The Autophagy Machinery Controls Cell Death Switching between Apoptosis and Necroptosis (original) (raw)

Dev Cell. Author manuscript; available in PMC 2017 May 23.

Published in final edited form as:

PMCID: PMC4886731

NIHMSID: NIHMS787344

Megan L. Goodall,1 Brent Fitzwalter,1 Shadi Zahedi,2 Min Wu,1,3 Diego Rodriguez,4 Jean M. Mulcahy-Levy,2 Douglas R. Green,4 Michael Morgan,1 Scott D. Cramer,1,5 and Andrew Thorburn1,5

Megan L. Goodall

1Department of Pharmacology, University of Colorado Denver, Aurora, CO 80045, USA

Brent Fitzwalter

1Department of Pharmacology, University of Colorado Denver, Aurora, CO 80045, USA

Shadi Zahedi

2Department of Pediatrics, University of Colorado Denver, Aurora, CO 80045, USA

Min Wu

1Department of Pharmacology, University of Colorado Denver, Aurora, CO 80045, USA

3Department of Molecular and Cellular Oncology, MD Anderson Cancer Center, Houston TX 77030, USA

Diego Rodriguez

4Department of Immunology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA

Jean M. Mulcahy-Levy

2Department of Pediatrics, University of Colorado Denver, Aurora, CO 80045, USA

Douglas R. Green

4Department of Immunology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA

Michael Morgan

1Department of Pharmacology, University of Colorado Denver, Aurora, CO 80045, USA

Scott D. Cramer

1Department of Pharmacology, University of Colorado Denver, Aurora, CO 80045, USA

Andrew Thorburn

1Department of Pharmacology, University of Colorado Denver, Aurora, CO 80045, USA

1Department of Pharmacology, University of Colorado Denver, Aurora, CO 80045, USA

2Department of Pediatrics, University of Colorado Denver, Aurora, CO 80045, USA

3Department of Molecular and Cellular Oncology, MD Anderson Cancer Center, Houston TX 77030, USA

4Department of Immunology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA

5SDC and AT share senior authorship.

Summary

Although autophagy controls cell death and survival, underlying mechanisms are poorly understood and it is unknown if autophagy just affects whether or not cells die or also controls other aspects of programmed cell death. MAP3K7 is a tumor suppressor gene associated with poor disease-free survival in prostate cancer. Here we report that Map3k7 deletion in mouse prostate cells sensitizes to cell death by TNF-related apoptosis inducing ligand (TRAIL). Surprisingly, this death occurs primarily through necroptosis, not apoptosis, due to assembly of the necrosome in association with the autophagy machinery, mediated by p62/SQSTM1 recruitment of RIPK1. The mechanism of cell death switches to apoptosis if p62-dependent recruitment of the necrosome to the autophagy machinery is blocked. These data show that the autophagy machinery can control the mechanism of programmed cell death by serving as a scaffold rather than by degrading cargo.

Graphical abstract

An external file that holds a picture, illustration, etc. Object name is nihms787344u1.jpg

Introduction

Macroautophagy (autophagy) maintains cellular homeostasis by catabolizing organelles and proteins to generate nutrients and macromolecular precursors (Kroemer et al., 2010). Autophagy occurs at a basal level in all cells and is up-regulated in response to diverse stressors (e.g. starvation, oxidative stress, drug treatment, ER stress) playing a critical role in development and disease (Green and Levine, 2014; Levine and Kroemer, 2008; Mizushima and Levine, 2010). Autophagy can both promote and inhibit cell death under different cellular contexts, and several mechanistic links between autophagy and apoptosis have been elucidated (Fitzwalter and Thorburn, 2015; Rubinstein and Kimchi, 2012). For example, autophagy promotes apoptosis by Fas Ligand/ CD95 because of its ability to degrade a negative regulator of CD95 signaling (Gump et al., 2014) but it can protect against Tumor Necrosis Factor-Related Apoptosis Inducing Ligand (TRAIL)-induced apoptosis by controlling the levels of a pro-apoptotic member of the BCL family (Thorburn et al., 2014). During developmental cell death, similar mechanisms whereby components of the apoptosis machinery are degraded by autophagy have also been identified (Nezis et al., 2010). Very little is known about how autophagy regulates other forms of programmed cell death (Galluzzi et al., 2015), such as necroptosis.

Necroptosis is best understood in response to Tumor Necrosis Factor (TNFα) and requires a cytosolic complex, known as the necrosome that is formed by the serine/threonine receptor interacting protein 3 (RIPK3) in complex with RIPK1, FADD, and caspase-8 (Han et al., 2011; Vandenabeele et al., 2010). Mixed lineage kinase domain-like protein (MLKL) is recruited to the necrosome and phosphorylated MLKL mediates plasma membrane lysis to induce necroptosis (Cai et al., 2014; Sun et al., 2012; Zhao et al., 2012). TNFα can also stimulate other secondary complexes to activate NFκB or, via the death-inducing signaling complex (DISC), promote apoptosis. All of these complexes can involve RIPK1, and the balance of activities within them is believed to control caspase-dependent and caspase-independent cell death (Arslan and Scheidereit, 2011; Fuchs and Steller, 2015). For instance, repression of the necroptotic pathway by apoptotic regulators, such as FADD and caspase-8, is essential for proper mammalian development (Kaiser et al., 2011; Oberst et al., 2011; Zhang et al., 2011). The importance of this balance of different modes of programmed cell death was elegantly shown by the finding that genetic ablation of Ripk1 in mice causes postnatal lethality that is only rescued with loss of both Ripk3 and either caspase-8 or Fadd (Dillon et al., 2014). This is likely due to the fact that RIPK1, which directly regulates caspase-8 activity in some circumstances (Bertrand et al., 2008; Dondelinger et al., 2013; Morgan et al., 2009; Wang et al., 2008), has also been shown to both positively and negatively regulate RIPK3 oligomerization and necroptosis (Dannappel et al., 2014; Orozco et al., 2014).

