Autophagy facilitates the release of immunogenic signals following chemotherapy in 3D models of mesothelioma
Carlo Follo1 | Yao Cheng2 | William G. Richards3 | Raphael Bueno3 |
V. Courtney Broaddus1

We have previously shown that nearly half of mesothelioma patients have tumors with low autophagy and that these patients have a significantly worse outcome than those with high autophagy. We hypothesized that autophagy may be beneficial by facilitating immunogenic cell death (ICD) of tumor cells following chemotherapy. An important
hallmark of ICD is that death of tumor cells is preceded or accompanied by the release of damage‐associated molecular pattern molecules (DAMPs), which then can stimulate an antitumor immune response. Therefore, we measured how autophagy affected the
release of three major DAMPs: high mobility group box 1 (HMGB1), ATP, and calreticulin
following chemotherapy. We found that autophagy in three‐dimensional (3D) models with low autophagy at baseline could be upregulated with the cell‐permeant Tat‐BECN1 peptide and confirmed that autophagy in 3D models with high autophagy at baseline
could be inhibited with MRT 68921 or ATG7 RNAi, as we have previously shown. In in vitro 3D spheroids, we found that, when autophagy was high or upregulated, DAMPs were released following chemotherapy; however, when autophagy was low or inhibited, DAMPs release was significantly impaired. Similarly, in ex vivo tumors, when autophagy was high or upregulated, HMGB1 was released following chemotherapy but, when autophagy was low, HMGB1 release was not seen. We conclude that autophagy can be upregulated in at least some tumors with low autophagy and that upregulation of autophagy can restore the release of DAMPs following chemotherapy. Autophagy may be necessary for ICD in this tumor.

KEYW ORD S
3‐dimensional, ATG13, DAMP, ex vivo, Tat‐BECN1
1Department of Medicine, Zuckerberg
San Francisco General Hospital and Trauma Center, University of California San Francisco, San Francisco, California
2Department of Thoracic Surgery, The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, Shaanxi, China
3Division of Thoracic Surgery, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Correspondence
V. Courtney Broaddus, Zuckerberg San Francisco General Hospital and Trauma Center, 1001 Potrero Avenue, Room 5K1, San Francisco, CA, 94110.
Email: [email protected]

Funding information
International Mesothelioma Program; Simmons Mesothelioma Foundation

⦁ | INTRODUCTION

Although autophagy, a lysosomal degradation pathway, has been linked to many vital cell processes, its role in cancer remains unclear. In several studies, autophagy appears to enhance cancer develop- ment and growth, whereas in others, autophagy appears to suppress

Abbreviations: 3D, 3‐dimensional; BECN1, BECN1/beclin 1; LC3B, MAP1LC3/LC3 microtubule‐associated protein 1 light chain 3 isoform B; MCS, multicellular spheroids; MPM, malignant pleural mesothelioma; NH4+, ammonium chloride; TFS, tumor fragment spheroids.
cancer.1,2 Such divergent findings continue to puzzle those who would want either to inhibit or augment autophagy in cancer patients. We suggest that the field has been constrained by three main challenges: the lack of clinically relevant models, the absence of biomarkers to measure the dynamic process of autophagy in fixed tumor tissues, and the lack of reliable ways to increase and decrease autophagy.3-7 In studying pleural mesothelioma, a particularly chemoresistant tumor without good therapeutic options, we have tried to address these limitations to investigate the role of autophagy in that tumor.

Molecular Carcinogenesis. 2019;1-16. wileyonlinelibrary.com/journal/mc © 2019 Wiley Periodicals, Inc. | 1

In our prior studies, we have established that 3D models, of cell lines and of tumor tissue, were relevant models of the actual tumor8; for example, our ex vivo tumor spheroids reflected the level of autophagy in the tumor from which they were derived.9 We also found that ATG13 puncta, indicative of the initiation of autophagy, had value as a static marker of the dynamic process of autophagy in all three mesothelioma settings: in vitro spheroids, ex vivo spheroids and fixed tumor tissue.9 Using this marker, we learned that mesothelioma tumors could be divided into those with low or even undetectable autophagy and those with high autophagy. Interest- ingly, those with low or absent autophagy showed a significantly worse outcome than those with high autophagy. We were left with two major questions: why is low autophagy associated with a worse outcome and can anything be done about those with low autophagy to improve their outcome?
One possibility that has been recognized recently is that autophagy may be involved in the antitumor immune response by facilitating immunogenic cell death (ICD), characterized by the
release of immunogenic signals, or damage‐associated molecular
pattern molecules (DAMPs), following chemotherapy.10-16 Thus, we asked whether autophagy might be beneficial for mesotheliomas by facilitating ICD. To study this, we sought a specific inducer of autophagy, one that would be effective in 3D models and hopefully in
patients. We discovered that a cell‐permeable peptide, Tat‐BECN1,17
could upregulate autophagy in our 3D models, both in vitro and ex vivo. With this valuable tool, along with the means of downregulating autophagy using MRT 68921 or ATG7 RNAi, as in our previous publications,3,18 we tested whether baseline autophagy was needed for DAMPs release and, if so, whether low baseline autophagy could be upregulated to restore this release. We assessed the release in association with cell death of three DAMPs that are the major hallmark of ICD: high mobility group box 1 (HMGB1), ATP, and calreticulin.11,19 We were able to show that autophagy appears to be required for the release of DAMPs by this tumor following chemotherapy. Importantly, in the low autophagy tumor models, autophagy could be upregulated leading to an increase in the release of DAMPs and in cell death following chemotherapy. These findings raise the possibility that tumors with high autophagy may respond better to therapy and intriguingly, that tumors with low or absent autophagy can be rendered more immune responsive by increasing their autophagy.

⦁ | MATERIALS AND METHODS

⦁ | Reagents and antibodies
Unless otherwise specified, analytical grade chemicals were from Sigma‐Aldrich Corp (St. Louis, MO).

⦁ | Antibodies
The following primary antibodies were used in immunofluorescence or immunoblotting studies: rabbit monoclonal anti‐ATG13 (Cell
Signaling Technology, Danvers, MA, 13468), rabbit monoclonal anti‐calreticulin (Cell Signaling Technologies, 12238), rabbit poly- clonal anti‐cleaved caspase 3 (Asp175) (Cell Signaling Technology,
9661), rabbit polyclonal anti‐HMGB1 (Cell Signaling Technology, 3935), mouse monoclonal anti‐cytokeratin clones AE1/AE3 (Dako North America, Carpinteria, CA, M3515), rabbit monoclonal anti‐ LC3B (Cell Signaling Technology, 3868), mouse monoclonal anti‐α‐
tubulin (T6074). The following secondary antibodies were used in immunoblotting: horseradish peroxidase‐conjugated goat anti‐mouse
IgG (Bio‐Rad, Hercules, CA, 170‐6516), horseradish peroxidase‐
conjugated goat anti‐rabbit IgG (Bio‐Rad, 170‐6515). The following
secondary antibodies or fluorescent dyes were employed in immunofluorescence: Alexa Fluor 546 goat anti‐rabbit IgG (Life
Technologies, Carlsbad, CA, A11010), biotinylated sheep anti‐mouse
IgG (GE Healthcare, Waukesha, WI, RPN1001V), NeutrAvidin Oregon Green 488 conjugate (Life Technologies, A6374), SYTOX
Green Nucleic Acid Stain (Life Technologies, S7020), and TO‐PRO‐3
iodide (Life Technologies, T3605).

⦁ | Autophagy modulators
The cell‐permeant peptide Tat‐BECN1 (Selleck Chemicals, Houston, TX, S8595) was employed as a potent and specific autophagy inducer.17,20,21 Tat‐BECN1 was diluted in water and employed at the working concentration of 10 µM. The working concentration of Tat‐BECN1 was established as the minimum concentration (out of three different concentrations: 1 µM, 10 µM, and 20 µM) that
activates autophagy in our system (not shown). An equivalent
amount of water (0.2% final concentration) was used as vehicle control for Tat‐BECN1 (‐).
MRT 68921 (Selleck Chemicals, S7949) was employed to inhibit
autophagy at 1 µM as determined in a previous study.3 An equivalent dilution of DMSO (0.02% final concentration) was used as vehicle control (−) for MRT 68921.
The following autophagy inducers and inhibitors were tested and failed to modulate autophagy in our system (not shown and)3: lithium
chloride (203637), spautin‐1 (SML0440), SAR 405 (ApexBio Technol-
ogy, Houston, TX, A8883), and 3‐methyladenine (M9281).

