Synergic Induction of Autophagic Cell Death in Anaplastic Thyroid Carcinoma

Abstract Anaplastic thyroid carcinoma (ATC) has poor prognosis, high mortality rate and lack of effective therapy. A synergic combination of PD-L1 antibody together with cell death promoting substances like deacetylase inhibitors (DACi) and multi-kinase inhibitors (MKI) could sensitize ATC cells and promote decay by autophagic cell death. The PD-L1-inhibitor atezolizumab synergized with panobinostat (DACi) and sorafenib (MKI) leading to significant reduction of the viability, measured by real time luminescence, of three different patient-derived primary ATC cells, of C643 cells and follicular epithelial thyroid cells too. Solo administration of these compounds caused a significant over-expression of autophagy transcripts; meanwhile autophagy proteins were almost not detectable after the single administration of panobinostat, thus supporting a massive autophagy degradation process. Instead, the administration of atezolizumab caused an accumulation of autophagy proteins and the cleavage of the active caspases 8 and 3. Interestingly, only panobinostat and atezolizumab were able to exacerbate the autophagy process by increasing the synthesis, the maturation and final fusion with the lysosomes of the autophagosome vesicles. Despite ATC cells could be sensitized by atezolizumab via the cleavage of the caspases, no reduction of cell proliferation or promotion of cell death was observed. The apoptosis assay evidenced the ability of panobinostat alone and in combination with atezolizumab to induce the phosphatidil serine exposure (early apoptosis) and further the secondary necrosis. Instead, sorafenib was only able to cause necrosis. The increase of caspases activity induced by atezolizumab, the apoptosis and autophagy processes promoted by panobinostat synergize thus promoting cell death in well-established and primary anaplastic thyroid cancer cells. The combined therapy could represent a future clinical application for the treatment of such lethal and untreatable solid cancer.


Introduction
The anaplastic thyroid carcinoma (ATC) accounts only for approximately 1-2% of thyroid cancers but it is one of the most aggressive diseases with a mean survival time of three to six months after the initial diagnosis (1,2). It is responsible for up to 50% cancer deaths in patients affected by thyroid cancer because of its de-differentiation associated with iodine/radioiodine refractoriness (RAI-R), concomitant loss of sodium/iodide symporter (NIS) and high metastatic potential (3)(4)(5). Multimodal treatment based on surgical resection in combination with radio-chemotherapy is recommended (6), but the prognosis remains poor (7). Therefore, molecularly targeted therapies including multi-kinase inhibitors (MKI) and programmed death (ligand)-1-inhibitors (PD-1/PD-L1) have gained importance in the last years and have shown beneficial for patients with ATC (8,9). Elevated expression levels of the immune checkpoint PD-1 and its ligand PD-L1 have been observed in many different tumor cells and associated tumor-infiltrating lymphocytes (10) and, for this reason, PD-1-/PD-L1-inhibitors were introduced for the treatment of bladder cancer, melanoma, renal cell carcinoma or pulmonary cancer (11)(12)(13). The PD-1 antibody pembrolizumab has shown high efficacy in combination with MKI or BRAF-inhibitors in ATC (14)(15)(16). Additionally, recent work demonstrated that tumor PD-L1 drives several molecular mechanisms involved in cell survival like autophagy (11). The suppression of tumor autophagy in melanoma and ovarian cancer mediated by PD-L1 inhibition suggested that autophagy could be a direct target of cancer-related PD-L1 (11). Autophagy is a lysosomal degradation pathway that prevents the accumulation of damaged proteins and organelles within cells, maintains homeostasis and plays an important role in organogenesis, metabolism and immune response (17)(18)(19)(20), but the relationship between autophagy and cancer remains complex. It acts not only as tumor suppressor resulting in the degradation of metabolic and structural proteins and subcellular organelles (21)(22)(23), but it also promotes cancer growth and survival by maintaining cellular energy production, by eliminating stress and by mediating drug resistance (19,20,24,25). Recent findings highlighted somatic mutations of ATC occurring at the mismatch repair system (26), which could sensitize ATC cells to PD-1 blockade regardless of the expression of PD-L1 (27). Instead, the link between autophagy and PD-L1 in ATC remains unknown yet. This study elucidates the molecular mechanism induced by the deacetylase inhibitor (DACi) panobinostat, sorafenib (MKI), showing high efficacy in primary and established ATC cells (28)(29)(30), and the PD-L1-inhibitor atezolizumab in single and combined administration to modulate autophagic mechanism, associated expression of PD-L1 and cell viability blockade in ATC.
Preparation of patient-derived human tumor tissue (PDTT) Preparation of patient-derived human tumor tissue (PDTT) from three patients who underwent surgery due to ATC was performed as previously described (30). All PDTT were obtained from resected tumor tissue of patients that have never received a previous neoadjuvant treatment (Table 1).

