GL261 cells expressing luciferase (GLUC2) were obtained from the lab of Dr. Michael Lim. They are derived from a chemically induced astrocytoma in C57BL/6 mice (29). KR158 cells were obtained from the labs of Drs. Tyler Jacks and Behnam Badie, and are derived from genetically engineered Nf1/Tp53 mutants (30). Cells were maintained in DMEM, 10% serum, 1% antibiotic, 1% sodium pyruvate and incubated at 37°C with 5% CO₂. bEND.3 cells were cultured in DMEM with serum as above. Primary patient-derived human glioma cells (GBM43) which carries Nf1 and Tp53 mutations were obtained from Dr. Jann Sarkaria at the Mayo Clinic from the xenograft cell line panel. To enrich for glioma stem-like cells (GSC) in GLUC2, KR158 and GBM43 cells, serum was reduced step-wise over a week as described previously (31). GSC were cultured in serum-free DMEM medium containing F12 supplement along with pyruvate, antibiotics, N2 supplement, EGF, FGF and heparin (31).

For single lucanthone treatment studies, GLUC2 and KR158 cells were plated at a density of 2,000 and 1,000 cells per well, respectively, in a 6-well plate and allowed to adhere overnight. They were then treated with 10 μM Lucanthone every 4 days for 12 days. On day 13, media were aspirated, and cells were fixed with 4% PFA for 10 minutes. Cells were then treated with 0.5% crystal violet solution for 20 minutes. Plates were washed and photographed.

For dual treatment studies (lucanthone and TMZ), GLUC2 and KR158 cells were plated at a density of 2,500 and 1,000 cells per well in a 12-well plate and allowed to adhere overnight. Cells were then treated with medium, TMZ, lucanthone, or a combination for 4 days. The media were aspirated, and the cells were washed with PBS once and incubated with standard medium for 3 days. The cells were fixed with PFA and treated with 0.5% crystal violet solution as above and photographed. Then lysis solution of 10% SDS in dH₂O was added to the plates overnight. To quantify relative crystal violet intensity, the absorbance of the crystal violet-containing supernatant was read under a spectrophotometer at 590 nm with a reference wavelength of 670 nm. Data are graphed as percent of control (medium only-treated cells).

Cells were plated in a 96-well plate and incubated overnight. Adherent tumor cells (2D cultures) were treated with lucanthone for 3 days and then subject to the MTT protocol as per manufacturer’s instructions (Promega). GSC (3D cultures) were treated with lucanthone for 5 days, as this allowed sufficient time for spheroids to grow in culture. Prior to addition of the MTT reagent, plates were imaged under confocal microscopy with the addition of Calcein-AM and Ethidium homodimer to mark live cells and dead cells, respectively.

GLUC2, KR158 and GBM43 cells were plated on glass-bottom 35mm plates overnight. They were then treated with medium or lucanthone for 48 hours. The cells were treated with 5μg/ml acridine orange for 15 minutes. Plates were washed with PBS 3x and then incubated in complete medium. Plates were then imaged for acidic vesicle accumulation (525/590nm) under confocal microscopy, according to manufacturer’s instructions (Cayman chemical).

For immunocytochemical analysis, GLUC2, KR158 and GBM43 cells were plated on glass coverslips overnight. Cells were treated with medium or lucanthone for 48 hours. The medium was aspirated and cells were fixed with 4% PFA for 10 minutes. Plates were then washed 3x with 0.3% TX-100 in PBS and wells were blocked with 3% normal goat serum/0.3% TX-100 in PBS for 1 hour. Cells were stained with primary antibodies overnight (LC3, Ki67, Nestin, Olig2, SOX2, CD133, p62, Cathepsin D, γH2AX). The primary antibody was removed, and cells were again washed 3x with 0.3% TX-100 in PBS after which time cells were incubated with fluorescent secondary antibodies for an hour at room temperature. Cells were then washed 3x with PBS, counterstained with DAPI and imaged under confocal microscopy. GSC were induced to adhere to glass slides by precoating glass slides with Geltrex for an hour.

