination with autophagy inhibitors could initiate cell death (Echeverry et al., 2015; Xu et al., 2021). Furthermore, in MPM tumor spheroids treatment with an inhibitor of autophagy initiation through ULK1 inhibition, MRT68921, increased chemosensitivity in spheroids with high autophagy steady-state level (Follo et al., 2018).

Nintedanib (BIBF1120) is a small-molecule tyrosine kinase inhibitor (TKI) that inhibits the activity of several receptor tyrosine kinases (TKs), such as vascular endothelial growth factor receptor (VEGFR 1–3), platelet-derived growth factor receptor (PDGFR-α and–β) and fibroblast growth factor receptor (FGFR 1–3). It also impairs the activity of certain non-receptor TKs including the SRC-kinase family member Lyn (tyrosine-protein kinase lyn), Lck (lymphocyte-specific tyrosine-protein kinase) and SRC (proto-oncogene tyrosine-protein kinase src). It binds to the ATP-binding pocket of these receptors in a competitive and reversible manner (Wind et al., 2019). Nintedanib was approved by the FDA for the treatment of idiopathic pulmonary fibrosis (IPF), a chronic lung disease, as it effectively interferes with lung fibroblast proliferation and migration (Rivera-Ortega et al., 2018). It is also used for the treatment of locally advanced, metastatic or recurrent non-small cell lung carcinoma in combination with docetaxel due to its antiangiogenic effect (Reck et al., 2014). The effect of nintedanib on MPM tumors has also been investigated. In preclinical experiments, nintedanib inhibited the proliferation and migration of MPM cells in vitro and decreased tumor burden and vascularization in vivo (Laszlo et al., 2018). In the first clinical study (LUME-Meso phase II/III trial), combination of nintedanib with standard-of-care chemotherapy provided a clinical benefit (Grosso et al., 2017). Unfortunately, the extended phase III trial did not confirm this initial observation (Scagliotti et al., 2019). These results highlight the need to further examine the effect of nintedanib in MPM cells in order to identify potential biomarkers of sensitive tumors and to explore possible resistance mechanisms. It was demonstrated that in lung cancer cells with hyperactivated FGFRs nintedanib can be sequestered into the lysosome as an intrinsic resistance mechanism. In the lysosome, the weakly basic nintedanib becomes protonated and trapped. Upon co-treatment with the lysosomal vacuolar H + -ATPase (V-ATPase) inhibitor bafilomycin A1 or chloroquine, the lysosomal accumulation of nintedanib decreased as lysosomal acidification was inhibited. Furthermore, both drugs sensitized the cells to nintedanib already at subtoxic doses (Englinger et al., 2017). In lung fibroblasts from patients with IPF nintedanib initiated autophagy in a Beclin 1-dependent and ATG7-independent manner (Rangarajan et al., 2016).

Dasatinib is an inhibitor of primarily the SRC family kinases (SRC, LCK, YES, FYN) and BCR-ABL, but it has an effect also on other receptor TKs, including c-KIT, PDGFR-α and PDGFR-β. It is approved for chronic myeloid leukemia and Philadelphia chromosome positive acute lymphoblastic leukemia (Lindauer and Hochhaus, 2014). c-SRC is highly expressed and activated in both MPM cell lines and MPM tumor samples, and dasatinib treatment induces cell cycle arrest and inhibits the migration and invasion of MPM cells (Tsao et al., 2007). Clinically, as a neoadjuvant agent, dasatinib did not show significant efficacy among patients with resectable MPM, even though, reduced SRC phosphorylation at Tyr419 after dasatinib treatment correlated with longer progression-free survival (PFS) (Tsao et al., 2018). Also, as a second line treatment after chemotherapy in patients with unresectable MPM it had limited efficacy (Dudek et al., 2012). Similarly to nintedanib, it was found that dasatinib can induce autophagy in some cancer cells. In ovarian cancer cell lines, dasatinib initiated autophagic cell death and decreased cell growth in an AKT, mTOR and Beclin 1-dependent manner (Le et al., 2010). In another study where the autophagy-inducing capacity of different TKIs was compared, dasatinib initiated autophagy in two lung cancer cell lines but not in an oral squamous cell carcinoma cell line and in this case, autophagy had a cytoprotective effect (Tanaka et al., 2020).

In this study we investigated the combined effect of dasatinib and nintedanib treatments on a panel of mesothelioma cell lines. Surprisingly, we found that nintedanib treatment did not inhibit SRC activation in MPM cells and paradoxically even increased phosphorylation of SRC in several cell lines. Combination treatment with SRC inhibitor dasatinib could reverse this effect in all cell lines but its impact on cell viability was cell line-dependent. In some cell lines the treatments initiated cell death while in others induced protective autophagy to promote resistance.