Spheroid formation
C643, patient 1, patient 2 and patient 3 PDTTs, and Nthy-ori-3-1 spheroids were formed on 50 ml of 1.5% peqGOLD Universal Agarose (PEQLAB Biotechnology GmbH, Erlangen, Deutschland) in a flat-bottom 96-well plate (SARSTEDT AG and Co. KG, N€ umbrecht, Germany) for four days without medium change. Ten thousands C643 and Nthy-ori-3-1 cells, 7,000 PDTTs cells were plated in 150 ml of medium in a humidified atmosphere containing 5% CO 2 at 37 C and placed on an orbital shaker with a shaking speed of 100 rpm overnight. Fifty microliters of medium with atezolizumab, sorafenib, or panobinostat and their combination were added to each well containing a single spheroid. The working concentration was 500 ng/ml of atezolizumab, 1 mM of sorafenib, and 10 nM of panobinostat.
Measurement of cell viability C643, PDTT of three patients, Nthy-ori-3-1 monolayer and spheroids viability was detected by the use of the Real Time GloV R kit purchased from Promega (Mannheim, Germany) by following the instructions of the manufacturer. Synergistic effect was determined by calculating the Combination Index (CI) by applying the formula: where D3 is the luminescence optical density (OD) of the combined treatment (panobinostat or sorafenib plus atezolizumab); D1 and D2 represent the luminescence OD of the single compounds. Value lower than 1 was considered as synergic; equal to one was considered additive; higher than 1 was considered antagonist.

Immunohistochemistry of paraffin embedded anaplastic thyroid tissue
Three mm thin sections of 4% formaldehyde fixed paraffin embedded thyroid tissue resected from patient 1, patient 2 and patient 3, were cut, rehydrated and de-paraffinized. The slides were then incubated for 45 minutes with rabbit monoclonal Anti-PD-L1 antibody (1:2000; Cell Signaling #13684, clone E1L3N). The calculation of the TPS, Cologne, IC and CPS Scores was assessed by a pathologist according to the previous findings (31).

Measurement of apoptosis/necrosis
The detection of apoptosis/necrosis was performed by luminescence/fluorescence after the administration of the RealTime-Glo TM Annexin V Apoptosis and Necrosis Assay (JA1011, Promega). Five thousand C643 cells were seeded in a 96-well plate. After 24h, 500 ng/ml of atezolizumab, 1 mM of sorafenib and 10 nM of panobinostat were administered to the cells, alone and in combination. The measurement was acquired by FLUOstar OPTIMA (BMG LABTECH, Ortenberg Germany) plate reader for up to 72h. The data were analyzed by Excel 2016 (Microsoft).

Stable transfection
C643 cells were stably transfected with an E. coli plasmid encoding for RFP-GFP-MAP1LC3B (ptfLC3 was a gift from Tamotsu Yoshimori (Addgene plasmid # 21074; http://n2t.net/addgene:21074; RRID:Addgene_21074) (32) by incubation with 20 mg/ml plasmid in serum-free medium and FUGENEV R HD Transfection Reagent (Promega). The selective agent G-418 (Roche Diagnostics Gmbh, Risch-Rotkreuz, Switzerland) was added to fresh medium threedays later. In the following weeks, transfected cells were separated from negative cells by scratching with a pipette under the fluorescence microscope and plating positive ones on a new dish with fresh medium containing 20 mg/ml of G-418.

Autophagy assay
Five thousand C643 cells, stably transfected with RFP-GFP-MAP1LC3B, were seeded in a round bottom low-attachment plate (corning spheroid microplate 4515, CORNING, Corning, NY, USA) for four days. Spheroids obtained with the transfected cells were treated with atezolizumab (500 ng/ml), 10 nM panobinostat, 100 mM sorafenib and 100 pM bafilomycin. The green and red fluorescence status was monitored by IncuCyteV R S3 Live-Cell Analysis System (Sartorius, G€ ottingen, Germany).