Immunoblotting was done as described previously (3). Briefly, cells were lysed in 50mM Tris-HCl (pH 7.4) with 1% Nonidet P-40, 0.25% sodium deoxycholate, 150mM NaCl, 1% SDS and 1mM sodium orthovanadate. Proteins were denatured by boiling with treatment with BME. Proteins were run on SDS-page gels, transferred to PVDF membranes (Immobilon; Millipore). Membranes were washed with Tris-buffered saline with 0.1% Tween 20 and blocked in a 5% non-fat dry milk powder for 1 hour. Membranes were then probed for LC3 (1:1000), p62 (1:1000), Olig2 (1:1000), SOX2 (1:1000) and B-Actin (1:2000; sigma Aldrich). Membranes were rinsed in TBS-T, probed with associated HRP-conjugated secondary antibodies and exposed to Pierce ECL substrate for 1 minute (Thermo Fisher Scientific) after which x-ray films were developed from membranes.

To prepare RNA, GLUC2 spheroids were spun down and lysed with Trizol and processed using the manufacturers protocol. To obtain cDNA, one microgram of RNA was reverse transcribed on a Veriti thermocycler using the High Capacity cDNA Reverse Transcription Kit. Amplification was performed on a StepOnePlus real-time PCR machine using a SYBR green kit (Applied biosystems). Primer sequences are as follows: GAPDH forward, 5′-GCACAGTCAAGGCCGAGAAT-3′; GAPDH reverse, 5′-GCCTTCTCCATGGTGGTGGA-3′; Olig2 forward, 5′- CAAATCTAATTCACATTCGGAAGGTTG -3′; Olig2 reverse, 5′- GACGATGGGCGACTAGACACC -3′. GAPDH was used as an internal control.

C57Bl6 mice were bred under maximum isolation on a 12:12 hour light:dark cycle with food ad libitum.

Gliomas were established in 3-4 month old male and female mice as described previously (3, 4, 32). GLUC2 GSC were dissociated with accutase and counted. Mice were anesthetized with 20mg/kg avertin, a midline incision was made in the scalp, the skin retracted and a small burr hole was drilled in the skull at the following stereotactic coordinates from bregma: -1mm anteroposterior and +2 mediolateral. 1x10⁵ GLUC2 GSC resuspended in PBS were injected over a period of 2 minutes at a depth of 3mm. At the end of the injection, the needle was kept in the injection site for a further 3 minutes. After needle removal, the incision was sutured and mice were placed on a heating pad until they fully recovered from anesthesia. During the disease course if mice were found to have lost more than 15% of their initial body weight, they were euthanized. All animal procedures were approved by the Stony Brook University Institutional Animal Care and Use Committee.

GSC engraftment was visualized using the IVIS spectrum in vivo imaging system 7 days after inoculation and again on days 14 and 21. Briefly, mice were anesthetized using continuous isofluorane exposure. Their scalps were shaved. Mice were injected i.p. with 150mg/kg D-Luciferin, carefully placed in the IVIS spectrum machine and imaged every 3-4 minutes for 40 minutes. Relative signal was quantified by a researcher blinded to the treatment, and luminescence ratios of day 21 to day 7 were calculated to approximate disease progression throughout the course of treatment.

Lucanthone was supplied by Dr. Robert Bases. Lucanthone was solubilized in 10% DMSO, 40% HPCD in PBS. After confirming the presence of gliomas on day 7, mice were randomly divided to control and treatment groups, and treated with either saline or 50mg/kg Lucanthone i.p. every day from day 7 to day 20. On day 21, tumors were visualized by bioluminescent imaging, as above.

Mice were anesthetized with 20mg/kg avertin and transcardially perfused with 30ml PBS followed by 30ml 4% PFA in PBS. Brains were removed and post-fixed in 4% PFA in PBS overnight. They were dehydrated for 48 hours in 30% w/v sucrose in PBS. Brains were then embedded in optimal cutting temperature compound (OCT, Tissue-Tek) and 20μm coronal sections throughout the entire tumor were taken on a Leica cryostat (Nusslock, Germany) and collected on Superfrost plus microscope slides. To determine tumor volume, serial sections were taken from each animal and subjected to hematoxylin and eosin stain. Tumor volume was calculated as tumor area x 20 μm thickness, x number of slides (33).

For immunohistochemical analysis, slides were brought to room temperature, washed 3x with 0.3% TX-100 in PBS and then blocked with 1% BSA/0.3% TX-100 in PBS for 1 hour. Slides were incubated overnight with appropriate primary antibodies ( Supplementary Table 1 ). The primary antibody was removed and slides were washed 3x 0.3% TX-100 in PBS and incubated with appropriate secondary antibodies for 1 hour. Slides were washed 3x with PBS, and counterstained with DAPI. Immunoreactivity was visualized by confocal imaging using the Leica SP8-x system, with white light and argon lasers.