Necroptosis is associated with inflammatory disease (Linkermann and Green, 2014; Pasparakis and Vandenabeele, 2015) and is important in the response to bacterial and viral infection (Cho et al., 2009). For instance, mice with deletions in Ripk3 or Mlkl are protected from inflammatory pancreatitis (He et al., 2009; Wu et al., 2013). A role for necroptosis in cancer is suggested because expression of RIPK3 is commonly silenced in cancers making most cancer cells unable to undergo necroptosis even though they are still capable of activating apoptosis (Koo et al., 2015). This suggests that necroptosis may be specifically selected against during tumor evolution, perhaps because factors that activate adaptive anti-tumor immunity are preferentially released by induction of necroptosis rather than apoptosis of tumor cells (Yatim et al., 2015).

MAP3K7 (also known as TGF-β-activated kinase 1, TAK1) is a serine/threonine protein kinase responsible for activating NF-κB signaling and mitogen-activated protein kinases downstream of death receptors. MAP3K7 is recruited to death receptor complexes through its interaction with RIPK1. Loss of MAP3K7 leads to hypersensitivity to cell death in response to TNFα (Arslan and Scheidereit, 2011; Dondelinger et al., 2013; Lamothe et al., 2013; Morioka et al., 2014; Vanlangenakker et al., 2011) and TRAIL (Choo et al., 2006; Lluis et al., 2010; Morioka et al., 2009) but the underlying mechanisms are incompletely understood. Interestingly, deletion of the MAP3K7 gene occurs in 30–40% of prostate tumors (both hemizygous and homozygous deletions are observed). Prostate cancer is the most commonly diagnosed cancer in the United States in men, and the second leading cause of male cancer related deaths affecting approximately 10% of patients diagnosed (Siegel et al., 2014). Deletion or low mRNA expression of MAP3K7 defines an aggressive subtype of prostate cancer with poor disease-free survival (Liu et al., 2007; Rodrigues et al., 2015; Wu et al., 2012). Thus, finding a way to kill tumor cells that have lost expression of MAP3K7 might improve treatment of a lethal subset of prostate cancer. Consistent with previous reports, we show here that a mouse cell model of Map3k7 null tumors is hypersensitive to cell death in response to TRAIL. Surprisingly however, although TRAIL normally kills by caspase-dependent apoptosis (Johnstone et al., 2008), prostate cells lacking MAP3K7 are capable of switching between apoptosis and necroptosis in response to TRAIL treatment and we found that TRAIL-induced necroptosis depends on the ability of RIPK1 to interact with p62/SQSTM1 and associate with the autophagosome machinery. These data define a mechanism by which the autophagy machinery controls switching between different types of programmed cell death in a regulated manner by serving as a scaffold for efficient formation and activation of the necrosome.

Results

Loss of Map3k7 sensitizes to TRAIL-induced death

Loss of MAP3K7 increases sensitivity to TRAIL in human cancer cells (Morioka et al., 2009). To characterize this sensitivity in an isogenic system that avoids complications with human cancer cells, which are often selectively resistant to necroptosis (Koo et al., 2015), matched mouse prostate epithelial cells (MPECs) with either intact floxed or Cre-deleted Map3k7 (Wu et al., 2012) were tested for their response to chemotherapeutics commonly used in prostate cancer (docetaxol, paclitaxel, and carboplatin) in addition to TNFα and TRAIL (Fig 1A). Loss of Map3k7 increased cell death by TNFα and TRAIL, but not to the other drugs. To assess the clinical relevance of TRAIL and TNFα sensitivity to MAP3K7 loss in human prostate cancers, publicly available prostate cancer CNA datasets were analyzed (Baca et al., 2013; Cerami et al., 2012; Gao et al., 2013; Taylor et al., 2010) (The results shown here are in part based upon data generated by the TCGA Research Network: http://cancergenome.nih.gov/). MAP3K7 (red bars) was homozygously co-deleted with TRAIL receptors (TNFRSF10A and B, blue and green bars respectively) in 5% of prostate cancers, but not with other death receptors such as TNFRSF1A and B or FAS/CD95 (Fig 1B–D). Furthermore, homozygous loss of TRAIL receptors occurs at the highest frequency in prostate cancers (Figure S1A). The loss of TRAIL receptors in conjunction with a specific gene, MAP3K7, whose loss selectively sensitizes cells to TRAIL suggests that prostate tumors where MAP3K7 is lost are under selective pressure to circumvent TRAIL induced cell death. Therefore, we focused on TRAIL rather than TNFα for further experiments.

