Autophagy inhibition signals through senescence to promote tumor suppression

ABSTRACT

Macroautophagy/autophagy, a stress-responsive cellular survival mechanism, plays important and context-dependent roles in cancer, and its inhibition has been implicated as a promising cancer therapeutic approach. The detailed mechanisms underlying the function of autophagy in cancer have not been fully understood. In this study, we show that autophagy inhibition promotes both the efficacy of chemotherapy for the treatment of glioblastoma (GBM) and therapy-induced senescence of GBM cells. As a specific cell fate characterized by permanent cell cycle arrest, senescence is also associated with the expression of a panel of specific secreted protein factors known as senescence-associated secretory phenotype (SASP). Intriguingly, we found that autophagy inhibition not only quantitatively enhanced GBM cell senescence but also qualitatively altered the spectrum of SASP. The altered SASP had increased potent activity to induce paracrine senescence of neighboring GBM cells, to skew macrophage polarization toward the anti-tumor M1 state, and to block the recruitment of pro-tumor neutrophils to GBM tumor tissues. Taken together, this study reveals novel functional communication between autophagy and senescence and suggests cancer therapeutic approaches harnessing autophagy blockage in inducing senescence-mediated antitumor immunity.

Introduction

Macroautophagy (referred to as autophagy hereafter) is a stress-responsive, lysosome-mediated catabolic process that plays important roles in a plethora of physiological and pathological processes [1,2]. The role of autophagy in cancer has been intensively studied and has been shown to be highly context dependent. While autophagy can suppress the transformation of normal cells to cancerous cells, once the transformation is accomplished, cancer cells often rely on autophagy to overcome numerous stressful conditions, such as nutrient deprivation and mechanical stress. As such, autophagy inhibition has emerged as a promising cancer therapeutic approach, especially in combination with other specific targeted therapies [3–6]. Remarkably, it has been demonstrated recently that cancer cell autophagy promotes immune evasion by degrading MHC-1 molecules in the cancer cells, and thus autophagy inhibition may enhance the therapeutic effect of immune checkpoint blockade [7]. Interestingly, recent work from us and others have shown that autophagy can also regulate another cellular process relevant to cancer, senescence [8–10].

Cellular senescence is considered as a permanent state of cell cycle arrest triggered by a variety of stimuli, such as DNA damage, telomere erosion, oncogene activation, and epigenetic stress [11,12]. Senescent cells are characterized by enlarged and flattened cellular and nuclear morphology, accompanied with a series of molecular modulations, such as, among others, upregulation of senescence-associated (SA) GLB1/β-galactosidase activity and secretion of a bioactive senescence-associated secretory phenotype (SASP) consisting inflammatory cytokines, chemokines, growth factors, and proteases. Senescence has been implicated in various deteriorating conditions such as cancer, neurodegeneration, inflammation, and aging [13–15]. In cancer, tumor cell senescence prevents tumor growth, whereas SASP may elicit both antitumor immunity and tumorigenic immunity, depending on the specific composition of SASP generated by senescent cells [16–18]. Although numerous genes (such as tumor suppressors TP53/p53 and RB1, and immune regulator NFKB/NF-κB) have been shown to regulate senescence, the exact mechanisms governing the execution and regulation of senescence are still poorly defined [19–23].

Recent evidence suggests that autophagy and senescence might functionally interact with each other. It has been reported that loss of nuclear LMNB1 (lamin B1), a critical event for senescence, is mediated by LMNB1 interaction with autophagy protein LC3 followed by its degradation through the autophagy-lysosomal pathway [24,25], and that inhibition of autophagy enhances cellular senescence in glioblastoma cells to impede gliomagenesis [10]. In the current study, we demonstrate that in combination with the chemotherapeutic agent temozolomide, autophagy inhibition promotes glioblastoma cell senescence and alters SASP expression pattern of the senescent cells, leading to reduction of tumor cell proliferation, paracrine induction of senescence in neighboring tumor cells, and establishment of antitumor immunity in the tumor microenvironment. As such, senescence mediates the anticancer effect of autophagy inhibition via both cell autonomous and non-cell autonomous mechanisms.

Results

Autophagy inhibition enhances the potency of chemotherapy and chemotherapy-induced senescence in a xenograft model for glioblastoma

Our previous finding that suppression of autophagy impedes glioblastoma (GBM) development [10] prompted us test the potential anticancer effect of autophagy inhibition in combination with the first-line chemotherapeutic agent for GBM, temozolomide (TMZ). For this purpose, we evaluated the effect of the combinatorial regimen in a xenograft mouse model. In this model, GBM tissues were derived from xenografted human GBM U87MG cells, and autophagy inhibition in these cells was achieved by a doxycycline (dox)-induced shRNA against ULK1, the autophagy-initiating protein kinase (Fig. S1) [26]. In agreement with clinical outcome, TMZ treatment significantly inhibited the growth of tumor xenografts. Strikingly, comparing to TMZ alone, the combinatorial regimen (TMZ plus dox) resulted in better outcome, especially when TMZ was administrated at a lower dosage (5 mg/kg) (Figure 1A-C). Further, immunohistochemistry staining (IHC) of tumor xenograft sections upon dox administration showed an absence of ULK1, accompanied with an accumulation of SQSTM1/p62 (Figure 1D), which is a substrate for autophagic degradation [27], indicating an effective autophagy inhibition.

Figure 1.

ULK1 knockdown potentiates the anticancer effect of temozolomide in a glioblastoma of xenograft mouse model. (A) Schematic plot for the experimental layout. Xenograft tumors were generated by subcutaneously injecting into nude mice U87MG cells expressing doxycycline-inducible ULK1-shRNA. ULK1 shRNA expression in xenograft tumors was induced by doxycycline diet for 7 days. Mice were subsequently treated with oral gavage of temozolomide (TMZ) for 5 days and tumor size was monitored for additional 23 days. (B) Xenograft tumor growth is shown as a function of time (left) and at endpoint (right) in absence or presence of doxycycline and treated with vehicle or TMZ (5 mg/kg or 25 mg/kg). (C) Images (Left) and tumor weight (Right) of xenograft tumors dissected at end point. (D-E) IHC staining of ULK1, SQSTM1, GLB1, TP53, γH2AX and CDKN1A in xenograft tumors following indicated treatments. Quantification of positive cells in each group is indicated in black. Representative experiments of n = 3.

We have shown that TMZ can induce GBM cell senescence in vitro [10] . Consistently, compared to tumor xenograft sections from mice treated with TMZ alone, those subject to the combinatorial regimen showed significant increase of cells positive for TP53, CDKN1A/p21, γH2AX, and GLB1 (galactosidase beta 1) (Figure 1E), indicating enhanced senescence. Taken together, our in vivo data suggests that autophagy inhibition potentiates chemotherapy-induced senescence of GBM tissues, which likely contributes to the enhanced anticancer effect of the combination treatment.