Data comparing two population means with a normal distribution were analyzed using Student’s t-test. Data with non-normal distributions were analyzed using a Mann-Whitney test. Differences in cumulative distributions were assessed with the Kolomogorov-Smirnov test. To assess for synergistic interactions, King’s synergy test was used (34–36). Blood vessel circularity was calculated using the equation Circularity=4*π*(area/(perimeter²)). Alpha value was set at 0.05 prior to starting experiments. Power analysis was used to determine the appropriate number of animals used in each experiment. Experiments were replicated with the two tumor lines. Statistical analysis was performed using Graphpad Prism (Graphpad Software Inc, La Jolla, CA).

To examine whether lucanthone affects the growth of the two murine glioma cell lines GLUC2 and KR158, lucanthone ( Figure 1A ) was added to glioma cultures at 10 μM every 4 days for 2 weeks ( Figure 1B ), which reflects concentrations observed in the serum of patients (26). The proliferation of both cell lines was hindered. To investigate whether the possible mechanism by which lucanthone acts on glioma cells engaged autophagy, we treated glioma cells with lucanthone for 48 hours, and then stained them with acridine orange, which accumulates in acidic vacuolar organelles and shifts from green to red fluorescence (37). In control conditions, only few lysosomes were present in the cell lines. After treatment with lucanthone, cultures in both cell lines exhibited a remarkable diffuse cytoplasmic staining of dilated lysosomes ( Figure 1C ) with a corresponding increase in LC3 punctae ( Figure 1D ). These data parallel what has been observed after treatment with chloroquine in other tumor types (37) and suggest that lucanthone targets lysosomes and affects autophagic function at clinically relevant concentrations.

We also assessed the levels of the autophagy cargo receptor p62 and Cathepsin D. P62 accumulates in cells in which autophagy has been functionally inhibited and Cathepsin D is a lysosomal aspartyl protease (27). Our data demonstrate that after 48 hours of lucanthone treatment, P62 and Cathepsin D increase in both glioma cell lines, though we note a higher relative increase of both proteins in KR158 cells ( Figures 1E, F ). These findings illustrate lucanthone’s ability to inhibit autophagy at clinically relevant concentrations.

To examine whether lucanthone exerts its functions by acting as an inhibitor of topoisomerase 2 or APE1, we assessed the extent to which lucanthone induced DNA damage in glioma cell lines. To that end, GLUC2 and KR158 cells were treated with lucanthone for 48 hours, after which levels of γH2AX, a DNA damage marker, were assessed (38). As a positive control, glioma cells were also treated with the FDA-approved topoisomerase 2 inhibitor etoposide. While etoposide produced a marked increase in γH2AX intensity, lucanthone only produced a minimal effect, indicating that it is exerting its effect most likely via autophagy inhibition. When the levels of cleaved caspase-3 were evaluated, only minimal induction of cleaved caspase-3 in GLUC2 and KR158 spheroids treated with 10 μM lucanthone for 48 hours were observed, indicating that lucanthone may not be inducing apoptosis in these glioma cell lines ( Figure S1 ), as was shown for another autophagy inhibitor, thymoquinone, which induces cathepsin-mediated, but caspase-independent cell death (39).

The interaction between lucanthone and TMZ was investigated by performing combination studies in vitro. First, we performed an MTT assay to determine minimally effective concentrations of lucanthone in both cell lines. Lucanthone exerted a dose-dependent reduction in cell viability, with an IC50 of approximately 11-13 μM ( Figure 2B ). Two-way ANOVA illustrated that both cell lines were similarly sensitive to lucanthone, implying that this drug may be useful regardless of driver mutations. These data also pointed towards the use of 1 μM lucanthone for the combination studies, since this concentration exerted minimal effects alone on both cell lines.