An external file that holds a picture, illustration, etc. Object name is nihms787344f1.jpg

MAP3K7 loss correlates with TRAIL receptor loss, but not other death receptor loss in prostate cancer

A) Cell viability of MAP3K7 floxed (white bars) and null (black bars) MPECs treated with chemotherapeutics for 24 hours. Data are represented as mean ± SD. B) Frequency of homozygous deletions of MAP3K7, TNFRSF10A, TNFRSF10B, TNFRSF1A, TNFRSF1B, and FAS in three prostate cancer datasets. C) Co-occurance and mutual exclusivity between MAP3K7, TNFRSF10A, TNFRSF10B, TNFRSF1A, TNFRSF1B, and FAS. D) Oncoprint of MAP3K7, TNFRSF10A, TNFRSF10B, TNFRSF1A, TNFRSF1B, and FAS (TCGA, n=333). Each bar represents a patient sample with 150 samples shown; the additional 183 had no changes to these genes. See also Figure S1.

TRAIL-induced cell death involves RIPK1-dependent necroptosis and apoptosis

TRAIL typically kills cells by canonical apoptosis that can be completely prevented by caspase inhibition. However, in _Map3k7_-null MPECs, the pan-caspase inhibitor Q-VD-OPh (QVD) failed to rescue cell viability after TRAIL treatment (Figure 2A–C, S2A). TRAIL-induced death was however blocked by the RIPK1 kinase inhibitor necrostatin (Fig 2A–C) or by knockdown of Ripk1 (Fig 2C, S2E). TRAIL-induced necroptosis was suggested in Map3k7 null cells by fast, temporally coincident annexin V and propidium iodide (PI) staining assessed by flow cytometry and live cell imaging on an IncuCyte with coincident staining starting about 4 hours after TRAIL treatment (Fig 2B, S2B–C) and little caspase 3/7 activity (Figure S2D). In contrast, staurosporine, which induces apoptosis, demonstrated annexin V positivity first (apoptosis), but little dual PI positivity (necroptosis) at early time points (Figure 2B, S2B, grey circles). Consistent with TRAIL-induced death being primarily by necroptosis, QVD did not significantly alter the amount of annexin V and PI co-staining, but, in the presence of necrostatin-1, the few cells that did die were largely annexin V positive and PI negative suggestive of apoptosis (Figure S2B). Knockdown of Ripk3 or Mlkl by shRNA as well as the RIPK3 inhibitor GSK’872 were all unable to rescue cell viability on their own, however each of these necroptosis-specific inhibitions did rescue cell viability when combined with QVD (Fig 2D–F, S2E). These data indicate that in Map3k7 null cells, TRAIL causes switching between necroptosis and apoptosis, both involving RIPK1. Necroptosis is the predominant mode of death in the absence of intervention, but the mechanism of cell death switches to apoptosis if the necrosome is targeted by removing MLKL or by removing or inhibiting RIPK3.

An external file that holds a picture, illustration, etc. Object name is nihms787344f2.jpg

Necroptotic cell death is Ripk1 dependent

A) Cell viability of Map3k7 floxed (white bars) and null (black bars) MPECs treated with TRAIL, QVD, and necrostatin-1 for 24 hours. B) Map3k7 null MPECs treated with TRAIL, Staurosporine, QVD, and necrostatin-1 and imaged in the presence of annexin V and PI on an IncuCyte in 4 fields of view every 4 hours for 24 hours with representative images at 4 hours with scale bars representing 30 μm. Co-localization of annexin V and PI determined using CellProfiler and Pearson correlation coefficient (r) determined. See also Figure S2. C) Cell viability of Map3k7 null MPECs treated with TRAIL, QVD, and necrostatin-1 for 24 hours with Ripk1 shRNA (black bars). Data are represented as mean ± SD. D) Cell viability of Map3k7 null MPECs treated with TRAIL, QVD, GSK’872, and necrostatin-1 for 24 hours. Data are represented as mean ± SD. E) Cell viability of Map3k7 null MPECs treated with TRAIL, QVD, and necrostatin-1 for 24 hours with Ripk3 shRNA (black bars). Data are represented as mean ± SD. F) Cell viability of Map3k7 null MPECs treated with TRAIL, QVD, and necrostatin-1 for 24 hours with Mlkl shRNA (black bars). Data are represented as mean ± SD.