⦁ | Chemotherapeutic agents
Carboplatin and pemetrexed were obtained from the University of California San Francisco pharmacy. The agents were diluted in medium and used at concentrations of carboplatin (200 µM) and pemetrexed (10 µM).3,9 Treatments were performed for either 16 hours (a time where apoptosis is not detectable in our system) or 24 hours. We chose these times of treatment because release of ATP and exposure of calreticulin have been reported to occur premortem at a preapoptotic or early apoptotic stage.22,23 Of note, the combined treatment with carboplatin plus pemetrexed did not
alter autophagy in our system (not shown), thus in this study autophagy modulation resulted only from Tat‐BECN1 or MRT 68921.

Carboplatin has been reported to stimulate ATP release and, to a lower extent, HMGB1 release and calreticulin exposure.24

⦁ | Spheroid generation and treatment
⦁ | Multicellular spheroids
Multicellular spheroids (MCS) were generated from human mesothe- lioma cell lines REN and M28, originally obtained from colleagues or ATCC.18 Cell lines were confirmed as mesothelial in origin by staining positively for mesothelioma markers (CALB2/calretinin, WT1) and negatively for other markers not seen in mesothelioma (TTF1). All cells were confirmed to be negative for mycoplasma every month by
PCR analysis as previously described.25 MCS were generated in nonadsorbent round‐bottomed 96‐well plates, as previously de-
scribed.26 Briefly, 96‐well plates were coated with a 5 mg/mL
solution of polyHEMA (P3932) in 95% ethanol and dried at 37°C for 48 hours. Plates were sterilized with ultraviolet light for 30 minutes before use. A total of 105 cells were added to each well and plates were centrifuged at 800 g for 5 minutes to bring the cells into contact at the bottom of each well. The plates were then placed in a 37°C humidified incubator with 5% CO2 for 24 hours to allow spheroids to form. For the next 24 hours, MCS (n = 10) were
transferred to each well of a polyHEMA‐coated 12‐well plate or left
in the original 96‐well plates for either the ATP assay or the caspase‐
Glo 3/7 assay. For the study of released HMGB1 and extracellular ATP (eATP), MCS were grown in medium without fetal bovine serum (FBS) for the last 24 hours. For the study of autophagic flux, MCS were exposed to ammonium chloride (10 mM, NH4+) for the last 4 hours of the experiment, a time validated for measuring autophagy in this setting.3,5,9 For autophagy modulation, MCS were exposed to
Tat‐BECN1, MRT 68921, or vehicle control for 24 hours. For studies
of cell death following chemotherapy, MCS were treated with C+P for the last 16 hours or 24 hours of the experiment. For RNA interference experiments, transient knockdowns were performed using Lipofectamine RNAiMAX Transfection Reagent according to
the manufacturer’s protocol (13778150; Thermo Fisher Scientific, Waltham, MA). In brief, M28 cells were transfected in Opti‐MEM (11058021; Thermo Fisher Scientific) and allowed to grow as
monolayers for 24 hours in complete medium. Cells were then
trypsinized and employed to generate spheroids in polyHEMA‐ coated 96‐well plates. Autophagic flux and HMGB1 release were then determined as described below. Scrambled (12935–400) and
ATG7 (s20651) siRNA were purchased from Thermo Fisher Scientific.

⦁ | Tumor fragment spheroids
TFS were generated from fresh tumor specimens resected from chemonaïve malignant pleural mesothelioma (MPM) patients by extrapleural pneumonectomy or pleurectomy procedures performed at Brigham and Women’s Hospital in Boston, MA, as previously described.8 For ex vivo spheroid cultures, tumor tissue was diced finely with scalpels to pieces smaller than 1 mm in diameter. These
TFS were cultured in 100 mm Petri dishes coated with 0.77% Noble agar (A5431) in Dulbecco’s modified Eagle medium supplemented with 5% fetal bovine serum and 1% penicillin‐streptomycin solution
for 2 or 3 weeks until rounded. Then, TFS were studied for 24 hours. At the start of the 24 hours period, TFS (n = 10) were transferred to
each well of a polyHEMA‐coated 12‐well plate and given fresh media.
For the study of released HMGB1, TFS were grown in medium
without FBS for the last 24 hours. For autophagy upregulation, where indicated, TFS were exposed to Tat‐BECN1, or the vehicle control for 24 hours. For studies of cell death following chemotherapy, TFS were
treated with C+P for the last 16 hours or 24 hours of the experiment.

⦁ | Autophagy studies
⦁ | Autophagy initiation in multicellular spheroids
MCS employed in this study were previously shown to have either
inactive autophagy initiation (REN), or active autophagy initiation (M28) at baseline.3,9 Following Tat‐BECN1, autophagy initiation was determined in MCS by ATG13 puncta analysis as previously
described.3,9 Briefly, MCS were disaggregated with Accumax
(Innovative Cell Technologies Inc., San Diego, CA, AM‐105) and 2× 104 cells were cytospun onto glass slides. Cells were then washed in PBS, fixed with 4% paraformaldehyde at 4°C, blocked and
incubated overnight with rabbit anti‐ATG13 antibody at 4°C (1:50). Cells were then incubated 2 hours with Alexa Fluor 546 goat antirabbit IgG (1:200) and TO‐PRO‐3 (1:1000; 910576; Life Technologies). Washes (3×) were performed with TBS (Amresco, Solon, OH, 0788)‐0.1% Tween 20 (Fisher Scientific, Hampton, NH, BP337) following each antibody incubation. Primary or secondary
antibodies and fluorescent dyes were diluted in antibody diluent solution (21544; EMD Millipore, Burlington, MA). In each experi- ment, negative control with the secondary antibody alone was included. Slides were mounted with ProLong Gold antifade (P36930; Life Technologies). Images were captured at 63× magnification with a Nikon C1 confocal microscope (Nikon Instruments Inc., Melville, NY). ATG13 puncta were either detectable in almost all the cells or absent and thus were not counted.

⦁ | Autophagic flux in multicellular spheroids
MCS in this study were previously shown to have either low
autophagic flux (REN), or high autophagy flux (M28) at baseline.3,9 Following Tat‐BECN1, or MRT 68921 autophagic flux was deter- mined in MCS by LC3B immunoblotting as previously described.3,9
4
Briefly, MCS were exposed to NH + for the last 4 hours before harvesting the spheroids to determine the accumulation of the
autophagy protein LC3 accordingly to the autophagic flux.5 MCS were washed twice with cold phosphate‐buffered saline (DPBS; GE Healthcare, Waukesha, WI, SH30028), harvested in radioimmuno- precipitation assay buffer (1% Nonidet P‐40 [74385]; 0.5% sodium deoxycholate [30970]; 1% SDS [L3771]) supplemented with protease
and phosphatase inhibitor cocktail (78442; Thermo Fisher Scientific)

and homogenized using an ultrasonic cell disruptor (Fisher Scientific, Hampton, NH). Protein concentration was assessed with DC Protein Assay (500‐0111; Bio‐Rad) and equal amounts of protein (30 µg of
total cell homogenates) were separated by sodium dodecyl sulfate‐ polyacrylamide gel electrophoresis and transferred onto a polyviny-
lidene fluoride membrane (162‐0177; Bio‐Rad). Following blocking with 5% nonfat milk (sc‐2324; Santa Cruz Biotechnology, Dallas, TX,), the filter was probed with designated primary and secondary
antibodies and developed with enhanced chemiluminescence sub- strate (34080; Thermo Fisher Scientific). Bands were imaged and intensities were measured by densitometry using the BioSpectrum
imaging system apparatus (UVP LLC, Upland, CA) equipped with the Vision‐WorksLS software (UVP LLC). The autophagic flux was
expressed as a ratio of normalized LC3B‐II band intensities following
4 4
NH + to the band intensities before NH +. Autophagic flux was determined in MCS grown in the complete medium following Tat‐
BECN1 (Figure 1B and 1C) to determine whether Tat‐BECN1
upregulate autophagy. Autophagic flux was also determined in MCS grown in medium without FBS following Tat‐BECN1 or MRT 68921 to assess whether serum deprivation altered the ability of autophagy
modulators to either upregulate or block autophagy (Figure 1).