It has been reported that GL261 and KR158 cells exhibit striking resistance to TMZ in vitro (40, 41). Therefore, we treated GL261 and KR158 cells with control medium, either drug alone, or both drugs for 4 days, and then allowed cultures to recover for 3 days before analysis. In this extended treatment format, 1 μM lucanthone alone, or 50 μM TMZ or 100 μM TMZ produced only a modest effect on GL261 and KR158 cells ( Figures 2B, C ). However, crystal violet intensity was markedly decreased when cells were treated with a combination of lucanthone and TMZ ( Figures 2B, C , p<0.05, King’s synergy test). Our data, in agreement with previous studies on breast tumor cells (23), suggest that even lower doses of lucanthone may be useful when paired with standard of care therapies to slow glioma progression.

To understand why lucanthone may augment the anti-tumor effects of TMZ, we tested for changes in the levels of γH2AX, a marker of DNA damage, in both cell lines. After 48 hours, changes in γH2AX intensity were evident in cultures treated with TMZ, but not in those treated with lucanthone ( Figures 2D-F ). Cultures treated with both drugs exhibited slightly increased levels of γH2AX compared to cultures treated with TMZ alone, but this increase did not become statistically significant.

Cancer stem cells are defined as progenitor-like tumor cells that repopulate the tumor after what is considered “successful” treatment, driving tumor recurrence and fatality. It is now accepted that cancer stem cells (termed here GSC) are rapidly dividing (42) and resistant to both TMZ and radiation (43, 44). Recent data reveal that GSC preferentially rely on autophagy for their survival and resistance to TMZ (45, 46). To that end, we enriched for stemness characteristics in GLUC2 and KR158 cell lines (please see Materials and Methods). Both glioma cell lines grew as partially suspended spheroids. 1 week after culturing cells in stemness medium, GLUC2 spheroids stained positive for the stemness markers nestin, SOX2 and Olig2, while KR158 spheroids stained positive for nestin, CD133 and SOX2 ( Figure S2 ). Cells staining positive for these markers also stained positive for the proliferation marker Ki67, demonstrating that these cells are indeed actively proliferating. Additionally, western blot analysis indicates that GLUC2 spheroids express higher levels of SOX2 and Olig2, while KR158 spheroids express higher levels of SOX2 than their adherent counterparts ( Figure S2 ).

After determining that these cells expressed stemness markers, they were treated with increasing concentrations of lucanthone. Remarkably, doses as low as 3 μM produced a strong oncolytic effect in these GSC. Lucanthone reduced spheroid area in both cell lines ( Figures 3A, B ). Further, treatment with lucanthone in a dose-dependent manner resulted in reduced numbers of spheroids formed in culture and reduced viability of the cultures ( Figures 3C, D ). These data show that lucanthone may preferentially kill cells left behind after treatment with modalities such as TMZ and radiation. Additionally, the IC₅₀ of lucanthone was approximately 2μM for KR158 and GLUC2 GSC. This is in contrast to an IC₅₀ of 11-13 μM in cells cultured with serum. These data indicate that stem-like glioma cells may be more susceptible to autophagy inhibiting drugs like lucanthone.

To gain mechanistic insight into how lucanthone reduces stemness, we allowed KR158 an GLUC2 GSC to form spheroids for 10 days, then treated spheroids with 10 μM lucanthone for 48 hours and assayed for alterations in levels of LC3and p62. By western blot analysis, we observed that lucanthone increased p62 levels in GLUC2 and KR158 GSC and increased levels of LC3-II as well ( Figures 3E, G ). We also observed increases in LC3 punctae in spheroids by immunocytochemistry and immunoblotting ( Figures 3E-G ). These data illustrate that the drug probably acted in a similar manner to that observed in adherent 2D cultures. It is worthy to note that in control conditions, LC3 punctae were also observed in spheroids, suggesting a higher level of baseline autophagy in GSC and a higher reliance on autophagy in general. In addition to assessing for changes in autophagic flux, we assessed for changes in the levels of stemness markers after treatment. We observed a strong reduction in Olig2 intensity in lucanthone-treated cultures ( Figures 3H, I ), while expression on nestin and SOX2 did not change. Using RT-qPCR we found that lucanthone reduced Olig2 mRNA expression in GLUC2 spheroids by >60% ( Figure 3I ). Minimal changes were also observed in Ki67 in these cultures.