Necroptosis depends on the autophagy machinery

Since we previously found that TRAIL-induced apoptosis is inhibited by autophagy (Thorburn et al., 2014), we tested how autophagy affects TRAIL induced death in the _Map3k7_-null cells. Basal and induced autophagic flux was functional in the floxed and null Map3k7 cell lines as shown by immunoblotting for LC3-II accumulation after inhibition of the lysosome and by quantitative flow cytometry (Gump et al., 2014; Gump and Thorburn, 2014) of cells expressing mCherry-GFP-LC3-II (Fig 2F). Inhibition of autophagosome fusion with the lysosome using either chloroquine (CQ) or Bafilomycin A1 (Baf) caused more TRAIL-induced death consistent with a protective effect of basal autophagy (Fig 3A). However, inhibition of early/mid-stages of autophagy with the PI3 kinase inhibitor Wortmannin or knockdown of Atg5, Atg7, or Becn1 led to increased cell viability (Fig 3A, S3 A–B). Rescue in cell viability was not observed with knockdown of Ulk1 or FIP200 (Fig S4B). However, neither of these gene products is essential for basal autophagy and FIP200 does not change recruitment of components of the autophagy machinery to earlier structures (Itakura and Mizushima, 2011). No rescue in cell viability was observed with knockdown of Rubicon, suggesting that autophagy rather than LC3-Associated Phagocytosis is responsible for these effects (Martinez et al., 2015) (Fig S3B). These data suggest that it is not the autophagy-mediated degradation of cargo that is important in determining the response to TRAIL treatment, because if this were the case one would expect a similar effect on cell death regardless of which stage of autophagy was inhibited. An alternate hypothesis to explain these results is that the autophagy machinery provides a scaffold for necroptosis signaling. To test this, we looked to see if autophagy proteins associated with necrosome proteins after TRAIL treatment (Fig 3B, S3C–D). As expected, the necrosome proteins RIPK3 and MLKL were co-immunoprecipitated with RIPK1, with more association in the Map3k7 null cells after treatment with TRAIL (Fig 3B, S3C–D). Proteins involved in early/mid-stage of autophagy including ATG5/ATG12, ATG7, and p62/SQSTM1 also co-immunoprecipitated with RIPK1 while proteins that are important for earlier signaling in the autophagy process (BECN-1 and ULK1) and those involved in later steps when autophagosomes fuse with lysosomes (STX17 and LAMP) did not (Fig 3B). In situ complex formation of the necrosome together with autophagy proteins was demonstrated by Proximity Ligation Assays (PLA) (Fig 3C–F), which gives a positive signal when antibodies recognizing two proteins of interest are within ~40 nm of one another, strongly suggesting interaction. Knockdown of the target proteins confirmed antibody specificity. We performed PLAs that were quantitated by a blinded observer to test associations between the autophagosome-associated protein LC3 and either RIPK1 or active Ser 345- phosphorylated MLKL (Rodriguez et al., 2016). Both LC3-RIPK1 and LC3- phospho- MLKL interactions were detected and both were increased by TRAIL treatment. We expanded this approach to develop a dual Proximity Ligation Assay (dPLA) (Fig 3E–F, S3E–G, SI1) to simultaneously test interactions for both ATG5-RIPK1 and ATG5-MLKL in the same cells using an antibody pair targeting ATG5 and RIPK1 and a separate antibody pair targeting ATG5 and MLKL (Figure S3E). Dual PLAs were performed with the initial PLA (interaction of ATG5 and RIPK1) in red, and the second PLA (interaction of ATG5 and MLKL) in green. Both PLAs were more robust in Map3k7 null cells than floxed cells (Figure S4G) indicating increased basal interactions in the null cells. Consistent with the co-immunoprecipitations, PLA signals were increased upon treatment with TRAIL (Figure 3C–F, S3). Necrostatin-1 treatment abrogated association of RIPK1 and MLKL with ATG5 (Figure 3E–F), indicating a requirement for RIPK1 kinase activity for complex formation. The PLA experiments therefore show that necrosome proteins associate with the autophagy machinery and this is potentiated by TRAIL leading to activation of MLKL in association with autophagy proteins. To test if these complexes were forming on an autophagosomal membrane, immuno-gold Transmission Electron Microscopy (TEM) was performed with antibodies targeted to ATG5 and RIPK1 (Figure 3G–H, S3H). A small gold particle (5nm) indicates ATG5 protein and a large gold particle (15nm) indicates RIPK1 protein. Small and large particles were observed close together and these complexes and the number of immuno-gold signals in the complexes increased upon stimulation with TRAIL and were found close to double membrane structures consistent with complex formation on autophagosomal structures.

An external file that holds a picture, illustration, etc. Object name is nihms787344f3.jpg

Autophagy machinery localizes with necrosome proteins

A) Cell viability of Map3k7 null MPECs treated with TRAIL, necrostatin-1, CQ, BafA1, or Wortmannin for 24 hours or with Atg5 and Atg7 shRNA. Immunoblots show knockdown of Atg5 and Atg7 with β-actin as loading control. Data are represented as mean ± SD. B) Immunoblots of necrosome proteins and autophagy proteins, ATG7, ATG5/12, p62, BECLIN-1, ULK1, STX17, and LAMP2, after a co-immunoprecipitation of RIPK1 in Map3k7 null MPECs with and without TRAIL treatment. C) Proximity Ligation Assay (PLA) of p62 with MLKL and phospho-MLKL in Map3k7 null MPECs treated with and without TRAIL. Scale bars represent 20 μm. D) Quantification of C. Data represented as mean ± SD. E) Dual PLA of ATG5 with RIPK1 (red) and ATG5 with MLKL (green) in Map3k7 null MPECs treated with TRAIL or necrostatin. Scale bars represent 20 μm. F) Quantification of D. Data are represented as mean ± SD. G) Immuno-gold TEM in Map3k7 null MPECs treated with and without TRAIL. Scale bars represent 100 nm. Large gold particles indicated by black arrows recognize antibodies to RIPK1 and small gold particles indicated by white arrows ATG5. H) Quantification of G. Data represented as mean ± SD. See also Figure S3.