⦁ | Autophagy initiation in tumor fragment spheroids
Following Tat‐BECN1, autophagy initiation was determined in TFS by ATG13 puncta analysis as previously described.3,9 Briefly, TFS
were collected, fixed in 10% formalin and embedded in 3% agarose. The agar pellets were embedded in paraffin. TFS sections (5 µm) were deparaffinized with xylene and rehydrated using an ethanol gradient. For antigen retrieval, slides were incubated in citrate buffer (C999) and heated in a pressure cooker for 10 minutes. Following antigen retrieval, sections were blocked with 1% bovine serum albumin (BSA)
(SH30574; GE Healthcare) in TBS‐0.1% Tween 20 for 30 minutes.
Slides were incubated overnight at 4°C with rabbit antibody anti‐
ATG13 (1:50); for staining mesothelioma cells, sections were then incubated with mouse anti‐cytokeratin (1:200) for 1 hour and, after washes, with biotinylated sheep antimouse IgG (1:200) for 1 hour.
Sections were then incubated 2 hours with Alexa Fluor 546 goat antirabbit IgG (1:200), NeutrAvidin Oregon Green 488 conjugate
(1:200), and TO‐PRO‐3 (1:1000; 910576; Life Technologies). Washes
(3×) were performed with TBS (0788; Amresco)‐0.1% Tween 20
(BP337; Fisher Scientific) after each antibody incubation. Primary or secondary antibodies and fluorescent dyes were diluted in antibody diluent solution (21544; EMD Millipore). In each experiment, a negative control with the secondary antibody alone was included. Slides were mounted with ProLong Gold antifade (P36930; Life Technologies). Images were captured at 63× magnification with a Nikon C1 confocal microscope (Nikon Instruments Inc). As described
before, because the ATG13 puncta cannot be individually counted, we count the percentages of cytokeratin‐positive cells containing
ATG13 puncta (ATG13‐positive MPM cells).3,9 At least 100 cytoker-
atin‐positive cells, obtained from 3 different TFS, were analyzed for
the presence of ATG13 puncta. A punctum was defined as an ATG13‐positive mainly circular cytoplasmic structure of approxi- mately 1 µm in diameter; this is thought to correspond to an early
autophagy structure (omegasome or phagophore). As previously determined, autophagy initiation correlates with autophagy levels in our 3D models.3,9 A cutpoint of ATG13 positivity of 6% was used to
identify TFS with either low autophagy (<6% of ATG13‐positive
MPM cells) or high autophagy (≥6% of ATG13‐positive MPM cells) based on our previous results showing a correlation between ATG13
positivity ≥6% and better patient outcome in mesothelioma.9

⦁ | Apoptosis studies
⦁ | Multicellular spheroids
In MCS, cell death following chemotherapy was measured by
cleavage of caspases 3 and 7, and nuclear condensation analysis. Cleavage of caspase 3 and 7 was assessed using the caspase‐Glo 3/
7 assay kit (G8093; Promega, Madison, WI) as previously
described.3 Briefly, MCS were treated as indicated in the polyHEMA‐coated 96 wells in which they were grown (200 μL final volume). One spheroid was used for each condition; studies
were performed in quadruplicate. After treatment, plates were spun at 400 g for 10 minutes RT; 100 μL of supernatant was carefully removed without disturbing the pellet and 100 μL of
complete caspase‐Glo 3/7 reagent was added to each well. Plates
were gently mixed using a plate shaker at 300 to 500 rpm for
1 minute and then incubated at RT protected from light for 1 hour. Luminescence was measured in a plate‐reading luminometer (Perkin Elmer, Waltham, MA). Nuclear condensation, indicative
of apoptosis, was assessed after Hoechst staining as previously
described.3 Briefly, MCS were disaggregated with Accumax (Innovative Cell Technologies Inc., AM‐105), washed with ice‐cold PBS, and then fixed with 2.5% glutaraldehyde. Cells were then
stained with 8 μg/mL of Hoechst 33342 (Life Technologies, H3570) and placed on slides. Apoptosis was quantitated by counting cells with distinctive signs of nuclear condensation and expressed as a
% of the total cells (% apoptosis). For each condition, at least 300 cells were counted in three different fields by two investigators blinded to the experimental conditions.

⦁ | Tumor fragment spheroids
In TFS, the presence of apoptotic cells was analyzed by detecting cleaved caspase 3 specifically in MPM cells by immunofluores- cence. TFS were collected and processed for immunofluorescence as described above (see autophagy initiation in tumor fragment
spheroids) but using a rabbit antibody anti‐cleaved caspase
3 (1:50) as the primary antibody. Mesothelioma cells were identified by positive cytokeratin staining and the percentage of
cleaved caspase 3‐positive MPM cells was determined for each
condition. At least 100 cytokeratin‐positive cells, obtained from
three different TFS, were analyzed for the presence of cleaved caspase 3 staining.

FIG U RE 1 In multicellular spheroids, Tat‐BECN1 increases autophagy in spheroids with low autophagy and does not alter autophagy in spheroids with high autophagy. A, MCS with either low autophagy (REN) or high autophagy (M28) at baseline were exposed to 10 µM Tat‐BECN1 (+), or the vehicle control (–) for 24 hours. Spheroids were disaggregated with trypsin and the cells cytospun on glass slides. Cells
adherent on glass slides were then fixed, stained for ATG13 (green), TO‐PRO‐3 to identify nuclei (blue), and imaged by confocal microscopy. In
the absence of Tat‐BECN1, ATG13 puncta were almost undetectable in REN but detectable in almost all the cells in M28. Following Tat‐BECN1,
ATG13 puncta were detectable in almost all the cells in both cell lines. Representative ATG13 puncta are indicated by arrows. Scale
bars = 10 µm. B, Cells were grown as in A. Where indicated, the cells were exposed to 10 mM ammonium chloride (NH4+) for 4 hours. LC3B expression was assessed by immunoblotting. As a loading control, filters were probed with the anti‐α‐tubulin antibody. Band intensities were determined by densitometric analysis. The autophagic flux is expressed as a ratio of normalized LC3B‐II band intensities following NH4+ to
before NH4+ (NH4+:CTRL). A representative immunoblot of three independent experiments is shown, with ratios shown below. C, Bars show the
autophagic flux of the indicated cell lines. Data were obtained from three independent experiments, one of which is shown in B. Asterisks indicate statistically significant differences of autophagic flux between MCS grown in either the presence or in the absence of Tat‐BECN1 (P < .01). Error bars, S.D. In REN MCS, the mean LC3‐II ratios in the presence of Tat‐BECN1 are significantly higher than in the absence of
Tat‐BCEN1 (P < .01). MCS, multicellular spheroids; SD, standard deviation [Color figure can be viewed at wileyonlinelibrary.com]

⦁ | Release of DAMPs
⦁ | HMGB1
HMGB1 released by MCS and TFS in the media was assessed at the end of the 24 hours by immunoblotting of media precipitates. MCS and TFS were grown in medium without FBS to allow analysis of protein precipitates by immunoblotting. At the end of experiments, media were collected, and protein precipitated by trichloroacetic acid/acetone precipitation. Protein precipitates were dissolved in 1× Laemmli sample buffer. The dissolved precipitates were then loaded on polyacrylamide gels for HMGB1 immunoblotting.

⦁ | Extracellular ATP
ATP released by MCS in the media was determined with the Enliten ATP assay system bioluminescence kit (FF2200; Promega) according to the manufacturer’s guidelines. Briefly, MCS were
treated as indicated in the polyHEMA‐coated 96 wells in which
they were grown (200 μL final volume). One spheroid was used for each condition; studies were performed in quadruplicate. After treatment, plates were spun at 400 g for 10 minutes RT; 100 μL of supernatant was carefully removed without disturbing the pellet and employed for the ATP assay. Luminescence was
measured in a plate‐reading luminometer (Perkin Elmer, Waltham,
MA). MCS were grown in medium without FBS to avoid ATP degradation. eATP concentrations were calculated according to a standard curve.