Given that ULK1 might possess autophagy-independent function, we sought to further validate the role of autophagy in senescence by testing other essential autophagy proteins in human GBM cell lines, U87MG and A172. By using senescence associate (SA)-GLB1 staining, senescent cells (SN cells) were visualized as enlarged blue cells. As expected, TMZ treatment caused an increased number of SA-GLB1⁺ cells (Figure 2A, B), Fig. S2A and S2C), loss of LMNB1, and upregulation of TP53 and CDKN1A, in both a dose-dependent and time-dependent manner (Figure 2C, Fig. S2B and S2D). C₁₂FDG-based fluorescence coupled with flow cytometry was also used to monitor TMZ-induced senescence (C₁₂FDG becomes fluorescent after cleavage by SA-GLB1) (Figure 2D, Fig. S2E). All these results indicate that TMZ can induce senescence in GBM cells. Importantly, when autophagy-essential gene, ATG3 or ATG5, was knocked out in these cells by CRISPR-Cas9 system, we observed complete blockage of autophagy (Figure 3A) and a significantly increased level of TMZ-induced senescence (Figure 3B-D).

Figure 2.

TMZ stimulates senescence of glioblastoma cells in a dose-dependent manner. (A-B) Cytochemical staining for SA-GLB1 activity of U87MG cells (A) and A172 cells (B) after an exposure to DMSO (as a control) or TMZ at indicated dosage (9 days for U87MG and 4 days for A172), and quantification of the percentage of U87MG cells positive for SA-GLB1 activity staining by Image J (Right). Scale bar: 200 μm. Unpaired t-test. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001; ns, no significance. Data presented as mean ± s.d. (n = 3). (C) Immunoblot of LMNB1, TP53, CDKN1A and GAPDH (loading control) in U87MG cells and A172 cells, after treating with indicated dose of TMZ for 8 days and 4 days, respectively. Representative experiments of n = 3. (D) Representatives of C₁₂FDG flow cytometric histogram of U87MG cells (Left) and A172 cells (Right) after exposure to TMZ of indicated dosage. The separated line indicates the cutoff in intensity level used to define senescence. Representative experiments of n = 3.

Figure 3.

Blockage of autophagy promotes TMZ-induced GBM cell senescence. (A) Immunoblot of SQSTM1, ATG3 (in U87MG cells) or ATG5 (in A172 cells), LC3, and GAPDH (loading control) in wild-type (WT) and ATG3^−/− U87MG cells (left) or WT and ATG5^−/− A172 cells (Right). Cells were treated as indicated with torin 1 (1 μM) for 6 h, bafilomycin A₁ (100 ng/ml) for 2 h, or a combination of torin 1 for 6 h with bafilomycin A₁ for the last 2 h. Representative experiments of n = 3. (B) Cytochemical staining for SA-GLB1 of WT and ATG3^−/− U87MG cells (Left) or WT and ATG5^−/− A172 cells (Right) after a 9-day (for U87MG cells) or 4-day (for A172 cells) exposure to DMSO or indicated dose of TMZ, scale bar: 200 μm; the percentage of cells staining positive for SA-GLB1 activity was quantified by Image J (bottom). (C) Immunoblot of ATG3 (in U87MG cells) or ATG5 (in A172 cells), LMNB1, TP53, CDKN1A, GAPDH in WT and ATG3^−/− U87MG cells (left) or WT and ATG5^−/− A172 cells (Right) after treatment of TMZ at indicated dosage. Representative experiments of n = 3. (D) Representatives of flow cytometric histogram detecting SA-GLB1 activity using C₁₂FDG, in WT and ATG3^−/− U87MG cells (Left) or WT and ATG5^−/− A172 cells (Right) after treatment with TMZ at indicated dosage. The separated line indicates the cutoff in intensity level used to define senescence. Representative experiments of n = 3.

Autophagy modulates SASP profile both quantitatively and qualitatively

In addition to the cell-autonomous arrest of proliferation, senescent cells can also impact its environment by secreting a cohort of bioactive molecules, i.e., SASP. Recent progress indicates that the composition of SASP can be highly context-dependent, and sometime may even exerts opposite biological effect on neighboring cells [12,18,28–30]. As our study indicates an effect of autophagy on GBM senescence quantitatively, we further asked whether autophagy can also have a qualitative effect on senescence via altering the composition of the SASP profile. For this purpose, we developed a C₁₂FDG-based cell sorting method to isolate senescent cells from cells exposed to TMZ (Figure 4A). Six cell subpopulations were obtained: A+/D-FITC^low, A-/D-FITC^low, A+/T-FITC^low, A+/T-FITC^high, A-/T-FITC^low, A-/T-FITC^high (A+: autophagy intact; A-: autophagy deficient; D: DMSO; T: TMZ). As expected, compared to FITC^low cells, FITC^high cells showed flattened and enlarged morphology and possessed increased SA-GLB1 activity, and alteration of the expression of LMNB1, TP53 and CDKN1A, all indications of senescence (Figure 4B, Fig. S3A and S3B). Therefore, we designate the six cell subpopulations as A+/D-NS, A-/D-NS, A+/T-NS, A+/T-S, A-/T-NS and A-/T-S (NS: Non-Senescent cells; S: senescent cells). Notably, an upregulated expression of TP53 and CDKN1A was also observed in T-FITC^low A172 cells, likely a response to TMZ-induced DNA damage (Fig. S3B).

Figure 4.

Autophagy modulates the SASP profile of senescent cells. (A) A schematic diagram showing the isolation of senescent U87MG cells after TMZ treatment. Fluorescence-activated cell sorting (FACS) based on C₁₂FDG staining was performed to isolate several subpopulations harboring distinguished SA-GLB1 activity from WT and ATG3^−/− U87MG cells exposed to DMSO or TMZ (15 μM for 9 days). Cells without C₁₂FDG labeling was served as a negative control for setting up a boundary in fluorescence histogram to define negative (C₁₂FDG^lo) and positive cells (C₁₂FDG^hi). (B) Cytochemical staining for SA-GLB1 activity of the sorted six subpopulations as described in (A). Representative experiments of n = 3. (C) Human cytokine array for conditioned media (CM) from sorted U87MG subpopulations of A+/D-C₁₂FDG^lo, A-/D-C₁₂FDG^lo, A+/T-C₁₂FDG^hi and A-/T-C₁₂FDG^hi (Left). As shown in Figure 4B, C₁₂FDG^hi cells were senescent cells (SA-GLB1 activity positive), we redefined these subpopulations as: A+/D-NS, A-/D-NS, A+/T-S and A-/T-S. (NS: non-senescent cells, indicating C₁₂FDG^lo; S: senescent cells, indicating C₁₂FDG^hi). Profiles of mean spot pixel density were created using a transmission-mode scanner and Image J (Right). Representative experiments of n = 3. (D) Real-time qPCR analysis of SASP genes in sorted six U87MG subpopulations of A+/D-NS, A-/D-NS, A+/T-NS, A-/T-NS, A+/T-S, and A-/T-S. Data are relative to expression of DMSO-treated proliferating WT U87MG (A+/D-NS), which is defined as 1, normalized to the average expression of housekeeping gene GAPDH. Unpaired t-test. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001; ns, no significance. Values are mean ± s.d. (n = 3).