Nintedanib (BIBF1120) is a small-molecule tyrosine kinase inhibitor (TKi) with a strong anti-angiogenic effect that could potentiate the effect of conventional chemotherapies. A promising phase II trial in mesothelioma patients showed that there are certain patients that could benefit from the addition of nintedanib. In order to identify these patients we need to find predictive biomarkers and to better understand potential resistance mechanisms. In in vitro kinase assays nintedanib inhibited the SRC kinase at pharmacologically relevant concentrations (IC50: 156 ± 40 nmol/L) (Hilberg et al., 2008). It was also demonstrated that nintedanib attenuates pulmonary fibrosis by reducing the activation in Wnt3a-induced myofibroblast through suppressing SRC kinase activation (Li et al., 2020). Furthermore, nintedanib decreased lung fibrogenesis and TGF-β1-induced epithelial-mesenchymal transition in mice after mechanical ventilation partially by inhibiting the SRC pathway (Grosso et al., 2017). It has been previously shown that SRC protein is strongly expressed in both MPM cell lines and tumor samples. Furthermore, the active, Y419 phosphorylated form of SRC was also present in cell lines and in high percentage of the tumors as well (Tsao et al., 2007). Similarly, we found that in all of the investigated eight mesothelioma cell lines SRC was strongly expressed and activated to a varying extent. However, nintedanib treatment did not decrease SRC activation in any of the MPM cell lines and in five of them we detected a paradoxical activation. Previously, it was shown that TKi can lead to SRC activation. In lung cancer cell lines insulin-like growth factor receptor (IGF-1R) inhibitor linsitinib treatment shortly decreased SRC activity but in a few hours SRC activity returned to an even higher level. However, combined treatment with linsitinib and dasatinib effectively lowered SRC activation for prolonged periods of time (Min et al., 2015). An inverse correlation between linsitinib sensitivity and phosphorylated SRC was also found. We determined the nintedanib sensitivity of our cell line cohort and found that 3 cell lines were highly sensitive (PF531, PF142, SPC212), two moderately sensitive (PF434, PF655) and three not sensitive. This is in good accordance with previous results that also showed that MPM cell lines strongly vary in their sensitivity to nintedanib (Laszlo et al., 2018). However, neither SRC base line phosphorylation, nor its further activation by nintedanib was dependent on the nintedanib sensitivity of the cells. To determine if combined nintedanib and dasatinib treatment can reverse the activation of SRC, first we analyzed the dasatinib sensitivity of the cells. We found that 3 cell lines were sensitive to dasatinib with an IC50 around 100 nM, while two were insensitive with an IC50 over 500 nM. This diversity in the sensitivity of MPM cell lines corroborated the results of previous studies (Tsao et al., 2007; Monica et al., 2016). Upon combination treatment dasatinib could reverse the effect of nintedadib on SRC activation already at the lower 50 nM concentration in all cell lines similarly to the results of the linsitinib and dasatinib combination (Min et al., 2015). We found that in 3 cell lines (SPC111, SPC212 and PF626), the combination treatment had an additive or synergistic effect and in these cells there was no cell death induction. While in 2 cell lines, there was an additive effect only at the higher treatment concentrations and in these cells dasatinib initiated cell death. Similarly, in other studies SRC inhibitors could induce cell death and cell cycle arrest in a subset of mesothelioma cell lines (Tsao et al., 2007; Indovina et al., 2012). It has been shown that TKIs can induce autophagy and this effect is independent from their target molecule. Tanaka et al. compared several TKI inhibitors, including dasatinib, and found that the autophagy inducing capacity of the inhibitors was cell line dependent and had a cytoprotective function (Tanaka et al., 2020). Since autophagy has been shown as a resistance mechanism in MPM cells, and both nintedanib and dasatinib were found to initiate autophagy in other cell types we analyzed if autophagy played a role in our new MPM cell panel, and especially in the drug resistant cell lines. Our autophagic flux analysis suggests that the cells more resistant to nintenadib induce a stronger autophagic response perhaps as a mechanism of defense against external insults. Although the autophagic response across the different cell lines is heterogeneous (perhaps also influenced by the large genetic diversity of the cell lines characteristic of MPM) there is a trend towards a stronger autophagic response upon treatment on the more resistant cells. This suggest that autophagy induction is a resistance mechanism in this context. Indeed, some of the more sensitive cells to both nintenadib and dasatinib (PF434 and PF531) do not induce autophagy upon treatment. However, in the more nintedanib resistant SPC111 cells combination treatment with autophagy inhibitor 3-MA had a synergistic effect. 3-MA inhibits the early steps of the autophagic process and was shown to increase the sensitivity of cancer cells to traditional chemo- and radiotherapy (Xiao et al., 2021). The correlation between tumorigenesis and autophagy is complex and context-dependent, and the first association between cancer and the autophagy machinery derived from observations in mice deficient for Beclin 1, since Becn1 ^+/− mice spontaneously develop multiple malignancies at a high rate (Qu et al., 2003). Later on, increased tumorigenesis was also described in other Atg-deficient mouse models (Poillet-Perez and White, 2019). Such effects are probably due to the implications of autophagy in maintaining genomic stability and suppressing oxidative stress, suggesting a protective role of autophagy in early stages of tumor development. Autophagy also has a number of pro-tumorigenic functions that are believed to fuel already established tumors (Mizushima and Levine, 2020), and therefore inhibition of autophagy can be beneficial for cancer treatment, also because tumor cells rely more on autophagy for survival than normal cells. In many instances chloroquine (or hydroxychloroquine) are being investigated as inhibitor of the latter steps of the autophagy pathway, although the specificity of such inhibitors is limited (Manic et al., 2014), and clinical trials using these drugs showed a rather limited effect (Amaravadi et al., 2019; Mulcahy Levy and Thorburn, 2020).