⦁ | Cell surface‐exposed calreticulin (EctoCRT)
EctoCRT on MCS cells was determined at the end of the 24 hours
by cell surface calreticulin immunofluorescence. MCS (n = 10) were disaggregated with Accumax (AM‐105; Innovative Cell Technolo- gies Inc). Cells were then blocked in PBS/5% BSA and incubated with the cell‐impermeant SYTOX Green (10 µM) to reveal cells with compromised membrane integrity either by cell death or by handling. Cells were then incubated with anti‐calreticulin antibody (1:50) for 30 minutes at room temperature under gentle agitation.
A total of 2 × 104 cells were cytospun onto glass slides, fixed with 4% paraformaldehyde at room temperature for 10 minutes, and incubate 2 hours with Alexa Fluor 546 goat antirabbit IgG (1:200)
and TO‐PRO‐3 (1:1000). As a positive control of intracellular
calreticulin staining, MCS exposed only to the vehicle control were blocked in DPBS/5% BSA containing 0.1% saponin to permeabilize the cell membrane, and then processed for cell surface calreticulin immunofluorescence. Primary antibody was diluted in PBS/1%
BSA and secondary antibody and TO‐PRO‐3 were diluted in
antibody diluent solution (21544; EMD Millipore). In each experiment, negative control with the secondary antibody alone was included. Slides were mounted with ProLong Gold antifade (P36930; Life Technologies). Images were captured at 63× magnification with a Nikon C1 confocal microscope (Nikon instruments Inc).
⦁ | Patients with mesothelioma
Patients whose tumor samples were used in these studies were participants in a consented and Institutional Review Board‐approved biorepository and clinical database (Dana Farber/Harvard Cancer Center protocol #98‐063). They underwent surgical resection of MPM at Brigham and Women’s Hospital in Boston, MA without preoperative
chemotherapy. Samples of fresh tumor tissue from five patients were obtained from discarded portions of the resection specimen, deidentified and transferred in 4°C media by overnight courier for the preparation of ex vivo tumor fragment spheroids. A histological subtype of the tumors included in this study: biphasic (TFS #1, #2, #4), epithelioid (TFS #3), epithelioid myxoid subtype (TFS #5).

⦁ | Statistical analysis
Data are expressed as mean ± standard deviation (S.D.). Differences among the indicated conditions were analyzed by two‐way analysis of variance using a multiple comparison test (Bonferroni correction). A
P value of <0.01 was considered significant. GraphPad Prism was employed for statistical analysis (GraphPad Software Inc., La Jolla, CA).

⦁ | RESULTS

⦁ | In multicellular spheroids with low autophagy at baseline, autophagy can be upregulated by the TAT‐BECN1 peptide
Autophagy initiation, as measured by ATG13 puncta analysis, correlates with autophagic flux in multicellular spheroids (MCS).3,9 Thus, ATG13
puncta were assessed in MCS exposed to the cell‐permeant peptide
Tat‐BECN1, or to the vehicle control for 24 hours (Figure 1A). In MCS
with low autophagy, as shown previously,3,9 no ATG13 puncta were detectable at baseline; however, following Tat‐BECN1, ATG13 puncta were easily detectable, indicating an upregulation of autophagy. In MCS
with high autophagy at baseline, Tat‐BECN1 did not alter ATG13 puncta, suggesting that autophagy cannot be further increased by Tat‐ BECN1 in this MCS.
These findings were confirmed by the measurement of the
autophagic flux, in which the degradation function of lysosomes is inhibited and the resulting accumulation of lipidated microtubule‐
associated protein 1 light chain 3 isoform B (LC3B‐II) is measured by
immunoblotting.5,27 As found by measurement of ATG13 puncta, in the MCS with low autophagy at baseline, Tat‐BECN1 increased the autophagic flux significantly whereas, in the MCS with high autophagy at baseline, Tat‐BECN1 did not increase autophagic flux further (Figure 1B and 1C).

⦁ | In multicellular spheroids with low autophagy at baseline, upregulation of autophagy restores cell death following chemotherapy
In our previous studies, mesothelioma 3D models with high autophagy at baseline have shown greater cell death following

chemotherapy than 3D models with low autophagy at baseline.3,18 However, we did not know whether upregulation of autophagy would increase cell death following chemotherapy. Therefore, Tat‐BECN1
was tested in this setting.
In MCS with low autophagy at baseline, Tat‐BECN1 signifi-
cantly increased apoptosis following chemotherapy, as measured by nuclear condensation (Figure 2A) and by caspase 3/7 activity
(Figure 2B). In MCS with high autophagy at baseline, Tat‐BECN1
did not further increase apoptosis following chemotherapy. Interestingly, when the autophagy of the low autophagy MCS was restored to the level of the high autophagy MCS
(Figure 1A‐C), the level of apoptosis was also restored to a similar
level (Figure 2).
Of note, as in our previous studies,3,18 apoptosis was observed following 24 hours of chemotherapy; no apoptosis was observed at 16 hours of treatment.

⦁ | In multicellular spheroids with low autophagy at baseline, upregulation of autophagy enhances release of DAMPs
Autophagy has been shown in other systems to regulate DAMPs, including the release of HMGB1,10,14-16 release of ATP (extracellular or eATP),10,13,15,16 or the exposure of calreticulin on the cell surface (ectoCRT).12 Now, with a tool to upregulate autophagy, we addressed
whether Tat‐BECN1 would increase the release of three DAMPs

FIG U RE 2 In multicellular spheroids with low autophagy, Tat‐BECN1 enhances cell death following chemotherapy. MCS with either low (REN) or high autophagy (M28) at baseline were exposed to 10 µM Tat‐BECN1 (+), or vehicle control (–) as in Figure 1. Where indicated, carboplatin plus pemetrexed (C+P) was added for the last 16 hours or 24 hours. A, Apoptosis measured by nuclear condensation. In MCS with
low autophagy at baseline (REN), apoptosis did not increase after C+P alone but, following Tat‐BECN1, apoptosis increased significantly. In MCS with high autophagy at baseline (M28), apoptosis increased after C+P; apoptosis did not increase further with the addition of Tat‐BECN1. Of note, apoptosis was observed only at 24 hours. (*P < .01, different compared with its own CTRL; **P < .01, different compared to C+P 24 hours
without Tat‐BECN1; n = 3; mean ± SD). B, Caspase 3/7 activity (expressed as relative luminescence units, RLU) in MCS grown as in A. In MCS with low autophagy at baseline (REN), caspase activation did not increase after C+P alone but, following Tat‐BECN1, increased significantly. In MCS with high autophagy at baseline (M28), caspase activation increased after C+P; caspase activation did not increase further with addition of
Tat‐BECN1. As above, caspase activation was observed only at 24 hours. (*P < .01, different compared to its own CTRL; **P< .01, different compared with C+P 24 hours without Tat‐BECN1; n = 3; mean ± SD). MCS, multicellular spheroids; SD, standard deviation

FIG U RE 3 In multicellular spheroids with low autophagy, Tat‐BECN1 enhances DAMPs release. MCS with low autophagy at baseline (REN) were grown as in Figure 2. A, HMGB1 in the media was detected by immunoblotting; tubulin was measured to confirm the absence of cell debris in the media. HMGB1 was clearly detectable following Tat‐BECN1 and 16 hours of C+P. Whole cell homogenates of MCS (H) grown in the presence of vehicle control and in the absence of C+P were loaded as an HMGB1 positive control. A representative immunoblot of three
independent experiments is shown. B, Extracellular ATP (eATP) was assessed in the media of MCS grown as in A. Compared with media of MCS exposed only to the vehicle control, eATP increased significantly following Tat‐BECN1 and 16 hours of C+P. (*P< .01, different compared to its own CTRL; n = 3; mean ± SD). C, MCS were grown as in A to detect calreticulin exposed on the plasma membrane (ectoCRT) by cell surface immunofluorescence. Sytox Green (GREEN), a cell‐impermeant marker, was used to reveal cells with compromised membrane integrity (either
by cell death or by handling). Anti‐calreticulin antibody (RED) was used to detect either ectoCRT in cells negative for Sytox Green, or
intracellular calreticulin in cells positive for Sytox Green. TO‐PRO‐3 (BLUE) identified nuclei. Saponin was used to permeabilize the cell
membrane as a positive control for Sytox Green and intracellular calreticulin staining. Calreticulin was undetectable in MCS exposed to the vehicle control (−), Tat‐BECN1 alone, or C+P alone. Calreticulin was found on the surface of intact cells (12% ± SD) exposed to Tat‐BECN1 and to C+P for 16 hours, consistent with a DAMP localization pattern. Following 24 hours, however, calreticulin staining was seen only in cells with
Sytox Green positivity, suggesting cell death. Images representative of three independent experiments are shown. The enlargement in the panel Tat‐BECN1 plus C+P 16 hours shows a representative cell with calreticulin exposed on the cell surface. Scale bars = 10 µm. MCS, multicellular spheroids; SD, standard deviation [Color figure can be viewed at wileyonlinelibrary.com]