Statistical analysis
Data were collected using Excel (Microsoft Office). Significance was calculated using the ttest for paired samples. p < .05 was considered significant.

Ethical approval
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Ethics Committee of University Hospital of Marburg (No.123/19). Informed consent was obtained from all subjects involved in the study.

Alteration of spheroids ultrastructure and cell viability of monolayer PDTTs
Spheroids and monolayer of three different PDTTs were treated for 96h with 1 mM of sorafenib, 1 and 10 nM of panobinostat based on the previous studies identifying the best working concentrations of sorafenib and panobinostat (29,30). Additionally, the effect on cell viability of 500 ng/ml of atezolizumab was monitored as single administration and in combination with panobinostat or sorafenib. The administration of 500 ng/ml of atezolizumab or 1 nM of panobinostat did not affect neither the spheroid morphology ( Figure 1(A) and Supplemental Figures 1 and 2) nor the cell viability ( Figure  1(B)) of the PDTTs. Instead, 10 nM of panobinostat caused the dismantling of the outer membrane and the 3D structure of the PDTTsderived spheroids and a significant reduction of cell viability (p < .05) already after 48h of treatment of monolayer PDTTs. Interestingly, the combination of 500 ng/ml of atezolizumab and 10 nM of panobinostat caused not only a significant (p < .05) reduction of cell viability in comparison to untreated cells but also significant (p < .05) to 500 ng/ml atezolizumab or 10 nM panobinostat solo treated PDTTs. Similar results were observed after combined administration of 500 ng/ml of atezolizumab and 1 mM of sorafenib (Figure 1(B)). The cytotoxicity could be observed as well in the PDTTs-derived spheroids treated with the combined administration of compounds ( Figure 1(A)).
The effect of the treatment with atezolizumab, panobinostat and sorafenib was further analyzed in spheroids established from C643 and Nthyori-3-1 (human follicular thyroid cells) cells, which have been included in the study to check whether the effects of those compounds is cancer related or generally applicable to cells of thyroid origin. As shown in Figure 2(A) (Supplemental Figures 1 and 3) both spheroids treated with 1 or 10 nM of panobinostat alone and in combination with 500 ng/ml of atezolizumab showed a darkening of the inner area up to a complete loss of its brightness, confirming the loss of the spheroid viability. Instead, the single administration of 500 ng/ml of atezolizumab or 1 mM of sorafenib did not cause visible alteration of the spheroid morphology. Additionally, a significant reduction of cell viability was observed (Figure 2(B)) in C643-and Nthy-ori-3-1-derived spheroids after the administration of 10 nM of panobinostat or 1 mM of sorafenib alone and in combination with 500 ng/ml of atezolizumab. Interestingly, the analysis of the Combination Index (Figure 2(C)) evidenced a synergic effect (value lower than 1) of the combined administration of panobinostat/sorafenib with atezolizumab.