To examine the impact of autophagy on SASP, we FACS-sorted those individual subpopulations by using wild-type (WT) or ATG3 KO U87MG cells, incubated them separately and harvested conditioned media (CM) from each subpopulation, and then analyzed the secretome of each CM by using a human cytokine array containing 105 cytokines (R&D Systems). Using this approach, we detected a clear accumulation of multiple canonical SASP components (such as IL6, IL8, soluble PLAUR/uPAR [sPLAUR/uPAR]) and a plethora of other cytokines in the CM from TMZ-induced senescent cells, in comparison with non-senescent counterparts (Figure 4C, Fig. S4A). Importantly, ATG3 KO caused a further increase of IL6, IL8, sPLAUR, CXCL1, CSF3/GCSF, CSF2, among others, in comparison with WT senescent cells. This result is consistent with our observation that autophagy inhibition promotes cellular senescence. Unexpectedly, ATG3 KO did not enhance the expression of all SASP components in the senescent cells; it even caused reduction of several SASP components, such as GDF15, PTX3, and CCL2/MCP1 (Figure 4C, Fig. S4A). The effect of autophagy inhibition on SASP profile in A172 cells is highly similar to that in U87MG cells, as analyzed by qPCR analysis (Figure 4D, Fig. S4B). Taken together, we demonstrate for the first time that autophagy modulates SASP both quantitatively and qualitatively.

Autophagy blockage promotes the anti-proliferative activity of SASP

What is the biological consequence of the alteration of SASP by autophagy inhibition? To tackle this question, we again generated CM from senescent cells with defined autophagy genotype and tested how these CM affect proliferating cancer cells. Compared to CM derived from non-senescent, autophagy-competent GBM cells, CM from their senescent counterparts significantly inhibited the proliferation of naïve GBM cells (Fig. S5A and S5B); inhibition of the proliferation was more pronounced when naïve GBM cells were incubated with CM derived from autophagy-defective senescent GBM cells (Figure 5A, Fig. S5A and S5B). IL8 and CXCL1, SASP components whose expression was significantly stimulated by autophagy inhibition (Figure 4C, D), were responsible for the observed proliferative inhibition, because neutralizing antibodies against IL8 and CXCL1 ablated the inhibitory activity of the CM derived from autophagy-defective senescent cells, but control antibody failed to do so (Figure 5B).

Figure 5.

The effect of SASP on GBM cell proliferation and paracrine senescence. (A) Individual CMs were derived from FACS-sorted cell populations as indicated, diluted with fresh media at a ratio of 2:1, and then incubated with U87MG cells. Under each condition, cell growth was assessed by cell counting at indicated time (Left) and by crystal violet staining at Day-9. Unpaired t-test. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001; ns, no significance. Values are mean ± s.d. (n = 3). (B) Individual CMs were derived from FACS-sorted cell populations as indicated, diluted with fresh media at a ratio of 2:1, and then incubated with U87MG cells, in presence of IgG, neutralizing antibody against human IL8 (2 μg/ml), CXCL1 (2 μg/ml). Under each condition, cell growth was assessed by cell counting at indicated time (Left) and by crystal violet staining at day-9. Unpaired t-test. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001; ns, no significance. Values are mean ± s.d. (n = 3). (C) Representative of flow cytometric histogram detecting SA-GLB1 activity using C₁₂FDG, in proliferating U87MG cells after a 9-day incubation with fresh media or CM derived from indicated FACS-sorted cell populations (diluted with serum-free (basal) media at a ratio of 1:4). The dotted line indicates the cutoff intensity used to define senescence. Representative experiments of n = 3. (D) Representative of flow cytometric histogram detecting SA-GLB1 activity using C₁₂FDG, in proliferating U87MG cells after a 9-day incubation with CM derived from sorted A+/T-S or A-/T-S cells (diluted with serum-free (basal) media at a ratio of 1:4) in presence of IgG, neutralizing antibodies against human IL8, CXCL1, and VEGF, as indicated. Representative experiments of n = 3.

Interestingly, upon incubation with the proliferation-inhibitory CM, some GBM cells adopted the morphological feature of senescence (we did not observe obvious cell death), suggesting that SASP might induce paracrine senescence under these conditions, and this may contribute to the observed inhibition of proliferation. To test this hypothesis, we used C₁₂FDG-based flow cytometry to measure senescence. Indeed, CM derived from senescent GBM cells induced senescence of naïve GBM cells, and CM from autophagy-defective senescent cells was a more potent inducer (Figure 5C, Fig. S5C and S5D). Further, we found that neutralizing antibody against VEGF, a reported promotor of paracrine senescence [30], as well as neutralizing antibody against IL8 or CXCL1, attenuated paracrine senescence (Figure 5D), indicating that these SASP components contribute to the induction of paracrine senescence.

Autophagy-modulated SASP regulates macrophage polarization

Besides tumor cells, there are numerous noncancerous cells in the tumor microenvironment (TME), such as various type of immune cells including macrophages. Macrophages are divided into two subclasses: “classically activated” M1 macrophages that mediate inflammation in response to pathogens and tumor cells, and “alternatively activated” M2 macrophages that exert immune suppressive phenotype by promoting angiogenesis and tissue remodeling [31]. As the process for macrophages to develop into a specific subclass, or macrophage polarization, is regulated by various cytokines, including some SASP components, we sought to determine how tumor cell autophagy may influence macrophage polarity via modulating tumor cell senescence and SASP. By incubating human monocytes-derived macrophages with CM from senescent GBM cells, we found that CM from autophagy-defective senescent cells produced a substantial increase in the M1 polarization markers CXCL10, CCR7, and IL12, and a decrease in the M2 polarization markers CCL22 and PPARG/PPARγ, in comparison with CM made from the WT senescent cells (Figure 6A, B, Fig. S6). Importantly, this effect was attenuated by neutralizing antibody against IL8 and CXCL1 (Figure 6C). In addition to those up-regulated SASP components, the down-regulated SASPs also called our attention considering the reported role of GDF15 and CCL2 in immunity [32,33]. As expected, supplement of recombinant human GDF15 and CCL2 significantly impaired M1 polarization caused by CM from autophagy-defective senescent cells (Figure 6D). Taken together, inhibition of autophagy in senescent GBM cells promoted macrophage M1 polarization, likely via altering expression of specific SASP components including IL8, CXCL1, GDF15, and CCL2.

Figure 6.

Autophagy inhibition promotes M1 macrophage polarization through modulating TMZ-induced SASP in GBM cells. (A) A schematic diagram and microscopic images representing the induction of U937 cells to macrophages by PMA (100 ng/ml). Scale bars: 200 μm. Representative experiments of n = 3. (B) Real-time PCR analysis of macrophage polarization markers (M1 markers: CXCL10, CCR7, IL12; M2 markers: CCL22, PPARG) in U937-derived macrophages after incubation with fresh media or CM from FACS-sorted U87MG cell populations A+/D-NS, A-/D-NS, A+/T-S, and A-/T-S (diluted with fresh media at a ratio of 3:2). Data are relative to gene expression levels of macrophages incubated with fresh media, normalized to the average expression of housekeeping gene GAPDH. Unpaired t-test. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001; ns, no significance. Values are mean ± s.d. from 3 replicates. (C-D) Real-time PCR analysis of macrophage polarization markers in U937-derived macrophages incubated with CM derived from FACS-sorted A+/T-S or A-/T-S U87MG cells (diluted with fresh media at a ratio of 3:2) in the presence of IgG, or neutralizing antibody against human IL8 or CXCL1, or recombinant human CCL2 or GDF15 protein as indicated. Data are relative to gene expression levels of macrophages incubated with CM from A+/D-NS (neutralizing abs) or A+/T-S (recombinant human protein) normalized to the average expression of housekeeping gene GAPDH. Unpaired t-test. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001; ns, no significance. Values are mean ± s.d. from 3 replicates.