following chemotherapy. For the measurement of HMGB1 and eATP, MCS were necessarily grown in medium without serum; we confirmed that the absence of serum did not change autophagy or
the response to Tat‐BECN1 (see Figure S1).
In MCS with low autophagy at baseline, DAMPs release was not
evident at baseline or after chemotherapy or Tat‐BECN1 alone (Figures 3A‐C). However, when Tat‐BECN1 was given with che- motherapy, HMGB1, and ATP were released and calreticulin was
expressed on mesothelioma cells (12% [±SD] of the cells). In sum, in MCS with low autophagy at baseline, the upregulation of autophagy enhanced DAMPs release following chemotherapy.
In MCS with high autophagy at baseline, HMGB1 was released following chemotherapy alone. Interestingly, HMBG1 release did not
change further after addition of Tat‐BECN1 (Figure 4), which we had
shown did not increase autophagy further in this higher autophagy setting (Figure 1). Thus, it appeared that autophagy at baseline was sufficient for this DAMP to be released by chemotherapy.
Interestingly, following autophagy upregulation or with high autophagy at baseline, DAMPs release was observed after 16 hours of chemotherapy, indicating that these immunogenic molecules were released premortem (see Figures 2-4).

⦁ | In multicellular spheroids with high autophagy at baseline, inhibition of autophagy impairs DAMPs release
In our prior studies, MRT 68921 was shown to be an effective nontoxic inhibitor of autophagy.3 Therefore, with this tool, we could ask whether inhibition of autophagy would impair DAMPs release in the MCS with high autophagy at baseline. As above, for the measurement of HMGB1 and eATP, MCS were necessarily grown in medium without serum; we determined that the absence of serum did not change autophagy or the response to MRT 68921 (see Figure S1).

FIG U RE 4 In multicellular spheroids with high autophagy, Tat‐BECN1 does not improve HMGB1 release. MCS with high autophagy at baseline (M28) were grown as in Figure 2. HMGB1 in
the media was determined by immunoblotting. Whole cell homogenates of MCS (H) grown in the presence of vehicle control and in the absence of C+P were loaded as an HMGB1 positive control. Tubulin was probed to confirm the absence of cell debris in
media. In the media, HMGB1 was clearly detectable following 16 hours of C+P. Addition of Tat‐BECN1 did not improve or alter HMGB1 release. A representative immunoblot of three independent
experiments is shown. MCS, multicellular spheroids
In MCS with high autophagy at baseline, all three DAMPs were released following chemotherapy alone (Figure 5A‐C), as already shown for HMGB1 (Figure 4). Following MRT 68921,
HMGB1 and ATP release was significantly decreased and calreticulin was no longer detectable following chemotherapy (Figure 5). Thus, in MCS with high autophagy at baseline, the inhibition of autophagy with a small molecule inhibitor impairs DAMPs release following chemotherapy. As we reported pre- viously,3 inhibition of autophagy in the high autophagy MCS increases cell death following chemotherapy (Figure S2). How- ever, this increase of cell death following autophagy inhibition did
not result in DAMPs release (Figure 5A‐C), suggesting that the
cell death we observed after inhibition of autophagy was not immunogenic.
In our previous studies, ATG7 downregulation was also shown to inhibit autophagy.18 To confirm the results obtained with MRT 68921, we assessed the effect of downregulating ATG7 on the release of HMGB1. For these experiments, M28 cells were transfected with either scrambled RNAi or ATG7 RNAi and grown into MCS. First, we confirmed that ATG7 was successfully downregulated and that, in those MCS with downregulated ATG7, autophagy was inhibited (Figure S3A and B). Then, we looked at the effect of ATG7 downregulation on HMGB1 release. In MCS with control transfection treated with chemotherapy, release of HMGB1 was again observed (Figure S3C), as in the previous experiments (Figures 4 and 5A). However, in MCS transfected with ATG7 RNAi, no HMGB1 release was observed (Figure S3C). Thus, we show that, with either a chemical or genetic approach, inhibition of autophagy impairs DAMPs release following chemotherapy.

⦁ | In ex vivo tumors, autophagy is upregulated by TAT‐BECN1 peptide
Ex vivo tumors (or tumor fragment spheroids, TFS) generated from resected mesothelioma tumor allow us to study the responses of human mesothelioma cells in their tumor microenvironment, surrounded by other nonmalignant cells and tumor stroma.8 We have developed methods for studying cell death and autophagy using immunofluorescence to confirm that the findings are localized to the tumor cells.3,8,9,28 In particular, for these studies, we have determined that autophagy in the ex vivo tumor is well represented by ATG13 puncta analysis.3,9
For these studies, we first asked if the autophagy in tumors with low or absent autophagy could be upregulated. Ex vivo tumors were generated from mesothelioma tumors that had not been exposed to
chemotherapy. All ex vivo tumors were exposed to Tat‐BECN1 for
24 hours (Figure 6A‐B). At baseline, and as previously reported,3,9 ex vivo tumors were shown to have either very low autophagy (TFS‐low autophagy, TFS #1 and #2) or high autophagy (TFS‐high autophagy, TFS #3‐5). Following Tat‐BECN1, autophagy increased significantly in
the low autophagy as well as the higher autophagy tumors. Thus, all the ex vivo tumors, not just the low autophagy tumors, responded to

FIG U RE 5 In multicellular spheroids with high autophagy, inhibition of autophagy impairs DAMPs release. MCS with high autophagy (M28) at baseline were exposed to 1 µM MRT 68921 (+), or vehicle control (−) for 24 hours. Where indicated, carboplatin plus pemetrexed (C+P) was added for the last 16 hours or 24 hours. A, At the end of the 24 hours, HMGB1 in the media was detected by immunoblotting; tubulin was measured to confirm the absence of cell debris in the media. HMGB1 was clearly detectable following 16 hours of C+P. HMGB1 was not detectable following the addition of MRT. Whole cell homogenates of MCS (H) grown in the presence of vehicle control and in the absence of C
+P were loaded as an HMGB1 positive control. A representative immunoblot of three independent experiments is shown. B, Extracellular ATP (eATP) was assessed in the media of MCS grown as in A. Compared with media of MCS exposed only to the vehicle control, eATP increased significantly following 16 hours of C+P and following 16 or 24 hours of C+P in the presence of MRT (*P< .01, different compared with CTRL; n = 3; mean ± SD). However, eATP measured in the presence of MRT was significantly lower than following 16 hours of C+P alone (**P< .01, different compared with 16 hours C+P; n = 3; mean ± SD). C, MCS were grown as in A to detect calreticulin exposed on the plasma membrane by
cell surface immunofluorescence. Sytox Green (GREEN), a cell‐impermeant marker, was used to reveal cells with compromised membrane integrity (either by cell death or by handling). Anti‐calreticulin antibody (RED) was used to detect either ectoCRT in cells negative for Sytox Green, or intracellular calreticulin in cells positive for Sytox Green. TO‐PRO‐3 (BLUE) identified nuclei. Saponin was used to permeabilize the cell membrane as a positive control for Sytox Green and intracellular calreticulin staining. Calreticulin was undetectable in MCS exposed to the
vehicle control (–), or MRT alone. Calreticulin was found on the surface of intact cells (15% ± SD) exposed to C+P for 16 hours, consistent with a DAMP localization pattern. No calreticulin exposed on the surface was found in MCS exposed to MRT and treated with C+P. At 24 hours, calreticulin staining was seen only in cells with Sytox Green positivity, suggesting cell death. Images representative of three independent experiments are shown. The enlargement in the panel C+P 16 hours shows a representative cell with calreticulin exposed on the cell surface.
Scale bars: 10 µm. DAMPs, damage‐associated molecular pattern molecules; MCS, multicellular spheroids; SD, standard deviation [Color figure
can be viewed at wileyonlinelibrary.com]

Tat‐BECN1 with a significant increase in autophagy, as measured by ATG13 puncta.
Following Tat‐BECN1, ATG13 positivity was greater than 10% in all ex vivo tumors (Figure 6A‐B). Of note, ATG13 positivity ≥6% was previously identified as the cutpoint to identify better patient
outcome.3,9
⦁ | In ex vivo tumors, induction of autophagy enhances cell death following chemotherapy
Knowing we could upregulate autophagy, we now asked whether upregulation of autophagy would increase cell death following
chemotherapy in ex vivo tumors. Therefore, Tat‐BECN1 was tested
in this setting.