Expression of autophagy markers in ATC cells
The alteration of autophagy genes has been shown to be related with thyroid cancer progression (33) and the clinical response to radioiodine therapy (34). Several proteins encoded by autophagy related genes are involved in the synthesis, elongation and maturation process of autophagosome vesicles (35). Thus, the status of their expression represent a valid tool for clarifying the autophagy status. The expression of autophagy transcripts UVRAG, SQSTM1, TFEB, BECN1, MAP1LC3B, PRKAA1_1, and PRKAA2_ 1 (36) was detected in C643 and three PDTTs (patient 1, patient 2, and patient 3) cells after 48h of treatment with 500 ng/ml of atezolizumab, 10 nM of panobinostat and 1 mM of sorafenib as single administration and in combination. As shown by the whisker box plots of Figure 3(A) that sum up the results obtained from all ATC cells, the treatment with 500 ng/ml of atezolizumab caused a significant over-expression of UVRAG, SQSTM1 and TFEB transcripts. The administration of 10 nM of panobinostat or 1 mM of sorafenib caused a significant (p < .05) overexpression of UVRAG, SQSTM1, TFEB, MAP1LC3B, PRKAA1_1, and PRKAA2_1, thus showing the strongest ability of both compounds to promote the expression of autophagy transcripts. Instead, the combination of 500 ng/ml of atezolizumab and 1 nM of panobinostat or 1 mM of sorafenib caused no significant change in the expression of the autophagy transcripts, except for the significant (p < .05) up-regulation of PRKAA2_1 after the administration of atezolizumab/panobinostat and TFEB after administration of atezolizumab/sorafenib.
To further assess the involvement of the autophagy process in the cell death mediated by atezolizumab, panobinostat, sorafenib and their combination, the protein level of the autophagy markers UVRAG, p62, Beclin, LC3B-I, LC3B-II and the protein level of the total AMPKa and its phosphorylated active form P-(Thr102)-AMPKa were detected after 48h of treatment of primary ATC cells derived from patient 2 and patient 3 (Figure 3(B)). The treatment with 500 ng/ml of atezolizumab caused an increase of all autophagy proteins in both patients' derived cells. Only LC3B-II and P-AMPKa decreased in patient 3 treated cells. The treatment with 10 nM of panobinostat caused a slight over-expression of UVRAG and AMPK-a protein level only in patient 2 derived cells. Interestingly, Beclin, p62, LC3B-I, LC3B-II, and P-AMPKa were significantly downregulated. The addition of 500 ng/ml atezolizumab to the administration of 10 nM of panobinostat rescued the down-regulation of the protein level caused by the single treatment with 10 nM of panobinostat. Thus causing a less pronounced down-regulation of all autophagy markers. The administration of 1 mM of sorafenib caused in both patients-derived cells a stable or increased expression of UVRAG, Beclin, LC3B-I, AMPK-a and P-AMPK-a proteins. The addition of 500 ng/ml of atezolizumab to 1 mM of sorafenib caused an increase of the protein level of all autophagy markers. Thus excluding the modulation of autophagy. The administration of 100 pM of bafilomycin, a well-known autophagy inhibitor, was performed to check the expression of the autophagy markers during autophagy blockade and to compare their protein level to the one of the different treatments. After 48 h of treatment with 100 pM of bafilomycin, the patient-2-derived cells showed an accumulation of all autophagy markers. Instead, the patient-3-derived cells were characterized by a slight decrease in comparison to untreated cells. Notably, the level of all autophagy markers was higher after administration of 100 pM of bafilomycin than after treatment with panobinostat alone and eventually in combination with atezolizumab (Figure 3(B)). Thus, confirming the inhibitory effect of bafilomycin and the stimulation of the autophagy process in the ATC cells treated with the DACi, which caused a strong significant down-regulation of the autophagy proteins because of the ongoing catabolic process. Similar effects have also been observed in patient 1, C643 and Nthyori-3-1 cells (Supplementary Figure 4).

Monitoring the autophagy process in C643 spheroids
The C643 spheroids were treated with solo and combined administration of 500 ng/ml of atezolizumab, 10 nM of panobinostat, 100 mM of sorafenib and 100 pM of bafilomycin. The autophagic process was monitored in C643 cells via stable plasmid transfection with GFP-RFP-MAP1LC3B. Double-labeled LC3B allows autophagic maturation and terminal degradation activity after fusion with the lysosome, where LC3B and the acid-sensitive GFP are degraded by the acidic milieu while the acid-stable RFP retains its fluorescence (37). Fluorescence status and spheroids morphology was continuously tracked with Incucyte for 96h. As shown in Figure 4 and Supplementary Figures 5-10, untreated C643 spheroids evidenced a basal autophagy process because of a stable green/red fluorescence. Instead, the administration of 100 pM of bafilomycin caused a reduction of the red fluorescence, which indicates an inhibition of the final fusion of the autophagosomes with lysosome. Thus impeding the autophagy process. The administration of 500 ng/ml of atezolizumab caused a slight inhibition of the autophagy process, confirmed by the reduction of the red fluorescence. Interestingly, the solo administration of 10 nM of panobinostat was able to increase not only the green fluorescence but also the red fluorescence. Thus confirming a prompt of the autophagy process in the C643 spheroids. Nonetheless, the combination of panobinostat and atezolizumab clearly showed a further increase of both fluorescent signals. Thus supporting the ability of both compounds to synergize and induce autophagy. Such effect was able to cause morphological changes in the spheroids, which showed a disruption of the outer membrane and the shrinkage of their volume, which indicates a block of cell proliferation and further cell decay.
The administration of 1 mM of sorafenib caused a block of the initial phases of the autophagy process, highlighted by the reduction of the green fluorescence. Even the addition of atezolizumab was not able to restore the autophagy process ( Figure 5 and Supplementary Figures 10  and 11).
These results evidenced that 10 nM of panobinostat was able to exacerbate the autophagy process thus prompting cytotoxic effects. This mechanism was more effective in combination with atezolizumab. Instead, sorafenib or bafilomycin caused no morphological change in the spheroids by inhibiting the autophagy process.