Despite multimodal treatment, the recurrence rate for glioblastoma is ~100%. It has also been proposed that glioma cells change throughout the course of treatment such that the cells that survive treatment are functionally different than the parental tumor (47–49). We tested whether Lucanthone was able to exert oncolytic effects on glioma cells that have been selected for their ability to resist the standard chemotherapy temozolomide, TMZ. To that end, we treated GLUC2 cells with two cycles (48 hours of treatment and 7 days recovery per cycle) of 250 μM TMZ and 3 cycles of 500 μM TMZ. After the selection, we noticed that the surviving cells started forming spheres in serum-containing medium, similar to the ones we observe when culturing these cells in stemness-promoting medium. These spheroids expressed the prototypic stemness gene CD133 whereas parental GLUC2 spheroids did not ( Figure 3J ), suggesting that glioma cells dynamically respond to genotoxic therapy by acquiring stem-like morphology and characteristics (47). Cells selected for TMZ resistance were also less sensitive to TMZ treatment than parental GLUC2 cells ( Figure S3 ). In spite of becoming more stem-like, these cultures were still markedly sensitive to 10 μM lucanthone ( Figure 3K ), suggesting that lucanthone could be used to slow the growth of TMZ-resistant malignant glioma cells.

To examine if lucanthone could target human glioma cells as well, we obtained patient-derived glioma cells from the Mayo Clinic (termed GBM43), which bear Tp53 and Nf1 mutations. After treatment with lucanthone, GBM43 cells exhibited a similar acridine orange cytoplasmic staining pattern as seen in GLUC2 and KR158 cells ( Figure 4A ). Additionally, there were modest increases in LC3 and P62 ( Figures 4B, C ), suggesting that autophagy was inhibited in these cells. After enriching for stem-like qualities in these cells, treatment with 10 μM lucanthone drastically reduced cell viability ( Figure 4D ) and completely inhibited spheroid formation in these cultures ( Figure 4E ). Taken together, these data show that lucanthone can be used to inhibit autophagy in mouse and human glioma cells.

To assess translational potential, the efficacy of lucanthone was investigated in a mouse model of glioma. GLUC2 GSC were allowed to form spheroids for 10 days in culture. The spheroids were mechanically dissociated and 100,000 GLUC2 cells were implanted in the striatum of mice. Tumors were allowed to form for 7 days. Tumor cell presence was confirmed using IVIS imaging system on day 7, after which mice were segregated into two groups: one group received saline every day until day 21 while the other group received 50mg/kg lucanthone every day until day 21. The animals were imaged on days 14 and 21 ( Figure 5A ). On day 14, 5 of the 7 control mice exhibited a 2-fold increase in luminescence. In contrast, only 1 of 8 lucanthone-treated mice experienced a two-fold increase in luminescence, suggesting that lucanthone mitigated tumor growth between days 7 and 14 (chi-squared test, p<0.05). By day 21, control (saline)-treated glioma-bearing mice experienced a ~200-fold increase in tumor luminescence compared to day 7, whereas lucanthone-treated mice experienced only a 10-fold increase in tumor luminescence ( Figures 5B, C ). Upon histological analysis, the tumors of lucanthone-treated mice were approximately 60% smaller than those of saline-treated animals ( Figures 5D, E ). Moreover, saline-treated mice experienced cachexia ( Figure 5F ), whereas lucanthone-treated mice did not experience significant weight loss throughout the course of treatment ( Figure 5F ).

Standard of care therapies for glioma enrich for tumor stem-like cells, which is thought to play a role in glioma recurrence (43, 44). However, the in vitro data described so far suggest that lucanthone may reduce stem-like qualities of glioma cells, rather than solely target non-stem glioma cells. To that end, we interrogated how lucanthone affects glioma stem-like cells in vivo. The expression of stemness genes such as Olig2 and SOX2 was assessed in experimental tumors. Initial examination revealed that the density of Olig2⁺ cells was highest near the periphery of the tumor ( Figures 6A, C, D ), though we did observe a significant number of Olig2⁺ cells near the core as well. These data agree with previous findings that Olig2⁺ glioma cells are present at increased numbers near the tumor periphery (50). According to the Ivy Glioblastoma Atlas, an anatomically annotated transcriptional dataset of human glioblastoma tumors (51), Olig2 expression is increased in areas of infiltrating tumor and cellular tumor, and reduced in areas of necrosis and around blood vessels ( Figure 6B ). These findings suggest that, with respect to spatial expression of Olig2, GLUC2 GSC may accurately reflect what is observed in the human disease.