Efficient necrosome activation, but not formation, depends on ATG5

To test if autophagy proteins are required for efficient necrosome formation and activation, RIPK1 was pulled-down and association with MLKL and p62 was determined after knockdown of Atg5 (Fig 4A). No significant change in RIPK1 association with p62 was observed with knockdown of Atg5, but a subtle reduction in RIPK1 association with MLKL was observed. To test for interaction in situ, PLAs using both MLKL and phospho-MLKL with p62 were performed after Atg5 knockdown (Figure 4B–C, S5). An increase in signal was observed in both PLAs with MLKL and phospho-MLKL upon TRAIL treatment, indicating both an increase in complex formation and activation. Knockdown of Atg5 slightly reduced the signal in the PLA detecting interaction between p62 and total MLKL but completely eliminated the TRAIL-induced p62-phospho-MLKL PLA signal. An immunoblot confirmed loss of phospho-MLKL in TRAIL-treated cells with Atg5 knockdown (Figure S4). These data suggest that ATG5 is more important for TRAIL-induced necrosome activation than it is for necrosome formation in the Map3k7 null cells.

An external file that holds a picture, illustration, etc. Object name is nihms787344f4.jpg

Atg5 required for necrosome activation

A) Immunoblots of ATG5, p62, and MLKL after a co-immunoprecipitation of RIPK1 in Map3k7 null MPECs with knockdown of Atg5, with and without TRAIL treatment. B) Proximity Ligation Assay of p62 with MLKL and pMLKL in Map3k7 null MPECs treated with and without TRAIL with knockdown of Atg5. Scale bars represent 20 μm. See also Figure S4. C) Quantification of B. Data are represented as a mean ± SD.

Binding of RIPK1 to p62 and subsequent recruitment to autophagy machinery determines if cells die by necroptosis or apoptosis

RIPK1 interacts directly with p62 (Sanz et al., 1999), suggesting a mechanism by which necrosome components could be recruited to the autophagosome. To test this, we immunoprecipitated RIPK1 after knockdown of p62/Sqstm1 and found a significant reduction in ATG5 association with RIPK1 and MLKL (Fig 5A). To test if these interactions were disrupted in situ, dual PLA was performed. This showed a decrease in association of ATG5 with RIPK1 (red) and MLKL (green) following p62/Sqstm1 knockdown, even in the presence of TRAIL (Fig 5B–C). Immuno-gold TEM following p62/Sqstm1 knockdown showed a reduction in ATG5-RIPK1 complex formation, size, and localization to membranes (Figure 5D–E, S5). Surprisingly however, knockdown of p62/Sqstm1 did not significantly increase cell viability upon TRAIL treatment. Instead, cell death was now completely rescued by QVD (Fig 6A), suggesting that removal of p62/SQSTM1 caused the cells to be only able to activate apoptosis not necroptosis. Switching of the cell death mechanism from necroptosis to apoptosis upon p62/Sqstm1 knockdown was confirmed by temporal analysis of annexin V and PI staining. Upon p62/Sqstm1 knockdown, a decrease in PI staining and its colocalization with annexin V was observed with QVD treatment in TRAIL treated cells (Fig 6B, S6). These data lead to two conclusions. First, p62/SQSTM1 is needed for RIPK1 to localize to the autophagy machinery and allow for efficient assembly and activation of the necrosome. Second, it is this recruitment that causes the TRAIL-treated cells to die by necroptosis rather than apoptosis. To test this, we tested for rescue of necroptosis by mutant p62 molecules lacking the ZZ domain that is required for RIPK1 interaction, or lacking the LIR domain that is required for LC3 interaction but is not required for recruitment of p62 to autophagosomal structures (Itakura and Mizushima, 2011). An shRNA targeting the 3’UTR of p62/Sqstm1 reduced endogenous p62 and constructs expressing wild-type p62, the ΔZZ mutant, or the ΔLIR mutant were added back (Fig. 6C). Removal of p62 allowed for rescue by QVD, but addition of the wild-type p62 or ΔLIR mutant did not. The ΔZZ p62 construct mimicked complete p62 loss in that QVD alone rescued cell viability. These results show that it is p62-mediated recruitment of RIPK1 to the autophagy machinery that results in activation of necroptosis rather than apoptosis upon TRAIL treatment.

An external file that holds a picture, illustration, etc. Object name is nihms787344f5.jpg

p62 required for Ripk1 association with autophagosome

A) Immunoblots of ATG5, p62, and MLKL after a co-immunoprecipitation of RIPK1 in Map3k7 null MPECs with knockdown of p62/Sqstm1, with and without TRAIL treatment. B) Dual Proximity Ligation Assay of ATG5 with RIPK1 (red) and ATG5 with MLKL (green) in Map3k7 null MPECs with knockdown of p62 treated with and without TRAIL. Scale bars represent 20 μm. C) Quantification of C. Data are represented as mean ± SD. D) Immuno-gold TEM in Map3k7 null MPECs with p62/Sqstm1 shRNA treated with and without TRAIL. Scale bars represent 100 nm. Large gold particles indicated by black arrows recognize antibodies to RIPK1 and small gold particles indicated by white arrows, ATG5. E) Quantification of D. Data represented as mean ± SD. See also Figure S5.