Expression and activity of caspases
The status of the caspases is essential to clarify further the cell death mechanisms induced by the single compounds and their combination in terms of activation of cell death executioners' enzymes. Their activity could be impeded by the expression of PD-L1 on the surface membrane of ATC cells. The protein level and the activity of caspases 8 and 3 were analyzed in C643 and Nthy-ori-3-1 cells. As shown in Figure 6(A), the administration (48h) of 500 ng/ml of atezolizumab caused the appearance of the low molecular weight (43-41 kDa, 18 kDa) active forms of caspase 8. Instead, no cleaved caspase 8 was detectable after the addition of 10 nM of panobinostat alone and in combination with atezolizumab. 1 mM of sorafenib was able to induce the cleavage of the caspase 8 active forms. Surprisingly, the combination of sorafenib and atezolizumab caused only the cleavage of the high molecular   (Figure 6(A,B)).
Nthy-ori-3-1 cells expressed basal active forms (43-41 kDa; 18 kDa) of caspase 8 and caspase 3, which were further expressed after treatment with 500 ng/ml of atezolizumab. The addition of 10 nM of panobinostat or 1 mM of sorafenib or 100 pM of bafilomycin caused the disappearance of the caspase 8 cleaved forms. Caspase 3 active form (18 kDa) was not induced by the single treatment with 10 nM of panobinostat ( Figure  6(B)). All the compounds caused, in Nthy-ori-3-1 spheroids, no increase of the activity of the caspases 8 and 3 ( Figure 6(A,B)).

Detection of apoptosis after solo and combined treatment
The ongoing apoptosis could be proven in C643 cells by the luminescence-based measurement of phosphatidylserine (PS) exposure (Figure 6(C)) and by the fluorescence-based (DNA-binding green fluorescent dye) measurement of secondary necrosis (Figure 6(D)). The cells evidenced an increase of PS exposure after 24 h of treatment with 10 nM of panobinostat alone and in combination with 500 ng/ml of atezolizumab. The administration of 1 mM sorafenib and or 500 ng/ml of atezolizumab caused no increase of the PS exposure related luminescence. The treatment with 10% of DMSO has been included as positive control of early apoptosis induction. The secondary necrosis was induced, after 48 h, by the administration of 10 nM of panobinostat and 1 mM of sorafenib alone and in combination with 500 ng/ml of atezolizumab, as evidenced by the significant increase of the fluorescence. Thus, confirming that the C643 cells are dying by apoptosis after solo treatment with panobinostat and in combination with atezolizumab. Instead, sorafenib was able to cause only necrosis. The solo administration of 500 ng/ml of atezolizumab caused no change in the luminescence/fluorescence values in comparison to the untreated cells

Detection of PD-L1 in anaplastic thyroid cancer cells
The expression of PD-L1 confers to cancer cells the ability to resist to the extracellular death signaling by inhibiting the apoptotic transduction signal and the further caspases cleavage. A basal PD-L1 protein level was detected in all primary ATC cells included in the study (Figure 7(A)). The administration of 500 ng/ml of atezolizumab, 10 nM of panobinostat or 1 mM of sorafenib (48h) caused a significant over-expression of the protein level of PD-L1 only in the patient 1 cells. The combination of atezolizumab and panobinostat was able to induce the over-expression of PD-L1 in the patient 1 and patient 3 cells. The combination of atezolizumab and sorafenib increased PD-L1 protein level in patient 1 and decreased PD-L1 protein level in patient 3 cells. 100 pM of bafilomycin caused a significant increase of PD-L1 in patient 1 cells (Figure 7(A)).