Autophagy blockage promotes antitumor immune surveillance in vivo

Our in vitro experiments suggest a role of autophagy in immune surveillance via modulating tumor cell senescence. We tested the effect of autophagy inhibition on antitumor surveillance in a human GBM cell-derived xenograft mouse model, in which tumor xenografts were established by subcutaneous transplantation of WT or ATG3 KO U87MG cells and treated by TMZ (Figure 7A). In line with its chemotherapeutic effectiveness for GBM and consistent with Figure 1, TMZ halted GBM growth (Fig. S7A), as demonstrated by GBM cell depopulation in H&E staining (Figure 7B). This was accompanied by an increase of SA-GLB1, TP53 and CDKN1A positive cells, modestly in autophagy-competent tumors and more pronounced in autophagy-defective tumors. Importantly, recruitment of ADGRE1/F4/80⁺ tumor-associated macrophages (TAM) was observed in TMZ-treated GBM tissues.

Figure 7.

Autophagy blockage enhances immune surveillance of GBM tissues in vivo upon TMZ treatment. (A) Experimental layout describing the generation of xenograft tumors by subcutaneously injecting WT or ATG3^−/− U87MG cells in nude mice. An oral gavage of TMZ (5 mg/kg or 25 mg/kg) or vehicle control for 5 days was followed. (B) A series of staining for glioma sections. SA-GLB1 activity staining in WT or ATG3^−/− xenograft tumors dissected 9 days post-treatment with vehicle or TMZ for 5 days. IHC staining of human SQSTM1, TP53, CDKN1A and mouse ADGRE1 in WT or ATG3^−/− xenograft tumors dissected 14 or 17 days post treatment with vehicle or TMZ for 5 days. Quantification of SQSTM1, TP53, CDKN1A, ADGRE1-positive cells and SA-GLB1-positive area in each group is indicated in black. Scale bar: 20 μm. Representative experiments of n = 3. (C) Immunofluorescence of mouse CD68 and NOS2 in WT or ATG3^−/− xenograft tumors dissected 17 days post-treatment with vehicle or TMZ (25 mg/kg) for 5 days. Ratio of M1 macrophages (CD68⁺ NOS2⁺) to total macrophages (upper) or total cells (lower) in each group is indicated in white. Scale bar: 20 μm. Representative experiments of n = 3.

TAMs are defined as macrophages that infiltrate and populate in the microenvironment of solid tumors [31]. Although TAM can exhibit phenotypes anywhere between tumoricidal M1 type and pro-tumoral M2 type, they are conventionally acknowledged as M2-polarized given the high correlation with high tumor grade and poor prognosis in many cancers, including human GBM [34–36]. Our observed TAM accumulation is consistent with previous clinic report of GBM chemotherapy and radiotherapy [37]. Of note, within the TMZ-treated cohorts, increased number of CD68⁺ TAMs were detected in tumor tissues derived from autophagy-defective GBM cells (Figure 7C). However, what is the M1/M2 classification of these accumulated TAMs? Immunofluorescence analysis showed that TAMs were positive of MRC1/CD206, a marker for M2 macrophage phenotype (Fig. S7B). However, in TMZ-treated, autophagy-defective tumor tissues, there was a stronger co-localization of CD68 with NOS2/iNOS, a marker for M1 macrophage phenotype in mice (Figure 7C). Importantly, depletion of these macrophages by administration of clodronate liposomes attenuated the inhibition of tumor growth (Fig. S7C and S7D), suggesting that upon inhibition of autophagy and TMZ treatment, TAMs were reeducated to predominantly display antitumor M1 phenotype.

We also analyzed other innate immune cells in GBM tissues following TMZ treatment. Surprisingly, neutrophil recruitment and accumulation was dramatically reduced in TMZ-treated, autophagy-defective GBM tissues, in line with the observed decreased number of blood vessels (Fig. S7E). As tumor-associated neutrophils (TANs) often play a tumor-promoting role by supporting angiogenesis and expanding cancer stem cell pool [38–40], the absence of neutrophils in TMZ-treated, autophagy-defective GBM tissues further suggests the effect of the combination of TMZ with autophagy inhibition on potentiating antitumor immunity.

Discussion

In this study, we show that autophagy inhibition can potentiate chemotherapy-induced tumor cell senescence in cell culture and xenograft models of GBM. Importantly, the impact of autophagy inhibition is beyond a quantitative enhancement of overall senescence; but rather, autophagy inhibition also altered the SASP profile in a manner non-correlative with overall senescence enhancement, with certain SASP components even decreased upon autophagy blockage. Further, we discovered that the alteration of bioactive SASP profile by autophagy inhibition can reshape the tumor microenvironment (TME), thus leading to the inhibition of GBM cell proliferation (Figure 5A, Fig. S5A and S5B), induction of senescence of neighboring GBM cells through a paracrine mechanism (Figure 5C, Fig. S5C and S5D), polarization of macrophages to antitumor M1 state (Figures 6B and 7C, Fig. S6), and blockage of the recruitment of tumor-promoting neutrophils to tumor tissues (Fig. S7E). All these events contribute to the establishment of a tumor-suppressive milieu. Further, our results suggest that SASP components IL8, CXCL1, GDF15, and CCL2, were particularly important for such modulation of TME.

This study raises many interesting questions. Particularly, how does autophagy regulate chemotherapy-induced senescence mechanistically? Work from Narita’s team suggested a senescence-promoting role for autophagy based on the observation that autophagy was activated in senescent cells via feedback inhibition of PI3K-MTOR signaling, and abrogation of this feedback delayed senescence [41]. They further propose that in senescent cells, autophagosomes degrade organelles and proteins to generate amino acids and other metabolites, which are utilized by MTOR for biogenesis of secretory factors; these secreted components may initiate and maintain senescence by both paracrine and autocrine manner [42]. Another mechanistic explanation is that autophagy may specifically degrade certain senescence regulators such as LMNB1 and ADAR1, thus altering senescence. Indeed, ADAR1 degradation by autophagy has been shown to promote senescence through CDKN2A/p16^INK4a upregulation at the translational level [9,25,43]. On the contrary, mounting evidence, including our current study, demonstrates autophagy can also play an anti-senescence role. For example, SQSTM1-dependent autophagic degradation of GATA4 is suppressed during senescence and the stabilized GATA4 enhances senescence and NFKB-mediated SASPs [21]. Upon chemotherapy, it is likely that autophagy is activated in cancer cells to triage certain harmful cellular components generated under this condition (e.g., protein aggregates and damaged organelles), and thus mitigating the induction of senescence; conversely, autophagy inhibition exacerbates chemotherapy-induced senescence. However, this mechanism fails to explain how autophagy blockage alters the SASP profile of senescent GBM cells in such as complicated way that some specific SASP members are even downregulated. As there are multiple, functionally distinctive SASP regulators, would autophagy degrade them differentially in a context-dependent manner? Consistent with this possibility, TNIP1 and GATA4, two newly identified SASP regulators, have been shown to be regulated in different manner to initiate NFKB-mediated SASPs [21,44]; and in addition to NFKB, a cohort of other transcription factors such as STAT3, AP-1, and CEBPB/C/EBPβ, as well as their modulators, also dictate the spectrum and expression level of SASP components [20–22,45].