An external file that holds a picture, illustration, etc. Object name is nihms787344f6.jpg

p62 regulates the switch between apoptotic versus necroptotic cell death

A) Cell viability of Map3k7 null MPECs treated with TRAIL, QVD, and necrostatin-1 with knockdown of p62/Sqstm1 (white bars). B) Map3k7 null MPECs treated with TRAIL, QVD, and necrostatin-1 with p62/Sqstm1 knockdown and imaged in the presence of annexin V and PI on an incucyte in 4 fields of view every 4 hours for 24 hours. Representative images at 4 hour timepoint with scale bars representing 100 μm. Co-localization of annexin V and PI determined using CellProfiler and Pearson correlation coefficient (r) determined. See also Figure S6. C) Cell viability of Map3k7 null MPECs treated with TRAIL, QVD, and necrostatin-1 with knockdown of p62/Sqstm1 and addition of either wild-type (WT) p62, mutant p62 lacking the ZZ domain (ΔZZ), or mutant p62 of the LIR domain (ΔLIR). Data are represented as mean ± SD.

Discussion

Regulation of cell death is critical in development and disease. However, while it has been recognized for many years that the decision whether or not a cell should die has critical biological consequences, until recently we have had a much poorer understanding of the importance of how a cell dies and how this is controlled. For instance, in mice the loss of cell death executioners that only affect apoptosis such as caspase-3, caspase-7, and Bax/Bak display relatively mild morphogenesis defects (Lakhani et al., 2006; Lindsten et al., 2000). However, loss of genes that have a role in both apoptosis and necroptosis such as caspase-8 or Fadd, lead to more severe phenotypes that can be rescued by coordinate loss with Ripk3 or Ripk1 (Kaiser et al., 2011; Oberst et al., 2011; Zhang et al., 2011). Therefore, there are important interactions between these two death processes that are crucial for proper development. Necroptosis is also critical in inflammation (Pasparakis and Vandenabeele, 2015) and the amount of necroptosis versus apoptosis in dying tumor cells may affect the release of factors that are required for anti-tumor immune responses (Yatim et al., 2015). Our work provides insights into these processes by defining a mechanism by which regulated switching between necroptosis and apoptosis can be achieved through recruitment of necrosome components to the autophagy machinery. This work also opens up another aspect of how autophagy regulates programmed cell death. It is becoming better understood how autophagy controls the likelihood of a cell dying (Fitzwalter and Thorburn, 2015). Our results show that in addition, the autophagy machinery can also control the way that the cells die by serving as a scaffold that allows more efficient activation of the necrosome leading to phosphorylation of MLKL that is able to cause cells to undergo necroptosis. Importantly, if this mechanism does not apply (e.g. due to lack of RIP3 expression), cells may still die but instead do so by activating caspase-dependent apoptosis. Our experiments suggest that the key step in recruitment of necrosome components to the autophagy machinery for efficient phosphorylation of MLKL and activation of necroptosis requires p62 but that this is independent of the LIR domain that binds LC3. Self-oligomerized p62 can bind to early autophagosomal structures independently of interaction with LC3 (Itakura and Mizushima, 2011). Mutation of the PB1 oligomerization domain in p62 resulted in a protein that was toxic when expressed in Map3k7 null cells, and so we were unable to directly test if p62 oligomerization was required for autophagosome recruitment of RIPK1 and necroptosis. The simplest explanation of our data is that p62 recruits RIPK1 to early autophagosomal structures and (because we see interaction between RIPK1 and proteins such as LC3 that are associated with more mature autophagosome structures) this complex is retained as the autophagosome matures allowing increased necrosome activation. Complexes that we saw by electron microscopy often appeared on quite mature structures consistent with this idea.

Other anti-cancer agents may also activate necroptosis involving autophagy (Basit et al., 2013; Kharaziha et al., 2015). Importantly however, the mechanism here applies under specific circumstances that suggest a potential for applying this knowledge in cancer treatment– in cells that have lost a specific tumor suppressor and in response to a death stimulus, TRAIL, which can be used clinically (Lemke et al., 2014) and is itself an important endogenous tumor and metastasis suppressor (Johnstone et al., 2008). Our data indicate that MAP3K7 loss sensitizes the cells to undergo necroptosis because the components of the necrosome (RIPK1, RIPK3 and MLKL) associate together even in the absence of exogenous signaling, however TRAIL receptor signaling and association with the autophagy machinery is needed for the necrosome to be efficiently activated leading to phosphorylation of MLKL. MAP3K7 loss defines an aggressive sub-type of prostate cancer, and sensitivity to TRAIL-induced cell death by necroptosis that could be further regulated using clinically-available autophagy inhibitors such as chloroquine may open up a new strategy for treating such tumors. Fig. 1 suggests that about half of patients whose tumors will have homozygous loss of the MAP3K7 gene retain TRAIL receptor expression and could therefore potentially benefit from such an approach. With over 200,000 new prostate cancer cases a year in the United States, this may amount to over 10,000 such patients. Additionally, by activating necroptosis, this approach might better activate an anti-tumor immune response (Yatim et al., 2015). Because these effects are mediated by the autophagy machinery serving as a scaffold rather than by degrading cellular material, inhibition of autophagy before or after formation of autophagosomes can have opposing effects showing that it is important not just if the autophagy pathway is inhibited, but where the pathway is inhibited. With autophagy inhibitors already in clinical trials for cancer treatment (Rebecca and Amaravadi, 2015) and suggestions of benefit in some patients (Levy et al., 2014) it is increasingly important to understand how autophagy modulates both the amount and type of tumor cell death, especially as new inhibitors targeting different steps in the autophagy pathway (Egan et al., 2015; Goodall et al., 2014; McAfee et al., 2012) could have opposing effects on these mechanisms.