Distribution of PD-L1 in anaplastic thyroid tissue
The tumor tissue, which was resected from the three patients included in the study and used for the isolation of primary cells, was stained for PD-L1 by immunohistochemistry and its expression was pathologically analyzed. As shown in Figure  7(B), all patients showed a basal expression of PD-L1. The Tumor Proportion Score (TPS) identified 35% (patient 1), 60% (patient 2) and 40% (patient 3) of positive cells. The Cologne score reached a value of 4 (patient 1 and 3) and 5 (patient 2). The Immune Cell (IC) score confirmed that only 0.5% (patient 2) of the immune cells were positively stained for PD-L1. Patient 1 and Patient 2 showed a negative staining of PD-L1 in the immune cells (IC score 0%). The Combined Positive score (CPS) was estimated at 35% (patient 1), 60% (patient 2) and 50% (patient 3). These results confirmed that the viable cancer cells were the only positive stained for PD-L1. Patient 1 was the only one characterized for a BRAFV600 mutation (Tables 1 and 2).

Discussion
The desolate prognosis of ATC has led to the development of a number of targeted new therapeutic approaches in recent years. Currently, multi-kinase inhibitors and immune checkpoint inhibitors have been approved for the treatment of ATC and have shown promising results (16,38,39). Further advances for the treatment of ATC have been achieved by the administration of deacetylase inhibitors (29,30,40,41).
The present study clarified the cytotoxic effect and the molecular mechanisms induced by a combined administration of the MKI sorafenib or the DACi panobinostat and the checkpoint inhibitor atezolizumab in monolayer and spheroids of C643 cells, human primary tumor cells and human thyroid epithelial cells. The solo administration of high concentration of panobinostat caused in all cells a significant reduction of the cell viability. Its combination with atezolizumab, based on the calculation of the combination index, exerted a synergic effect in spheroids derived from C643 and Nthy-ori-3-1 cells. Despite the single administration of sorafenib or atezolizumab caused no alteration of the viability, their combination was synergic too and lead to a significant reduction of the cell viability. Notably, the spheroids derived from anaplastic and epithelial cells were strongly affected by the treatment with panobinostat and atezolizumab, supporting that their combination is responsible for inducing not only a reduction of cell viability but also cell death. Instead, sorafenib alone and in combination with atezolizumab was not able to attack and dismantle the outer membrane and the ultrastructure of the spheroids.
To date, pembrolizumab, a PD-1 targeting antibody, is the only checkpoint inhibitor investigated for the treatment of ATC. Its combination with the MKI lenvatinib has improved the longterm remission of patients suffering for ATC (16). However, the molecular mechanisms induced by immune checkpoint inhibitors in anaplastic thyroid cancer-especially their influence on autophagy and apoptosis-is still unknown.
The involvement of autophagy as main process responsible for the reduction of cell viability of anaplastic thyroid cells was confirmed in our study by the significant over-expression of the transcript level of autophagic genes UVRAG, SQSTM1, TFEB, BECN1, MAP1LC3B, PRKAA1_ 1, and PRKAA2_1 after solo administration of atezolizumab, panobinostat and sorafenib. Moreover, the analysis of the level of the autophagic proteins clarified the implication of each single compound and their jam interaction. The primary ATC cells were characterized by a significant down-regulation of the protein level of all autophagy markers after administration of 10 nM of panobinostat, which confirmed a strong ongoing autophagy process leading to cell death. The peculiar mechanism of action of panobinostat to modulate autophagic cell death has been already described in several solid cancers (18,22,23,42) thus supporting the results highlighted in anaplastic thyroid cells. The administration of atezolizumab and sorafenib alone or in combination did not cause alteration similar to the administration of panobinostat. Especially the administration of the autophagy blocker bafilomycin showed that the protein level of the autophagy proteins was comparable to atezolizumab or sorafenib, lowered in comparison to untreated cells but, higher in comparison to panobinostat. Thus supporting the strong effect of panobinostat to induce autophagic cell death. Furthermore, the autophagy process has been monitored in stably transfected C643 cells derived spheroids. These results confirmed the ability of panobinostat alone and further in combination with atezolizumab to exacerbate the autophagy process thus leading to significant alteration of the spheroids, dismantling of the outer membrane and finally spheroid shrinkage, which could be attributed to cell death. Instead, the administration of sorafenib, also in combination with atezolizumab, caused a block of the early autophagy process, thus keeping unaltered the spheroid morphology likewise after the administration of the autophagy inhibitor bafilomycin or atezolizumab alone. The analysis of the caspases protein level and activity could better clarify the role exerted by atezolizumab, panobinostat and sorafenib. Interestingly, atezolizumab or sorafenib solo administration was able to induce the cleavage of the active form of the caspases 8 and 3 in C643 cells, which could intrinsically inhibit the processing of the autophagy proteins (43). Instead, panobinostat alone and in combination with atezolizumab was responsible for the suppression of both caspases protein level without affecting their activity, which was stable or even up regulated in C643 spheroids.