Another question concerns the role of autophagy in modulating immunity via senescence associated secretome. Tumor cell senescence may trigger an immune status that is either tumor suppressive or tumorigenic, as individual SASP components may regulate tumorigenesis differentially and some even in opposite direction [33,46]. In a pten^−/− prostate tumor senescence model, which is characterized by enrichment of immune-suppressive SASP such as CXCL1 and CXCL2, SASP recruits immune-suppressive myeloid cells; however, reprograming SASP via JAK-STAT3 or reeducating TAMs via CXCR2 blockade led to an anti-tumor environment [47–49]. In our study, we show that autophagy inhibition promotes chemotherapy-induced GBM cell senescence and results in a SASP profile eliciting antitumor immunity. However, does autophagy inhibition always result in a SASP profile that is tumor suppressive, or would it trigger a pro-tumor immunity under some other conditions? This question is obviously relevant to potential autophagy-targeted cancer therapy; and to answer this question, again, we first need to understand the mechanism by which autophagy differentially modulates the expression of individual SASP components.

Collectively, this study provides new insights into the potential of autophagy-targeted cancer therapy. Cancers with specific genetic background, especially those driven by oncogenic RAS, have been shown to be addictive to autophagy [50–52]. Indeed, autophagy inhibition has been reported to enhance the anticancer effect of multiple therapeutic agents, such as inhibitors of the RAS-MAPK pathway [5,6,53]. The efficacy of autophagy inhibition is attributed to the suppression of its pro-survival function, which can be activated by many therapeutic agents. Similarly, here we show that autophagy inhibition can result in more potent inhibition of GBM growth in combination with TMZ, the first-line drug for the treatment of this devastating malignancy. Importantly, we found that autophagy inhibition can also contribute to cancer treatment by potentiating tumor cell senescence, under certain clinically relevant conditions as investigated in this study. Cell autonomously, autophagy inhibition enhances GBM cell senescence to attenuate tumor growth. Non-cell autonomously, SASP, a consequence of GBM cell senescence, triggers paracrine senescence of neighboring tumor cells and elicit antitumor immunity by reeducating TAMs to M1 polarization and reducing neutrophil recruitment. Also relevant to antitumor immunity, an earlier study indicates that autophagy inhibition can render pancreatic cancer cells more susceptible to anticancer immunity, by preventing autophagic degradation of cancer cell MHC-1 (thus a cell-autonomous mechanism) [7]. Because of such multifaceted antitumor capability of autophagy inhibition, autophagy-targeted cancer therapy should be further pursued through developing potent and specific autophagy inhibitors, testing additional novel combination therapies (including combining with immune therapies), and identifying precise biomarkers that predict the responsiveness of individual patients to autophagy-targeted therapy, alone or in combination with other specific therapeutic regimens.

Materials and methods

Plasmid

For construction of tet-on shRNA-ULK1, an oligonucleotide targeting ULK1 sequence was inserted into the pTRIPZ lentiviral vector (Thermo Scientific Open Biosystems, RHS4750) according to the manufacturer’s protocol. For construction of LentiCRISPRV2-gRNA-ATG5, a gRNA oligonucleotide targeting human ATG5 was annealed and cloned into pLentiCRISPRV2 (Addgene, 52,961; deposited by Feng Zhang) according to the manufacturer’s protocol. For ATG3 gRNA expressing vector, we inserted gRNA oligo targeting ATG3 into PLX-sgRNA (Addgene, 50,662; deposited by Eric Lander, David Sabatini). All the oligonucleotides used for cloning are listed in Table 1.

Table 1.

Oligonucleotides list.

shRNA Sequence
ULK1	5’-CGCCCTTTGCGTTATATTGTAT −3’
sgRNA Sequence
ATG5	5’-TCAGGATGAGATAACTGAAA-3’
ATG3	5’-TTACCAGACTCCACGATTA-3’
Primers for Real-time PCR
Gene Name	Gene ID	Forward Primer (5’-3’)	Reverse Primer (5’-3’)
GAPDH	2597	TCCTCTGACTTCAACAGCGAC	TCTCTCTTCCTCTTGTGCTCTT
IL6	3569	TGCAATAACCACCCCTGACC	GTGCCCATGCTACATTTGCC
IL8-Pair#1	3576	TGTCTGGACCCCAAGGAAAAC	AAGTTTCACTGGCATCTTCACTG
IL8-Pair#2		AGAGAGCTCTGTCTGGACCC	TTTGCTTGAAGTTTCACTGGCAT
IL1A	3552	CTGGGAAACTCACGGCACTA	GTGAGACTCCAGACCTACGC
CXCL1-Pair#1	2919	CAAAACCTCGGAGTCCTGCT	GTCGTTCCAGTACCAACCGT
CXCL1-Pair#2		AACCGAAGTCATAGCCACAC	GTTGGATTTGTCACTGTTCAGC
CXCL1-Pair#4		AGCTTGCCTCAATCCTGCATCC	TCCTTCAGGAACAGCCACCAGT
CSF3/G-CSF	1440	TCCAGGAGAAGCTGGTGAGTGA	CGCTATGGAGTTGGCTCAAGCA
MIF	4282	AGAACCGCTCCTACAGCAAGC	GGAGTTGTTCCAGCCCACATTG
BSG/EMMPRIN	682	GGCTGTGAAGTCGTCAGAACA	ACCTGCTCTCGGAGCCGTTCA
SERPINE1	5054	CTCATCAGCCACTGGAAAGGCA	GACTCGTGAAGTCAGCCTGAAAC
IL1B	3553	CAGCTACGAATCTCCGACCAC	GGCAGGGAACCAGCATCTTC
PLAUR/uPAR-Pair#1	5329	CCACTCAGAGAAGACCAACAGG	GTAACGGCTTCGGGAATAGGTG
PLAUR/uPAR-Pair#2		CGGGCTCCAATGGTTTCCA	CAGAGTGAGCGTTCGTGAGTG
GDF15	9518	GACCCTCAGAGTTGCACTCC	GCCTGGTTAGCAGGTCCTC
CSF2	1437	TCCTGAACCTGAGTAGAGACAC	TGCTGCTTGTAGTGGCTGG
CSF1	1435	TGGCGAGCAGGAGTATCAC	AGGTCTCCATCTGACTGTCAAT
SPP1/OPN	6696	CTCCATTGACTCGAACGACTC	CAGGTCTGCGAAACTTCTTAGAT
PTX3	5806	AGGCTTGAGTCTTTTAGTGCC	ATGGATTCCTCTTTGTGCCATAG
CCL2-Pair#1	6347	AGAATCACCAGCAGCAAGTGTCC	TCCTGAACCCACTTCTGCTTGG
CCL2-Pair#2		GGCTGAGACTAACCCAGAAAG	GGGTAGAACTGTGGTTCAAGAG
CCL2-Pair#3		CAGCCAGATGCAATCAATGCC	TGGAATCCTGAACCCACTTCT
IL12	3593	AGTGTCAAAAGCAGCAGAGG	AACGCAGAATGTCAGGGAG
CXCL10	3627	GTGGCATTCAAGGAGTACCTC	TGATGGCCTTCGATTCTGGATT
CCR7-Pair#1	1236	CAGCCTTCCTGTGTGGTT	AGGAACCAGGCTTTAAAGT
CCR7-Pair#2		CAACATCACCAGTAGCACCTGTG	TGCGGAACTTGACGCCGATGAA
CCL22-Pair#1	6367	TCCTGGGTTCAAGCGATTCTCC	GTCAGGAGTTCAAGACCAGCCT
CCL22-Pair#2		ATTACGTCCGTTACCGTCTG	TAGGCTCTTCATTGGCTCAG
TNF-Pair#1	7124	TGGGATCATTGCCCTGTGAG	GGTGTCTGAAGGAGGGGGTA
TNF-Pair#2		CATTGCCCTGTGAGGAGGAC	CGACCCTAAGCCCCCAATTC
PPARG/PPAR γ	5468	TTCAGAAATGCCTTGCAGTG	CCAACAGCTTCTCCTTCTCG