FIGU RE 6 In ex vivo tumors, Tat‐BECN1 increases autophagy. TFS were exposed to 10 µM Tat‐BECN1 (+), or the vehicle control (−) for 24 hours. TFS were then fixed, embedded in paraffin, stained for ATG13 (GREEN), cytokeratin to identify mesothelioma cells (CK,
RED), TO‐PRO‐3 to identify nuclei (BLUE), and imaged by confocal microscopy. Autophagy initiation, which is representative of the levels of autophagy in ex vivo tumors, was expressed as the percentage of ATG13‐positive MPM cells. A, Representative images of 5 TFS with either very low autophagy at baseline (0%, TFS #1 and #2) or high levels of autophagy at baseline (≥6% of ATG13‐positive MPM cells, TFS #3‐5) are shown, with the percentage of mesothelioma cells with ATG13 puncta indicated in parentheses. Magnified
views of the regions in the dashed boxes show representative cells with ATG13 puncta (arrow). Scale bars = 20 μm. B, Bars show the average percentage of ATG13‐positive MPM cells (ATG13+MPM cells) calculated from three repetitions of the experiment shown in A.
The percentage of ATG13‐positive MPM cells increased significantly following Tat‐BECN1 in all the TFS, confirming the ability
of Tat‐BECN1 to induce autophagy ex vivo (*P< .01, different compared to CTRL; n = 3; mean ± SD). MPM, malignant pleural
mesothelioma; SD, standard deviation; TFS, tumor fragment spheroids [Color figure can be viewed at wileyonlinelibrary.com]

FIG U RE 6 Continued

Apoptosis was assessed in the 5 chemonaïve ex vivo tumors following chemotherapy and Tat‐BECN1 by cleaved caspase 3 immunofluorescence (Figure 7A‐B). In ex vivo tumors with low
autophagy at baseline, there was no increase in apoptosis following chemotherapy alone, as previously shown.3 However, following
Tat‐BECN1, apoptosis following chemotherapy increased significantly.
In ex vivo tumors with higher autophagy at baseline, apoptosis was observed following chemotherapy alone; nonetheless, following Tat‐ BECN1, the apoptosis following chemotherapy increased significantly.
In all ex vivo tumors, apoptosis was observed following 24 hours of chemotherapy; apoptosis was not observed at 16 hours of treatment.

⦁ | In ex vivo tumors, upregulation of autophagy enhances the release of HMGB1
Now we were prepared to determine whether the upregulation of autophagy would increase the release of DAMPs in association with cell death from chemotherapy. For the ex vivo tumors, we studied HMGB1 alone, due to technical difficulties studying eATP and ectoCRT in the ex vivo model. Release of HMGB1 in the media was determined by HMGB1 immunoblotting of media precipitates (Figure 8).
In the ex vivo tumors with low autophagy at baseline (TFS #1 and #2), HMGB1 was not released following chemotherapy alone.
However, following Tat‐BECN1, HMGB1 release was evident
following chemotherapy. In the ex vivo tumors with high autophagy at baseline (TFS #3‐5), HMGB1 was released after chemotherapy
alone and, following Tat‐BECN1, HMGB1 release following
chemotherapy increased further. Of note, as in the MCS, HMGB1 release from ex vivo tumors with autophagy was observed after 16 hours of chemotherapy, indicating that this immunogenic mole-
cule is released premortem in ex vivo tumors (see Figures 7 and 8). Importantly, our data show that Tat‐BECN1 increases autophagy and HMGB1 release following chemotherapy in all the ex vivo
tumors, regardless of their level of autophagy at baseline.
4 | DISCUSSION

We had previously shown that a higher autophagy was an independent and strong marker of better patient outcome in mesothelioma.9 The reason for the association of higher autophagy with outcome was not known; nor was it clear whether autophagy was a driver of the outcome and could thus be modulated to improve outcome. Here we show that autophagy facilitates ICD, which can stimulate the immune response to a tumor. While the underlying mechanism by which autophagy affects the release of DAMPs and cell death is unknown, the clinical consequences of this observation are potentially important.
To show a role for autophagy, we needed a reliable method of modulating autophagy in our models. For their potential clinical relevance, we sought small molecules that could inhibit or upregulate autophagy, without nonspecific effects on DAMPs release or on cell death, and tested many of them in our 3D models (see Materials and Methods). MRT 68921, a small blocker of ULK1/2, was employed to inhibit autophagy, as we have previously shown,3 and its effect
confirmed by genetic inhibition using ATG7 RNAi.18 The cell‐permeable peptide Tat‐BECN1, which has been shown to enhance autophagy and cell death following chemotherapy in other
tumors,17,20,21 was tested here for the first time in 3D models and was found to be highly effective. The peptide contains an active region of beclin 1, a key autophagy regulator protein, and is thought to overcome the sequestration of beclin 1 by the negative regulator
of autophagy GAPR‐1.29 In MCS with low autophagy at baseline,
Tat‐BECN1 increased autophagy whereas, in the MCS with high
autophagy at baseline, Tat‐BECN1 had no effect, suggesting that, in this setting, endogenous beclin 1 was not sequestered. In the ex vivo models, Tat‐BECN1 increased autophagy in all tumors. In the ex vivo tumor models with low autophagy at baseline, the successful upregulation of autophagy by Tat‐BECN1 indicates that autophagy was suppressed but not defective. Interestingly, for the ex vivo tumor models with high autophagy, the fact that Tat‐BECN1 was able to

FIG U RE 7 In ex vivo tumors, Tat‐BECN1 enhances cell death following chemotherapy. TFS were exposed to Tat‐BECN1 (+) or vehicle control (−) as in Figure 6. Where indicated, carboplatin plus pemetrexed (C+P), was added for the last 16 hours or 24 hours. TFS were then fixed, embedded in paraffin, stained for cleaved caspase 3 (GREEN), cytokeratin to identify mesothelioma cells (CK, RED), TO‐PRO‐3 to identify nuclei (BLUE), and imaged by confocal microscopy. A, Representative images of the 5 TFS are shown, with the percentage of mesothelioma cells
positive for cleaved caspase 3 indicated in parentheses. Representative MPM cells positive for cleaved caspase 3 puncta are indicated by arrows. Scale bars: 20 μm. B, Bars show the average percentage of cleaved caspase 3‐positive MPM cells (cleaved casp.3+MPM cells) calculated from three repetitions of the experiment shown in A. Cleaved caspase 3‐positive MPM cells were found only following 24 hours of C+P alone or
in combination with Tat‐BECN1. In TFS #1 and 2, cleaved caspase 3‐positive MPM cells were detectable only when C+P was added concurrently with Tat‐BECN1. In the TFS #3 to 5, cleaved caspase 3‐positive MPM cells were detectable following 24 hours of C+P alone; addition of Tat‐BECN1 increased further the percentage of cells positive for cleaved caspase 3. These data show that Tat‐BECN1 potentiates
cell death following chemotherapy ex vivo. (*P< .01, different compared to its own CTRL; **P< .01, different compared to C + P 24 hours without Tat‐BECN1; n = 3; mean ± SD). MPM, malignant pleural mesothelioma; SD, standard deviation; TFS, tumor fragment spheroids [Color figure can be viewed at wileyonlinelibrary.com]