The mechanism behind this differential response observed in our MPM cell lines remains to be determined, but using autophagy inhibitors may be an option to overcome these effects. Indeed, targeting autophagy vulnerabilities in MPM patients could be an option to improve the efficacy of the treatments.

Autophagy is also regulated at the transcriptional level by several transcription factors, including TFEB (Fullgrabe et al., 2016). TFEB is a master regulator of lysosomal biogenesis, and members of the MiT-TFE family of transcription factors are found in chromosomal translocations (maintaining the open reading frame but controlled by a stronger promoter) in several malignancies, characterized by an increased expression. Under normal, nutrient-rich conditions, TFEB is phosphorylated and bound to 14-3-3 proteins, thus retained at the cytosol. TFEB respond to multiple stress conditions by de-phosphorylation and translocation to the nucleus (Puertollano et al., 2018). A recent publication suggested that TFEB de-phosphorylation and nuclear translocation in a model of Niemann-Pick type C disease (NPC) upon c-Abl inhibition with dasatinib (Contreras et al., 2020). To our knowledge, our report is the first one describing a similar effect on MPM cells. In addition, we also report for the first time the modulation of TFEB localization by nintedanib, being both, the effects of dasatinib and nintedanib independent of mTORC1. A deeper understanding on the role of such nuclear translocation remains to be studied since its consequences on expression of the lysosomal and autophagy target genes is rather mild and very heterogeneous across different cell lines.

We found that autophagy induction by both dasatinib and nintedanib was independent of the AKT1-mTORC1-S6 pathways in MPM cells. Nintedanib treatment did not influence the activation of this pathway in any of the tested cell lines. The effect of dasatinib was dose and cell line dependent. Autophagy was induced after 50 nM dasatinib treatment. However, substantial decrease in the phosphorylation of AKT and S6 was only detected at much higher concentrations (300 nM). This was unexpected as in ovarian cancer cells dasatinib increased autophagy through a mechanism involving the decreased phosphorylation of AKT, mTOR, p70S6K, and S6 (Le et al., 2010). Similarly, dasatinib decreased AKT and S6 activation and increased autophagic cell death in glioma cells (Milano et al., 2009). Of note, dasatinib concentrations were over 100 nM in both of these studies. Although autophagy often facilitates drug resistance and limits therapy response, in certain cases, it can increase antitumor immune response after chemotherapy. Follo et al. have found that in 3D models of MPM tumors with low basal autophagy, upregulation of autophagy increased the release of damage-associated molecular pattern molecules after chemotherapy and initiated immunogenic cell death (Follo et al., 2019). It is important to note that nintedanib was applied in combination with chemotherapy in the referenced study.

Our study corroborates the finding that mesotheliomas are very heterogeneous and likely analysis of unique mutational and transcriptional changes will be required to identify adequate treatments. The three major histological subtypes of MPM have profound prognostic significance. The epitheloid subtype is the most common and confers the best prognosis while the rare sarcomatoid type has the worst outcome. The biphasic subtype contains cells with both characteristics. In our work, we used 2 cell lines with sarcomatoid (PF434, PF531), two with biphasic (SPC111, SPC212) and one with epithelioid origin. Importantly, we found that the two sarcomatoid cell lines were not only highly sensitive to both dasatinib and nintedanib treatments but in these cells the treatments induced cell death. After the successful phase II trial—that included 25% non-epithelioid cases - the failed extended phase III study was limited to MPM patients with epithelioid morphology. Our findings further support the notion that this change in the trial protocol might had an impact on the phase III study and highlights the need for further studies (Nowak et al., 2020).