Materials and Methods

Reagents and Cell Lines

Mouse Prostate epithelial cells (MPECs) with floxed and Cre-deleted Map3k7 (Wu et al., 2012)) were maintained and cultured in DMEM/F12 (Gibco, 11320) with Fraction 5 BSA (Sigma), Cholera Toxin (List Biologicals, 101B), Bovine Pituitary Extract (Hammond Cell Tech, 1078-NZ), Gentamicin, Insulin (Sigma), Vitamin E, Transferrin, Trace Elements, EGF (Collaborative Research), and 1% FBS at 37°C with 5% CO2.

Dataset analyses

Array comparative genomic hybridization data and clinical data for the MSKCC (Taylor et al., 2010), TCGA (http://cancergenome.nih.gov/), and Broad/Cornell-2 (Baca et al., 2013) prostate cancer datasets were downloaded from the cBioPortal for Cancer Genomics (Cerami et al., 2012; Gao et al., 2013). Data were processed as described in the source publications.

Cell Viability Assays

Cells were seeded at 2,000 cells per well, in 96-well plates (Corning, Corning, NY) and incubated overnight or 500 cells per well with shRNA for 48 hours for knockdown experiments. Cells were treated in a dose response to TRAIL (R&D Systems, 375-TEC), [20μM] QVD (R&D Systems, OPH00101), [30 μM] Necrostatin-1 (Enzo Life Sciences, BML-AP309-0020), [40 μM] CQ (Sigma, C6628), [100 nM] BafilomycinA1 (Sigma, B1793), [100nM] Wortmannin (Sigma, W1628), or [0.5 nM] Staurosporine (Sigma, S5921) for 24 hours. All experiments were performed at least three times in triplicate and the proportion of cells per treatment group was normalized to control wells. LDH release was quantitated using the Cytoscan-LDH Cytotoxicity Assay Kit (G-Biosciences, 786-210) according to manufacturer’s instructions and absorbance was read at 490nm using a Benchmark Plus microplate spectrophotometer (BioRad). Viable cells were measured using the Cell Titer-Glo luminescent cell viability assay (Promega, G7572) following the manufacturer's protocol and luminescence was read using a Modulus Microplate reader.

Immunoblotting

Cells were seeded at 150,000 cells per well and treated with 1 ng/mL and 100 ng/mL TRAIL, [20μM] QVD, [30μM] Necrostatin-1, [10 nM] Rad001, or [40μM] CQ for 24 hours or starved with Earl’s Based Salt Solution (EBSS) (Sigma, E2888) for 4 hours prior to collection. Cell lysates were harvested after treatments and time-points indicated using RIPA buffer (4M NaCl, 0.01% NP-40, 2% Sodium Deoxycholate, 10% SDS, 1M Tris, 500mM NaF, 0.5M EDTA, dH20) with phosphatase inhibitors (Roche, 04693116001). Membranes were blocked in 5% BSA or 5% milk in 1x TBS-Tween, depending on antibody, and probed with primary antibodies at manufacturer recommended concentrations. Anti-β actin (Sigma, A5541) was used as the protein loading control. For list of antibodies see Supplemental Experimental Procedures.

shRNA Infections

A pLKO.1 vector system was utilized for RNAi of necrosome and autophagy related proteins and were acquired from the Functional Genomics Facility at the University of Colorado Cancer Center. Lentiviruses were prepared according to protocols published at the RNAi Consortium webpage (http://www.broadinstitute.org/rnai/trc/lib). Cells were plated at 20,000 cells/well in 6 well plates and transduced with lentivirus using 8ug/mL polybrene. After 48–72 hours, cells were plated for assays following cell counts listed. In 96 well assays, knockdown was done in the plate at 500 cells/well for 48 hours. TRC numbers for shRNAs are listed in Supplemental Experimental Procedures.

Co-immunoprecipitations

MPEC cells were cultured at 1 x 106 in 15 cm plates with and without shRNA for 72 hours at 37°C with 5% CO2. A higher concentration of TRAIL was used for shorter time points to increase the speed of necrosome formation. Cells were treated with 500 ng/mL TRAIL for 2 hours and lysates harvested in IP buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, and 1 mM β-glycerolphosphate) with phosphatase inhibitors. Lysates were spiked with 8 μg of RIPK1 antibody (BD Transduction Lab, 610459) and incubated overnight at 4°C. Protein A/G agarose beads (Santa Cruz, #sc-2003) were added and incubated for 2 hours and washed with IP buffer at least three times. Samples were run on a polyacrylamide gel and immunoblotted.