FIG U RE 8 In ex vivo tumors, Tat‐BECN1 enhances the release of
HMGB1. TFS were grown as in Figure 7. At the end of the 24 hours, HMGB1 in the media (M) was detected by immunoblotting;
pan‐keratin in the TFS homogenates (H) was measured to assess the
amount of MPM cells present in the TFS for each condition. In TFS #1 and #2, HMGB1 was only detectable in the media following Tat‐ BECN1 and 16 hours of C+P. In TFS #3 to 5, HMGB1 was detectable in the media following 16 hours of C+P alone; addition of Tat‐BECN1 further increased the amount of released HMGB1. These data show that Tat‐BECN1 enhances the release of HMGB1 ex vivo. MPM, malignant pleural mesothelioma; TFS, tumor fragment spheroids

increase autophagy further indicates that the baseline autophagy was not at a maximal level. This result is consistent with our finding that mesothelioma cells negative for ATG13 can be found in all the ex vivo tumors, even in those with high autophagy at baseline, and thus in all the ex vivo tumors there may be mesothelioma cells where autophagy can be restored.
With these valuable tools, we could then modulate autophagy and learn its contribution to the release of DAMPs by mesothelioma multicellular spheroids and ex vivo tumor. With the increase in autophagy, the low autophagy MCS regained the ability to release DAMPs and low/absent autophagy ex vivo tumors regained the ability to release HMGB1. Importantly, this release was only seen following chemotherapy; high autophagy alone did not cause DAMPs release, but appeared to prime the mesothelioma for a release of DAMPs following chemotherapy, as is characteristic of an ICD. Consistent with our view that autophagy upregulation facilitates ICD, a study in lung cancer has shown that autophagy upregulation with caloric restriction mimetics increases the release of ATP following chemotherapy and improves the antitumor efficacy of chemotherapy
in vivo.30 And, although the three DAMPs have not previously been studied together, the involvement of autophagy has been described separately for HMGB1,10,14-16 ATP,10,13,15,16 or calreticulin.12 Here, at least with one of our models (MCS), we were able to show for the first time that autophagy is required for the release of all three DAMPs after chemotherapy and that autophagy upregulation improves such release.
In mesothelioma, DAMPs release has been described previously; however, the possible relationship to autophagy in those studies was not explored. Interestingly, in two of the studies,31,32 the release of HMGB1 was induced by agents, a bromodomain inhibitor or curcumin, which have been shown to induce autophagy.33-35 In one of these studies, in addition, the release was shown to activate immune cells in vitro and to restore
an immune‐active environment in vivo.32 In other studies, HMGB1
release was found at baseline, without any stimulus, and was associated with inflammation and necrosis and, in patients with mesothelioma, the HMGB1 level in serum was shown to be a poor prognostic factor.36,37 In our hands, the release of HMGB1, and the other two DAMPs would appear to be a beneficial signal based on the association of the underlying autophagy with a better patient outcome. The DAMPs release observed in our study would also be unlikely to increase serum levels, so it may be that HMGB1 at a low level, released as a pulse following a stimulus, would have beneficial effects on the immune response, whereas HMBG1 released at a level high enough to raise serum levels would not. Perhaps, differences in timing and in the intensity of the signal can
explain how DAMPs can act in a dual role, either as pro‐tumor or
antitumor effectors22; such differences in signaling are well known to account for different biological responses in general.38-40
The release of all three DAMPs preceded cell death, suggesting that it was a specific response and not a nonspecific leakage from dead cells. In the ICD literature, and as we found, two of the DAMPs, eATP, and ectoCRT, are reported to be released before cell death.11,41 Interestingly and unlike our results, HMGB1 release is reported to follow cell death.11,41 Although the exact timing of HMGB1 release in our study may differ from the consensus for ICD, our findings are in agreement with others showing that HMGB1 can be released before cell death42 or even without cell death32 and still be able to stimulate the immune response.
The increase in autophagy enhanced not only DAMPs release but also cell death. We recognize that the relationship of autophagy and cell death is complex, with some studies showing that inhibition of autophagy enhances cell death,43-46 including an earlier study of our own,3 and others that appear to show the opposite.47-49 Our data in this study indicate that when autophagy is low, chemotherapeutic cell death is also low. Restoration of autophagy appears to restore cell death to the same level as seen in the higher autophagy models. Although we did not pursue the mechanisms underlying the connection between autophagy and apoptosis, we suggest that the known crosstalk between these two important processes50 requires autophagy to be active at baseline.

One limitation of our study is that we did not demonstrate that the DAMPs release can stimulate an immune response. However, in the study of Riganti et al32 in mesothelioma, and in another study on autophagy upregulation in a murine model of lung cancer,30 DAMPs release similar to what we observed has been shown to activate immune cells and improve antitumor efficacy of chemotherapy. Another limitation of our work is that we found only one autophagy inducer effective in our system. Future identification of more autophagy upregulators may allow us to improve autophagy further and perhaps also the DAMPs release.
In conclusion, we found that autophagy facilitates DAMPs release and cell death following chemotherapy in mesothelioma 3D models. We also showed that baseline autophagy could be upregulated to enhance ICD. Currently, immunotherapy is considered a potentially important therapy for malignant mesothelioma,51 a recalcitrant tumor for which few therapeutic options exist. We propose two possible clinical benefits of our findings: on the one hand, autophagy biomarkers such as ATG13 might help identify patients who would
respond better to chemotherapy and perhaps immune‐directed
therapy51 and, on the other hand, autophagy upregulation of poorly immune‐responsive tumors could perhaps render them more responsive. Understanding more about the relationship between
autophagy and ICD may lead to clinically useful approaches.

ACKNOWLEDGMENTS

The authors wish to thank Julianne Barlow and Corinne Gustafson, PhD from BWH for helping in the collection and shipping of the mesothelioma tumors samples. The authors also wish to thank the Meso Foundation for the previous support leading to the testing of autophagy modulators. In particular, the authors wish to thank the Simmons Mesothelioma Foundation for its generous support over the years; WGR and RB were supported by the International Mesothe- lioma Program at BWH.

CONFLICT OF INTERESTS

The authors declare that they have no conflict of interests.