Cell lines

U87MG (HTB-14), A172 (CRL-1620), and HEK293T (CRL-3216) cells were purchased from the American Type Culture Collection and grown in DMEM (MSKCC media core facility, New York, NY, USA) supplemented with 10% FBS, 100 U/ml penicillin-streptomycin, 2 mM L-glutamine and 1 mM sodium pyruvate (MSKCC media core facility New York, NY, USA). U937 cells were kindly provided by Dr. Ming Li (Memorial Sloan Kettering Cancer Center, New York, NY, USA) and grown in RPMI1640 (MSKCC media core facility, New York, NY, USA) supplemented with 10% FBS, 100 U/ml penicillin-streptomycin and 2 mM L-glutamine. Cells were maintained in a humidified incubator at 37°C with 5% CO₂ and regularly tested for mycoplasma infection.

Induction of senescence

Temozolomide (TMZ) was purchased from Sigma-Aldrich (T2577-100 MG) and dissolved in DMSO to yield 50 mM stock solution which was kept at −20°C until use. To induce senescence, A172 and U87MG cells were incubated with TMZ in culture media containing FBS (10% FBS for A172 and 2% FBS for U87MG) for 4–12 days as indicated.

CRISPR-Cas9-mediated gene knockout

To generate constitutive ATG3 knockout U87MG cells, cells were transfected with a mixture of two plasmids, encoding Cas9 (Addgene, 52,962; deposited by Feng Zhang) and gRNA-ATG3 (or a non-target gRNA in a case of control clones) by electroporation (Amaxa® Cell Line Nucleofector® Kit T, Amaxa Nucleofector II device). Media was changed 6 h post-transfection. Forty-eight hours after transfection, cells were trypsinized and seeded on 100-mm² plates at a density of 150 cells/plate. Approximately 2 weeks later, single clonal populations were isolated using sterile cloning discs (SP BEL-ART, F37847-0001).

To construct ATG5 knockout A172 cells, cells in 35-mm² culture dishes were infected with lentiviruses containing pLentiCRISPRv2-gRNA-ATG5 or pLentiCRISPR v2-sgNT (Addgene, 138,681; deposited by Reuben Shaw) in the presence of polybrene (Millipore Sigma, TR-1003-G) at a concentration of 10 μg/ml. Cells were given fresh media 12 h post-infection and were placed under puromycin selection at a concentration of 1 μg/ml 48 h post-infection for 3 days. Single colonies derived from single cell were expanded as described above.

Lentiviruses were produced by co-transfection of LentiCRISPRv2-gRNA-ATG5 or LentiCRISPRv2-gRNA-sgNT with psPAX2 (Addgene, 12,260; deposited by Didier Trono) and VSV-G (Addgene, 8454; deposited by Bob Weinberg) in HEK293T cells in 60 mm² culture plates using Lipofectamine 3000 reagent (Invitrogen, L3000015). Media was changed 6 h post-transfection and refreshed with 3 ml fresh media 24 h post-transfection. Media containing lentivirus was collected 36 h and 48 h post-transfection. This media was passed through a 0.45-μm filter to remove cell debris and stored at −80°C until use.

For doxycycline-inducible ATG3 knockout U87MG cells, U87MG-tet on-Cas9 cells showing a high Cas9 expression with doxycycline and the least basal Cas9 leakage without doxycycline were infected with lentivirus expressing gRNA-ATG3 (Sigma-Aldrich, HS5000004183 and HS5000004184). Cells were given fresh media 12 h post-infection and placed under puromycin selection 48 h post-infection for 3 days. Clonal populations were expanding and verified as described above and maintained in DMEM supplemented with tetracycline-negative FBS to avoid basal Cas9 leakage.

All the constructed autophagy-deficient GBM cells described above, along with their corresponding parental cells were treated with torin 1 (1 μM; Cedarlane Labs, 10,997–10) for 6 h, bafilomycin A₁ (100 ng/ml; Selleck Chemicals, S1413) for 2 h, or a combination of torin 1 for 6 h with bafilomycin A₁ for the last 2 h. Immunoblots of SQSTM1, ATG3, ATG5 or LC3 were conducted to evaluate autophagy flux.

SA-GLB1 staining

SA-GLB1 staining was performed at pH 6.0. Cells were seeded in 6-well plates at a density of 2 × 10⁴ cells/ml and exposed to TMZ to induce senescence as described above. Cells were washed three times with PBS (MSKCC media core facility New York, NY, USA), fixed, and stained for SA-GLB1 activity according to the manufacturer’s instruction (Cell Signaling Technology, 9860).

For cryosections of xenograft tumors, SA-GLB1 staining was completed as previously described [18,54]. In brief, freshly frozen sections were fixed with 0.5% glutaraldehyde in PBS for 15 min, washed with PBS supplemented with 1 mM MgCl₂, and stained overnight in freshly prepared staining solution containing 40 mM citric acid monohydrate (Millipore Sigma, 5949–29-1), 40 mM sodium dihydrogen phosphate monohydrate (Millipore Sigma, 10,049–21-5) buffer at pH 6.0, 5 mM each of potassium hexacyanoferrate (II) trihydrate (Millipore Sigma, 14,459–95-1) and potassium hexacyanoferrate (III) (Millipore Sigma, 13,746–66-2), 150 mM sodium chloride (Merck, 7647–14-5), 2 mM magnesium chloride hexahydrate (Millipore Sigma, 7791–18-6), and 1 mg/ml X-gal (Millipore Sigma, 7240–90-6). 3–5 high-power fields per well/section were counted to quantify the percentage of SA-GLB1⁺ cells.