Taken together, the results highlighted that panobinostat triggered autophagy but not atezolizumab or sorafenib, which instead caused the cleavage of the caspases. The non-apoptotic cleavage of the caspases, mediated by atezolizumab, could exert an inhibitory effect on the panobinostat-mediated autophagy process by blocking the catabolic process of its regulatory proteins, as it has been already shown (43,44), thus confirmed by the accumulation of the protein level of the autophagy markers after the combined administration of atezolizumab and panobinostat in ATC cells. Unfortunately, the cleavage of the caspases induced by atezolizumab was not sufficient to cause cell death. So, the combination with panobinostat resulted effective by overcoming the inhibitory effect exerted by atezolizumab or sorafenib on autophagy and by prompting, further, the caspases activity and to concert both mechanisms into reduction of cell proliferation and finally decay. Thus, supporting the previous findings highlighting the powerful of a combination of immune checkpoint inhibitors with MKI or DACi in order to target the molecular mechanisms at different level and prompt the tumor cells to die (16,28,29,35,44). Additionally, the detection of apoptosis by luminescence/fluorescence confirmed that the induction of apoptotic cell death could be solely attributed to panobinostat alone and in combination with atezolizumab. Instead, no induction of apoptosis could be observed after the administration either of atezolizumab or soarafenib. The last was only able to induce necrosis of ATC cells.
PD-L1 expression was detectable not only in viable tumor cells of the ATC patients included in the study but also in the primary ATC derived cells. The up-regulation of its protein level could be attributed to the effect exerted by the administration of panobinostat and sorafenib. Thus, supporting the possibility to sensitize such solid cancer to a therapy directed against PD-L1 as already highlighted by previous findings (9,(45)(46)(47). So far, PD-L1 is responsible for the activation of intracellular ERK (48), which leads to proliferation and survival and exert inhibitory effects on cell death stimuli. The inhibition of PD-L1 mediated by atezolizumab could block the ERK, thus favoring the accumulation of caspases and their further cleavage. Thus supporting the apoptotic signaling mediated by panobinostat. However, ATC are characterized by profound genetic alterations that could modulate the sensitivity to the treatment. For this reason, a deep analysis of the genetic background (49,50) and its implication with the activity of molecular mechanisms involved in cell death, together with further in vivo application of these compounds, would represent a further step for the development of a personalized therapy for patients affected by ATC. In particular, recent findings evidenced the loss of CDKN2A and CDKN2B locuses, together with the canonical mutations occurring at BRAF, TP53, ARID1A and RB1 (51). A further study highlighted the alteration of PIK3CA, ALK genes as well as BRAF V600 (52). The genetic alterations of those genes implicated in the cellular mechanism of proliferation and cell death could represent druggable target for the future therapy. Recently, a new therapy has been established for ATC patients carrying a wild type BRAF, based on the administration of lenvatinib (MKI) and pembrolizumab (monoclonal antibody against PD-1) (15). Instead, patients carrying a BRAFV600 alteration have been clinically treated with dabrafenib and trametinib (53). These new patient-oriented therapy could offer new advantages for the patient prognosis. A further personalization of the therapy by targeting recently discovered alterations could drastically improve the current poor prognosis of patient affected by ATC.

Conclusions
The combination of checkpoint inhibitors and DACi, being able to drive epigenetic modifications, or multi-kinase inhibitors could offer new perspectives for the treatment of ATC. The ability of these compounds to determine the reduction of the tumor growth and to suppress the proliferation of tumor cells is characterized by the activation of canonic and alternative mechanisms of cell death that synergize together to finally overcome the aggressiveness of ATC and terminally prompt the tumor cells to die. A further contribution of the tumor environment could exert a strategic role in such mechanisms and it still needs to be further investigated.