In contrast to the abundant Olig2 expression observed in saline-treated mice, we noted a striking reduction in Olig2 positivity around the periphery of lucanthone-treated tumors and near the core of these tumors. Two-way ANOVA revealed that in both treatment conditions, Olig2 intensity is higher near the tumor border, and that lucanthone resulted in reduction of Olig2 intensity at the tumor periphery and in the tumor core ( Figure 6E ). Ki67 positivity was similar in both treatment conditions. Additionally, SOX2 expression was not significantly different between treatment conditions, which parallels the result when individual spheroids were treated with lucanthone in vitro. While lucanthone did not significantly modulate γH2AX in vitro, γH2AX positivity was modestly increased in vivo in lucanthone-treated tumors ( Figure S4 ). Increases in γH2AX were most likely restricted to glioma cells, as most of the cells that exhibited increases in γH2AX were not staining for the GAM marker, F4/80 ( Figure S4 ).

In addition to assessing for tumor-cell specific effects of lucanthone in vivo, the extent to which other cell types in the tumor microenvironment may have been functionally affected by lucanthone treatment was examined. Previously, evidence has been provided that in addition to directly targeting tumor cells, chloroquine (another autophagy inhibitor) normalized the formation of blood vessels in the tumor microenvironment by directly acting on endothelial cells (52). Chloroquine augmented Notch1 signaling in endothelial cells, and as a consequence, reduced the blood vessel tortuosity and increased blood vessel patency. Because lucanthone and chloroquine exert their effects by a similar mechanism, we hypothesized that lucanthone may also modulate blood vessel formation in developing gliomas. To examine this possibility, tumor sections were stained for CD31, an endothelial cell marker. Blood vessel area, luminal area and overall blood vessel circularity were assessed. Interestingly, large blood vessels were observed in control tumors, though many of them exhibited a small luminal area. Accordingly, there were multiple tortuous blood vessels with minimal circularity. In lucanthone-treated tumors, the blood vessels were smaller, but those blood vessels typically showed an increased luminal area and the blood vessels themselves were more circular, suggesting that lucanthone may indeed be functionally affecting angiogenesis in the tumor microenvironment, potentially by acting directly on endothelial cells ( Figures 7A-D ). CD31 intensity was also diminished in lucanthone-treated tumors ( Figures 7A-D ). To examine if lucanthone acted directly on endothelial cells, we treated bEND.3 cells with lucanthone for 72 hours. Lucanthone exerted a dose-dependent effect on the cells, as at 20μM it significantly reduced bEND.3 cells viability after incubation for 72 hours ( Figure S5 ).

To interrogate functional outcomes of normalized tumor vasculature, the extent to which tumors exhibited evidence of hypoxia was assessed. In addition to proteins such as Hif1a/Hif2a, there are multiple other proteins induced in areas of tumor hypoxia, including Carbonic Anhydrase IX (CAIX) and Glut1 (53). Tumors in both treatment conditions displayed little CAIX positivity. While control-treated tumors displayed remarkable Glut1 positivity, specifically in necrotic tumor areas ( Figure 7E ), lucanthone-treated tumors displayed minimal Glut1 positivity ( Figure 7F ). Quantification of Glut1 intensities is shown in Figure 7G . Glut1 expression in control tumors also mirrors expression patterns observed clinically ( Figure 7H ). These data illustrate that in addition to tumor-cell specific effects, lucanthone may modulate additional parameters of the tumor microenvironment. While Glut1 was reduced throughout the tumor, we observed that another glucose transporter, Glut4, was expressed throughout the tumor in saline- and lucanthone-treated conditions ( Figure S6 ), suggesting that glucose transporter expression is not globally affected. As well, we observed an increase in the amount of cytotoxic T cells in the center of tumors in mice treated with lucanthone, which suggests that there may be a relief in the immunosuppressive nature fostered by gliomas after treatment ( Figure 7I ).

Targeting lysosomes is thought to exert effects on multiple cell types in the glioma microenvironment, potentially including GAM. Therefore, we assessed for differences in myeloid cell populations by staining for P2RY12 and TMEM119. In accordance with our previous work (32), P2RY12+ cells appeared mainly around the rim of gliomas in both treatment conditions ( Figure S7A ). However, we detected TMEM119+ cells throughout control-treated tumors and to a lesser extent in lucanthone-treated tumors ( Figures S7B, C ).