C₁₂FDG based flow cytometry

Fluorescent detection of SA-GLB1 activity was conducted according to a protocol described in [55]. In brief, Cells were seeded in 24-well plate at a density of 2 × 10⁴ cells/ml and exposed to TMZ or DMSO to induce senescence. Lysosomal alkalinization (~pH 6) was induced by 1-h pretreatment with 100 nM bafilomycin A₁ (Selleck Chemicals, S1413) and C₁₂FDG (Invitrogen, D2893), a GLB1 substrate that fluoresces after cleavage, was added to the media at a final concentration of 33 μM for 1 h. Stained cells were washed twice with ice-cold PBS, trypsinized, and spun down at 500 x g for 5 min. Each sample was resuspended in 200 μl PBS containing DAPI for flow cytometry analysis within 1 h, detecting FITC-positive cells.

Isolation of senescent cells and collection of conditioned media

Fluorescence-activated cell sorting (FACS) based on C₁₂FDG staining was employed to isolate senescent cells. WT and autophagy-deficient GBM cells were seeded in 150-mm² culture dishes at a density of 2 × 10⁴ cells/ml and exposed to TMZ or DMSO for 3 days (A172) or 9–12 days (U87MG). FACS analysis was performed by the Memorial Sloan Kettering Flow Cytometry Core Facility. Samples overlapping the control were defined as C₁₂FDG^lo, and samples with no or partial overlap with the control were defined as C₁₂FDG^hi, and considered potentially senescent cells. By this means, we eventually obtained 6 subpopulations. Cell pellets were then prepared for western blot or real-time PCR.

To collect conditioned media, sorted cells were spun down at 500 x g for 5 min and resuspended in complete culture media for cell counting. Cells were adjusted to 1–2 × 10⁵ cells/ml to seed in 6-well plates and changed to 2 ml fresh basal/complete culture media the next day to remove floating dead cells. After 48 h, cells were trypsinized and counted, conditioned media was collected, clarified by centrifugation and the volume used was normalized to cell number.

Real-time PCR

Total RNA was extracted using the TRIzol reagent (Invitrogen, 15,596,026). Briefly, 20% chloroform was added to each sample and mixed well by shaking for 30s before 3-min incubation. A colorless upper aqueous phase was obtained after 15-min spin at 13,000 x g, at 4°C and transferred to a new tube and mixed with an equal volume of isopropanol. Tubes were incubated for 10 min at room temperature, followed by centrifugation for 15 min at 13,000 x g, at 4°C pellet RNA. RNA was washed in 75% ethanol, air dried and resuspended in nuclease-free water (Ambion, AM9932). Complementary DNA (cDNA) was obtained using iScript Reverse Transcription Supermix (Bio-Rad, 1,708,891). Real-time PCR were performed in triplicates using the iQ SYBR Green supermix (Bio-Rad, 1,708,882) on the CFX Connect Real-Time PCR Detection System (Bio-Rad). The PCR program was as follows: 95°C, 30s; 40 cycles (for each cycle 95°C, 15s; 55°C, 40s). Primers are listed in Table 1. GAPDH served as the endogenous normalization control for human samples.

Western blot

Cells were lysed in RIPA buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% NP-40 [Millipore Sigma, 74,385], 0.5% sodium deoxycholate [Thermo Fisher Scientific, 89,905] and 0.1% SDS [Millipore Sigma, 151–21-3]) supplemented with phosphatase inhibitors (Millipore Sigma, P0044-5 ml and P5726-5 ml) and protease inhibitors (Calbiochem, 103,476–89-7, 26,305–03-3 and 9087–70-1) and protein concentration was determined by Bradford assay (Bio-Rad, 5,000,006). Cell lysates were resolved on SDS-PAGE gels and transferred to nitrocellulose membranes. Membranes were blocked in 5% nonfat milk + TBST (diluted 20x TBS [Thermo Fisher Scientific, J75892-K8] with ddH₂O and supplemented with 0.1% TWEEN 20 [Millipore Sigma, P2287-4 L]) for 1 h at room temperature, incubated overnight with primary antibodies. Membranes were then washed three times and incubated with secondary antibodies conjugated to horseradish peroxidase (goat anti-mouse [Thermo Fisher Scientific, 31,430]; donkey anti-rabbit [Thermo Fisher Scientific, 31,458]) for 1 h. After three final washes in TBST, membranes were visualized by chemiluminescence using Clarity Western ECL Substrate (Bio-Rad, 170–5061) on an Amersham Imager 600 (GE Healthcare Life Sciences).

Human cytokine array

Conditioned media (serum-free DMEM) from four sorted U87MG subpopulations: A+/D-C₁₂FDG^lo, A-/D-C₁₂FDG^lo, A+/T-C₁₂FDG^hi and A-/T-C₁₂FDG^hi, also known as A+/D-NS, A-/D-NS, A+/T-S and A-/T-S (A+: autophagy intact; A-: autophagy deficient; D: DMSO; T: TMZ) were normalized based on cell number by diluting with basal DMEM. Aliquots (200 μl) of CM were analyzed using a commercial Proteome Profiler Human XL Cytokine Array Kit (R&D system, ARY022B).

In vitro cell growth assay

Filtered conditioned media (complete culture media) collected as described above was applied to U87MG or A172 cells labeled with H2B-mCherry to measure GBM cell growth. In brief, 1,500 cells labeled with H2B-mCherry were plated in 96-well plates and incubated in triplicate with a mixture of CM and fresh complete media at a ratio of 2:1 one day later. H2B-mCherry-positive nuclei were counted every 8 h using a Cytation 5 imager (BioTek). Cells cultured in complete media served as a control.

To visualize growth of U87MG cells exposed to CM, U87MG cells were plated in 24-well plates at a density of 15,000 cells/ml and incubated with CM as described above. After a 9-day incubation, cells were washed in PBS, fixed in 95% methanol, and incubated for 30 min in 0.2% crystal violet with 2% methanol. Crystal violet was removed, and wells were washed in tap water until the solution ran clear and allowed to dry for imaging.

U937-derived macrophages

U937 monocytic cells were differentiated into macrophages by incubation with 200 ng/ml phorbol 12-myristate 13-acetate (PMA; Millipore Sigma, P8139) for 48 h in 24-well plates at a density of 4 × 10⁵ cells/ml.

Treatment by recombinant human proteins and neutralizing antibody studies

Conditioned media from sorted subpopulations was supplemented with recombinant human CCL2/MCP1 protein (R&D system, 279-MC-010/CF), recombinant human GDF15 protein (R&D system, 9279-GD-050) or preincubating for 1 h with neutralizing antibodies against human CXCL8/IL8 (R&D system, MAB208-100), human/primate CXCL1/GRO alpha/KC/CINC-1 (R&D system, MAB275-100), human/primate VEGF (R&D system, MAB293-100) or mouse IgG1 isotype control (R&D system, MAB002) before adding to cells. Cell growth, macrophage polarization and paracrine senescence assays were conducted afterward.

In vivo xenograft mouse model

Athymic nude mice (nu/nu, 5–6 weeks old) were obtained from Charles River Laboratory for human tumor xenograft studies. All protocols for animal experiments were approved by the MSKCC Institutional Animal Care and Use Committee.