Proximity Ligation Assays (PLAs)

Cells were seeded at 10,000 cells per well in an 8-well chamber coverslide (Millipore, PEZGS0816) and incubated overnight at 37°C in 5% CO2. Cells were treated with TRAIL for 2 hours and fixed with 3.7% formaldehyde. The PLA pair of monoclonal rabbit ATG5 (Abcam, ab108327) and mouse RIPK1 (BD Transduction Lab, 610459), PLA pair of polyclonal ATG5 (Novus, NB110-53818) and MLKL (Millipore, MABC604), PLA pair p62/Sqstm1 rabbit (Abgent, AP21183B) and rat MLKL or mouse phospho-Ser345-MLKL (Rodriguez et al., 2015), and LC3 (Novus Biologicals, NB100-2220) were incubated overnight at optimized dilutions based on immunofluorescence and the PLA was performed to manufacturer’s directions (Sigma, DUO92008, DUO92002, DUO92004) and signals quantitated by a blinded observer. For the dual PLA see Supplemental Experimental Procedures.

Flow Cytometry

MPEC cells (both floxed and Cre deleted) constitutively expressing mCherry-GFP-LC3 were seeded at 2 x 105 in a 6-well and incubated overnight. Cells were exposed to control media, Rad001, or BafA1 (for baseline inhibited autophagy) for 24 hours or EBSS starvation media for 4 hours prior to collection. Flow data was acquired on a Gallios561 and analyzed using Summit v5.1 (Beckman Coulter, Fort Collins, CO). Autophagic flux was determined by the ratio of mCherry:GFP. For Annexin V/PI, Caspase3/7, or Reactive Oxygen Species (ROS) Assays see Supplemental Experimental Procedures.

IncuCyte Cell Death Assay

MPEC cells were seeded at 500 cells per well in a 96-well plate (Costar, Corning, NY). Cells were cultured at 37° and 5% CO2 and monitored using an IncuCyte Zoom (Essen BioScience, Ann Arbor, MI). Cells were exposed to either TRAIL, QVD, or Necrostatin-1 for 24 hours in the presence of Annexin V and PI and images were captured at 4-hour intervals from 4 separate regions per well using a 10x objective. Each experiment was done in triplicate and co-localization was determined by exported green and red masks in CellProfiler (Broad Institute, www.cellprofiler.org).

Immuno-gold Transmission Electron Microscopy (TEM)

MPEC cells were seeded in 10 cm plates and incubated overnight at 37°C in 5% CO2 and treated as described then trypsinized, washed, pelleted, and resuspended in 0.5% glutaraldehyde and 4% formaldehyde fixatives in 0.1M cacodylate buffer. Cell pellets were embedded in 2% agarose, postfixed in osmium tetroxide, and dehydrated with an acetone series. Cell samples were infiltrated and embedded in LR White resin and polymerized at 60°C for 24 hours. Ultrathin sections of 70nm were generated and placed on nickel coated grids then incubated for 1 hour in 5%BSA + 5% goat serum in 1x PBS, washed, and incubated in primary ATG5 (rabbit) and RIPK1 (mouse) antibodies (listed above) overnight at 4°C. Grids were washed and incubated i n secondary goat-anti-Rabbit 6nm gold antibody (Electron Microscopy Sciences, 25103) and goat-anti-mouse 15nm gold antibody (Electron Microscopy Sciences, 25132) for 1 hour. Sections were examined using a Transmission Electron Microscope by the Electron Microscopy Center (University of Colorado Denver, Anschutz Medical Campus). Autophagosomal structures were identified as double membrane structures. Images were quantified using Nikon NIS Elements software, thresholded excluding anything outside of 0.6-1 circularity and larger than 20nm in size. Regions of Interest (ROIs) were drawn around anything containing 2 or more gold particles for quantification. Accounting for gold particle standard deviations, small gold particles (ATG5) were considered anything 4–8nm and large gold particles (RIPK1) were considered anything 10–18nm. Large complexes were defined as containing more than 3 particles, with at least one of each size. Complexes near a membrane were considered those within the ROI centered around the membrane.

p62 constructs

Human p62 cDNA was inserted into the pEGFP-C1 vector from Clontech, and site directed mutagenesis was performed to delete the ZZ domain of p62. Primers used were 5’-GCACCCCAATGTGATCCACCAA and 5’-GAATGCGAGCTTGGTGGATCAC. Mutations to the LIR domain were D335/336/337A based on Itakura et al. Primers used were 5’-GACAGATGGGTCCAGGC ATCAGCTCCTCCTGAACAGT, 5'-ACTGTTCAGGAGGAGCTGATGCCTGGACCCATCTGTC, 5'-GATGGGTCCAGGCAGCAGCTCCTCCTGAA, and 5'-TTCAGGAGGAGCTGCTGCCTGGACCCATC

​

Highlights

Acknowledgments

We thank Jenny Mae Samson for help with some experiments and Alicia Pastor from the Electron Microscopy core at Michigan State University for help with immuno-gold TEM. Supported by NIH grants RO1CA150925, RO1CA190170, RO1CA199741, R21CA187354, 1F32CA196080, and Shared Resources supported by P30CA046934.

Footnotes

Author Contributions

MLG, SC and AT conceived the project. MLG designed and executed most of the experiments and analyzed data. AT, MM, and SC contributed to experimental design. BF, SZ, MW and JML performed some experiments and helped with analysis. DR and DG provided essential reagents. MLG and AT wrote the paper with contributions from all the authors.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References