ORCID

Carlo Follo http://orcid.org/0000-0001-9711-2655

REFERENCES

⦁ Bhutia SK, Mukhopadhyay S, Sinha N, et al. Autophagy: cancer’s friend or foe? Adv Cancer Res. 2013;118:61‐95.
⦁ Eskelinen EL. The dual role of autophagy in cancer. Curr Opin
Pharmacol. 2011;11:294‐300.
⦁ Follo C, Cheng Y, Richards WG, Bueno R, Broaddus VC. Inhibition of autophagy initiation potentiates chemosensitivity in mesothelioma.
Mol Carcinog. 2018;57:319‐332.
⦁ Janku F, McConkey DJ, Hong DS, Kurzrock R. Autophagy as a target for anticancer therapy. Nat Rev Clin Oncol. 2011;8:528‐539.
⦁ Klionsky DJ, Abdelmohsen K, Abe A, et al. Guidelines for the use and interpretation of assays for monitoring autophagy (3rd edition).
Autophagy. 2016;12:1‐222.
⦁ Levy JMM, Towers CG, Thorburn A. Targeting autophagy in cancer.
Nat Rev Cancer. 2017;17:528‐542.
⦁ Rosenfeldt MT, Nixon C, Liu E, Mah LY, Ryan KM. Analysis of
macroautophagy by immunohistochemistry. Autophagy. 2012;8:963‐969.
⦁ Kim KU, Wilson SM, Abayasiriwardana KS, et al. A novel in vitro
model of human mesothelioma for studying tumor biology and apoptotic resistance. Am J Respir Cell Mol Biol. 2005;33:541‐548.
⦁ Follo C, Barbone D, Richards WG, Bueno R, Broaddus VC. Autophagy
initiation correlates with the autophagic flux in 3D models of mesothelioma and with patient outcome. Autophagy. 2016;12:1180‐1194.
⦁ Hou W, Zhang Q, Yan Z, et al. Strange attractors: DAMPs and
autophagy link tumor cell death and immunity. Cell Death Dis. 2013;4:e966‐e966.
⦁ Kroemer G, Galluzzi L, Kepp O, Zitvogel L. Immunogenic cell death in
cancer therapy. Annu Rev Immunol. 2013;31:51‐72.
⦁ Li DD, Xie B, Wu XJ, et al. Late‐stage inhibition of autophagy
enhances calreticulin surface exposure. Oncotarget. 2016; 7:80842‐80854.
⦁ Martins I, Wang Y, Michaud M, et al. Molecular mechanisms of ATP
secretion during immunogenic cell death. Cell Death Differ. 2014;21:79‐91.
⦁ Thorburn J, Horita H, Redzic J, Hansen K, Frankel AE, Thorburn A.
Autophagy regulates selective HMGB1 release in tumor cells that are destined to die. Cell Death Differ. 2009;16:175‐183.
⦁ Zhang Q, Kang R, Zeh HJ, 3rd, et al. DAMPs and autophagy: cellular
adaptation to injury and unscheduled cell death. Autophagy. 2013;9:451‐458.
⦁ Zhong Z, Sanchez‐Lopez EandKarin M. Autophagy, inflammation, and
immunity: a troika governing cancer and its treatment. Cell. 2016;166:288‐298.
⦁ Shoji‐Kawata S, Sumpter R, Leveno M, et al. Identification of a candidate
therapeutic autophagy‐inducing peptide. Nature. 2013;494:201‐206.
⦁ Barbone D, Follo C, Echeverry N, et al. Autophagy correlates with the therapeutic responsiveness of malignant pleural mesothelioma in 3D models. PLoS One. 2015;10:e0134825.
⦁ Pasquereau‐Kotula E, Habault J, Kroemer G, Poyet JL. The
anticancer peptide RT53 induces immunogenic cell death. PLoS One. 2018;13:e0201220.
⦁ Kraya AA, Piao S, Xu X, et al. Identification of secreted proteins that reflect autophagy dynamics within tumor cells. Autophagy.
2015;11:60‐74.
⦁ Sharif T, Martell E, Dai C, et al. Autophagic homeostasis is required for the pluripotency of cancer stem cells. Autophagy. 2017;13:264‐284.
⦁ Hernandez C, Huebener P, Schwabe RF. Damage‐associated mole-
cular patterns in cancer: a double‐edged sword. Oncogene. 2016;35:5931‐5941.
⦁ Showalter A, Limaye A, Oyer JL, et al. Cytokines in immunogenic cell death: applications for cancer immunotherapy. Cytokine. 2017;97:
123‐132.
⦁ Bezu L, Gomes‐de‐Silva LC, Dewitte H, et al. Combinatorial strategies
for the induction of immunogenic cell death. Front Immunol. 2015;6:187.
⦁ Harasawa R, Mizusawa H, Nozawa K, Nakagawa T, Asada K, Kato I. Detection and tentative identification of dominant mycoplasma
species in cell cultures by restriction analysis of the 16S‐23S rRNA intergenic spacer regions. Res Microbiol. 1993;144:489‐493.
⦁ Barbone D, Yang TM, Morgan JR, Gaudino G, Broaddus VC. Mammalian target of rapamycin contributes to the acquired apoptotic resistance of human mesothelioma multicellular spheroids.
J Biol Chem. 2008;283:13021‐13030.
⦁ Mizushima N, Yoshimori TandLevine B. Methods in mammalian autophagy research. Cell. 2010;140:313‐326.

⦁ Abayasiriwardana KS, Barbone D, Kim KU, et al. Malignant mesothelioma cells are rapidly sensitized to TRAIL‐induced apoptosis by low‐dose anisomycin via Bim. Mol Cancer Ther. 2007;
6:2766‐2776.
⦁ Garcia‐Valtanen P, Ortega‐Villaizan Mdel M, Martinez‐Lopez A, et al. Autophagy‐inducing peptides from mammalian VSV and fish VHSV rhabdoviral G glycoproteins (G) as models for the development of
new therapeutic molecules. Autophagy. 2014;10:1666‐1680.
⦁ Pietrocola F, Pol J, Vacchelli E, et al. Caloric restriction mimetics enhance anticancer immunosurveillance. Cancer Cell. 2016;30:147‐160.
⦁ Miller JM, Thompson JK, MacPherson MB, et al. Curcumin: a double hit
on malignant mesothelioma. Cancer Prev Res (Phila). 2014;7:330‐340.
⦁ Riganti C, Lingua MF, Salaroglio IC, et al. Bromodomain inhibition exerts its therapeutic potential in malignant pleural mesothelioma by promoting immunogenic cell death and changing the tumor
immune‐environment. Oncoimmunology. 2018;7:e1398874.
⦁ Sakamaki JI, Wilkinson S, Hahn M, et al. Bromodomain protein BRD4 is a transcriptional repressor of autophagy and lysosomal function.
Mol Cell. 2017;66:517‐532.
⦁ Zhao G, Han X, Zheng S, et al. Curcumin induces autophagy, inhibits
proliferation and invasion by downregulating AKT/mTOR signaling pathway in human melanoma cells. Oncol Rep. 2016;35:1065‐1074.
⦁ Zhu Y, Bu S. Curcumin induces autophagy, apoptosis, and cell cycle
arrest in human pancreatic cancer cells. Evid Based Complement Alternat Med. 2017;2017:5787218‐13.
⦁ Jube S, Rivera ZS, Bianchi ME, et al. Cancer cell secretion of the
DAMP protein HMGB1 supports progression in malignant mesothe- lioma. Cancer Res. 2012;72:3290‐3301.
⦁ Yang H, Rivera Z, Jube S, et al. Programmed necrosis induced by
asbestos in human mesothelial cells causes high‐mobility group box 1 protein release and resultant inflammation. Proc Natl Acad Sci USA. 2010;107:12611‐12616.
⦁ Behar M, Hao N, Dohlman HG, Elston TC. Dose‐to‐duration encoding
and signaling beyond saturation in intracellular signaling networks.
PLoS Comput Biol. 2008;4:e1000197.
⦁ Bluthgen N. Signaling output: it’s all about timing and feedbacks. Mol Syst Biol. 2015;11:843‐843.
⦁ Purvis JE, Lahav G. Encoding and decoding cellular information
through signaling dynamics. Cell. 2013;152:945‐956.
⦁ Kepp O, Senovilla L, Vitale I, et al. Consensus guidelines for the detection of immunogenic cell death. Oncoimmunology. 2014; 3:e955691.
⦁ Zhao T, Ren H, Wang X, et al. Rituximab‐induced HMGB1 release is
associated with inhibition of STAT3 activity in human diffuse large B‐cell lymphoma. Oncotarget. 2015;6:27816‐27831.
⦁ Levy JM, Thompson JC, Griesinger AM, et al. Autophagy inhibition improves chemosensitivity in BRAF(V600E) brain tumors. Cancer
Discov. 2014;4:773‐780.
⦁ Liu Q, Zhu D, Hao B, Zhang Z, Tian Y. Luteolin promotes the sensitivity of cisplatin in ovarian cancer by decreasing PRPA1‐ medicated autophagy. Cell Mol Biol (Noisy‐le‐grand). 2018;64:17‐22.
⦁ Piya S, Andreeff M, Borthakur G. Targeting autophagy to overcome chemoresistance in acute myleogenous leukemia. Autophagy.
2017;13:214‐215.
⦁ Wang J, Tan X, Yang Q, et al. Inhibition of autophagy promotes apoptosis and enhances anticancer efficacy of adriamycin via augmented ROS generation in prostate cancer cells. Int J Biochem
Cell Biol. 2016;77:80‐90.
⦁ Ji MM, Wang L, Zhan Q, et al. Induction of autophagy by valproic acid enhanced lymphoma cell chemosensitivity through HDAC‐ independent and IP3‐mediated PRKAA activation. Autophagy.
2015;11:2160‐2171.
⦁ Kou B, Liu W, Xu X, et al. Autophagy induction enhances tetrandrine‐
induced apoptosis via the AMPK/mTOR pathway in human bladder cancer cells. Oncol Rep. 2017;38:3137‐3143.
⦁ Lin CI, Whang EE, Donner DB, et al. Autophagy induction with RAD001
enhances chemosensitivity and radiosensitivity through Met inhibition in papillary thyroid cancer. Mol Cancer Res. 2010;8:1217‐1226.
⦁ Rubinstein AD, Kimchi A. Life in the balance ‐ a mechanistic view of
the crosstalk between autophagy and apoptosis. J Cell Sci. 2012; 125:5259‐5268.
⦁ Ye L, Ma S, Robinson BW, Creaney J. Immunotherapy strategies for
mesothelioma ‐ the role of tumor specific neoantigens in a new era of precision medicine. Expert Rev Respir Med. 2019;13:181‐192. Tat-BECN1