Xenografts with tet-on system

Nude mice were subcutaneously transplanted on the right flank with 5 × 10⁶ ULK1 iKD or ATG3 iKO U87MG cells in 0.1 ml (50 μl PBS + 50 μl Matrigel [Thermo Fisher Scientific, CB-40234]). When tumors reached a size of 100 ~ 300 mm³, mice were randomly assigned into two groups. One group was fed a diet containing doxycycline (625 mg/kg; Envigo Tekald, TD.08541), while the other was provided Picolab Rodent Diet 5053 (LabDiet, Purina; 5053). After 7 days, mice were assigned into three subgroups for daily oral gavage with vehicle or TMZ (5 mg/kg and 25 mg/kg) for 5 days. The gavage was provided using a 1-ml syringe and an 18-gauge ball-tipped animal feeding needle (Cadence Science, 9923). Tumor growth was monitored every other day by calipers and tumor volume was estimated using the formula: volume = (width² x length)/2. Mice transplanted with ULK1 iKD U87MG cells were euthanized 23 days post-gavage and dissected tumor xenografts were fixed in 10% formalin for paraffin section, immunohistochemistry. Mice transplanted with ATG3 iKO U87MG cells were euthanized 9 days post-gavage, tumor xenografts were dissected and immersed into OCT for cryosectioning and staining for GLB1 activity.

Xenografts without tet-on system

Nude mice were subcutaneously transplanted on the right flank with either control or ATG3 KO U87MG cells as described above. When tumors reached a size of 100 ~ 400 mm³, mice were divided into three subgroups for daily oral gavage with vehicle or TMZ (5 mg/kg and 25 mg/kg) for 5 consecutive days followed by 13 ~ 17 days off treatment. Tumor growth was monitored every other day and tumor xenografts were dissected for frozen and paraffin sections for IHC and IF staining. For macrophage depletion mouse model, nude mice were subcutaneously injected with 6 × 10⁶ ATG3 KO U87MG cells on the right flank. When tumors reached a size of 100 ~ 400 mm³, all xenografted mice were daily treated with TMZ (5 mg/kg) for 5 consecutive days. Macrophage depletion was achieved by intravenous injection with clodronate liposomes or control liposomes (Liposoma, CP-020-020) at a dosage of 100 µL/10 grams of animal weight one day before the TMZ treatment. The administration of clodronate liposomes was given every other day for 3 weeks. Tumor growth was monitored every day.

Immunohistochemistry and immunofluorescence

Dissected tumor xenografts were fixed overnight in 4% paraformaldehyde, embedded in paraffin, and cut into 5-μm thick sections for hematoxylin and eosin staining, and immunohistochemical or immunofluorescence staining according to standard protocols. Primary antibodies used for staining are listed in Table 2 and anti-rabbit (R&D system, CTS005), anti-rat (R&D system, CTS107) or anti-mouse HRP-DAB system (R&D system, CTS002) were applied for 1 h and visualized with DAB, or fluorophore-conjugated secondary antibodies (Life Technology, A11007; Santa Cruz Biotechnology, sc-2024; Molecular Probes, A-21206) were applied for 1 h at room temperature. Fluorescent antibody-labeled slides were mounted with Prolong Gold Antifade reagent with DAPI (Invitrogen, P36935) after counterstaining with DAPI. From 3 ~ 7 tumors from each group were selected; at least 3 sections from each tumor have been entirely imaged, staining positive area and cells were analyzed and quantified by using ImageJ in blinded fashion.

Table 2.

Antibody list.

Antibody Name	Vendor	Catalog No.	Ratio
Anti-TP53/p53(DO-1)	Santa Cruz Biotechnology	sc-126	1:1000 for WB
			1:200 for IHC
Anti- TP53/p53 (7F5)	Cell Signaling Technology	2527S	1:1000 for WB
Anti-CDKN1A/p21	Abcam	ab109520	1:200 for IHC
Anti- CDKN1A/p21	Cell Signaling Technology	2947S	1:1000 for WB
			1:200 for IHC
Anti-γ-H2AX	Cell Signaling Technology	9718	1:200 for IHC
Anti-GLB1/beta-galactosidase	Abcam	ab203749	1:200 for IHC
Anti-LMNB1/lamin B1(D4Q4Z)	Cell Signaling Technology	12586S	1:1000 for WB
Anti-ULK1	Sigma-Aldrich	A7481	1:200 for IHC
Anti-ULK1(D9D7)	Cell Signaling Technology	6439S	1:1000 for WB
Anti-ATG5	Sigma-Aldrich	A0731	1:1000 for WB
Anti-ATG3	Cell Signaling Technology	3415S	1:1000 for WB
Anti-human SQSTM1/p62	R&D	MAB80281-100	1:1000 for WB
			1:250 for IHC
Anti-human LC3	Sigma-Aldrich	L7543	1:1000 for WB
Anti-GAPDH (14C10)	Cell Signaling Technology	2118S	1:1000 for WB
Anti-Cas9 (7A9-3A3)	Cell Signaling Technology	1469s	1:1000 for WB
Anti-PECAM1/CD31	Novus	AF3628-SP	1:200 for IHC
Anti-mouse MRC1/MMR/CD206	R&D	AF2535	1:200 for IF
Anti-ADGRE1/F4/80	Bio-Rad	MCA497GA	1:200 for IHC
Anti-CD68	Bio-Rad	MCA1957	1:200 for IF
Anti-NOS2/iNOS	Novus	NB300-605	1:200 for IF
Anti-ELANE/neutrophil elastase	Cell Signaling Technology	90120S	1:200 for IHC

Statistical analysis

Data were analyzed by using GraphPad Prism 8 Software. Statistical significance was evaluated using either Student’s t-test with two tails, and an assumption of equal variance or two-way ANOVA.

Funding Statement

This work was supported by the Office of Extramural Research, National Institutes of Health [R01CA204232, R01CA258622, R01CA244581]; Office of Extramural Research, National Institutes of Health [F31CA247112].

Abbreviations

ADGRE1/F4/80: adhesion G protein-coupled receptor E1; BSG/EMMPRIN: basigin (Ok blood group); CM: conditioned media; CXCL: C-X-C motif chemokine ligand; CSF: colony stimulating factor; CCL2: C-C motif chemokine ligand 2; CCL22: C-C motif chemokine ligand 22; CCR7: C-C motif chemokine receptor 7; FGF19: fibroblast growth factor 19; FACS: fluorescence-activated cell sorting; GDF15: growth differentiation factor 15; GBM: glioblastoma multiform; IL6: interleukin 6; IL8: interleukin 8; IHC: immunohistochemistry staining; MRC1: mannose receptor C-type 1; NOS2: nitric oxide synthase 2; MIF: macrophage migration inhibitory factor; Non-Sen: non-senescent cells; PECAM1/CD31: platelet and endothelial cell adhesion molecule 1; PPARG: peroxisome proliferator activated receptor gamma; PTX3: pentraxin 3; PLAUR/uPAR: plasminogen activator, urokinase receptor; RB1: RB transcriptional corepressor 1; Sen: senescent cells; SASP: senescence-associated secretory phenotype; SERPINE1: serpin family E member 1; SPP1/OPN: secreted phosphoprotein 1; TMZ: temozolomide; TAM: tumor-associated macrophages; TME: tumor microenvironment; TNF: tumor necrosis factor; VEGF: vascular endothelial growth factor.

Disclosure statement

A.M.M. and X.J. are inventors of patents related to autophagy and cell death. X.J. is an consultant and equity holder of Exarta Therapeutics and Lime Therapeutics.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2022.2155794