Research ArticleCell Biology

Regulated proteolysis of p62/SQSTM1 enables differential control of autophagy and nutrient sensing

See allHide authors and affiliations

Sci. Signal.  04 Dec 2018:
Vol. 11, Issue 559, eaat6903
DOI: 10.1126/scisignal.aat6903

Cleaving a different function for p62

The scaffold protein p62 has a critical role in autophagy, the regulated degradation of proteins and organelles, and xenophagy, an autophagic process that clears invading pathogens and that may require the activation of Toll-like receptors (TLRs). Sanchez-Garrido et al. (see also the Focus by Martens) found that in skin fibroblasts, stimulation of TLR3, which detects double-stranded RNA typical of microbial pathogens, resulted in the generation by the protease caspase-8 of a proteolytic fragment of p62 that the authors called p62TRM. Instead of functioning in autophagy or xenophagy, p62TRM enhanced the ability of the amino acid leucine to activate mTORC1, a multiprotein complex that couples cellular growth and proliferation to nutrient availability. This function was lost in p62 variants with mutations in the caspase-8 cleavage site, one of which is associated with frontotemporal dementia. Thus, proteolytic cleavage of p62 generates a fragment with a distinct cellular function from that of the full-length protein, and loss of this cleavage event may account for the symptoms of patients with mutations in the gene encoding p62.


The multidomain scaffold protein p62 (also called sequestosome-1) is involved in autophagy, antimicrobial immunity, and oncogenesis. Mutations in SQSTM1, which encodes p62, are linked to hereditary inflammatory conditions such as Paget’s disease of the bone, frontotemporal dementia (FTD), amyotrophic lateral sclerosis, and distal myopathy with rimmed vacuoles. Here, we report that p62 was proteolytically trimmed by the protease caspase-8 into a stable protein, which we called p62TRM. We found that p62TRM, but not full-length p62, was involved in nutrient sensing and homeostasis through the mechanistic target of rapamycin complex 1 (mTORC1). The kinase RIPK1 and caspase-8 controlled p62TRM production and thus promoted mTORC1 signaling. An FTD-linked p62 D329G polymorphism and a rare D329H variant could not be proteolyzed by caspase-8, and these noncleavable variants failed to activate mTORC1, thereby revealing the detrimental effect of these mutations. These findings on the role of p62TRM provide new insights into SQSTM1-linked diseases and mTORC1 signaling.


The scaffold protein p62, also called SQSTM1 (encoded by SQSTM1), is involved in several homeostatic processes such as macroautophagy (henceforth referred to as autophagy) (13). The many domains within p62 protein facilitate protein-protein interactions for its various cellular roles (fig. S1A). For example, p62 promotes autophagic turnover of ubiquitylated cargo by binding LC3-like proteins and ubiquitin chains, nuclear factor κB (NF-κB) activation through atypical protein kinases C (aPKCs), tumor necrosis factor (TNF) receptor–associated factor 6 (TRAF6) and receptor-interacting serine/threonine kinase 1 (RIPK1), redox stress response through the E3 ligase kelch-like ECH-associated protein 1 (KEAP1), and cell growth and anabolism through the Ras-related guanosine triphosphate (GTP) binding (RAG) family of guanosine triphosphatases (GTPases) that regulate the mechanistic target of rapamycin complex 1 (mTORC1). As an autophagy adaptor, p62 facilitates both antimicrobial immunity and homeostasis. During xenophagy, it targets intracellular pathogens and restricts their replication, and during mitophagy, it facilitates the turnover of damaged mitochondria (4). More than 30 disease-linked nonsynonymous mutations in human SQSTM1 are linked to hereditary inflammatory diseases (57). These include Paget’s disease of the bone, frontotemporal dementia (FTD), amyotrophic lateral sclerosis (ALS), and distal myopathy with rimmed vacuoles. Most mutations map to the ubiquitin binding–associated region (UBA) (fig. S1A) and interfere with ubiquitin binding, which, in some cases, alter autophagy responses (57). Because of the multifaceted nature of p62 and ubiquitous expression, the mechanisms by which natural mutations affect some of the functions of p62 in specific cell types and cause disease have remained unclear. Therefore, a better understanding of context-specific signaling through p62 is essential.

Cross-talk between autophagy and Toll-like receptors (TLRs) promotes antimicrobial immunity (810). Unlike other TLRs, TLR3 uses only the adaptor protein TRIF [TIR-Toll/interleukin-1 (IL-1) receptor domain–containing adaptor protein inducing interferon-β (IFN-β)], which recruits RIPK1 for downstream signaling. TLR3 detects double-stranded RNA and stimulates proinflammatory cytokine induction through NF-κB and mitogen-activated protein kinase pathways, and type I IFN production through IFN regulatory factor 3. Natural loss-of-function or dominant-negative deficiencies in human TLR3 signaling genes cause pediatric susceptibility to herpes simplex virus infections, which point to a key role for TLR3 in humans (11). The interaction of the kinase RIPK1 with the protease caspase-8 determines the outcomes of TLR3 signaling in a context-specific and cell type–dependent manner (12). During antimicrobial responses, RIPK1 becomes polyubiquitylated, restrains caspase-8 activity, and promotes cell survival, gene expression, and immunity (1215). In contrast to normal fibroblasts, macrophages, and dendritic cells, TLR3 signaling in several cancer cell types, such as melanomas and breast cancer cells, can trigger necrotic cell death (1620). These findings have led to an interest in developing polyinosinic-polycytidylic acid [poly(I:C)], a synthetic TLR3 ligand, as an anticancer adjuvant (21, 22). Therefore, understanding the mechanisms of human TLR3 signaling is also important from a therapeutic point of view.

The molecular basis for the functional separation of the cellular pool of p62 in different pathways remains poorly understood. For example, nutrient starvation of cells triggers autophagic turnover of cytosolic contents to sustain metabolism; p62 serves as a selective cargo receptor for autophagy and is itself turned over within autophagolysosomes. On the other hand, during nutrient-replete conditions, such as plentiful amino acid supply, p62 is involved in activating the kinase in mTORC1 (2325). mTORC1 integrates nutrient availability and growth factor signaling and phosphorylates substrates such as the p70 ribosomal protein S6 kinase 1 (p70S6K1) and eIF4E-binding protein to promote protein translation, anabolism, and cell growth (26, 27). Notably, mTORC1 phosphorylates and inhibits proteins that initiate autophagy. Autophagy and mTORC1 thus oppose each other during nutrient-deplete and nutrient-replete conditions (28). However, because p62 can be rapidly turned over by starvation-induced autophagy (29), how p62 participates in the physiologically opposing mTORC1 pathway when nutrients become available is not clear.

Previous in vitro work suggested that caspases can cleave several autophagy proteins, including p62 (30, 31). While studying TLR3 signaling in human donor–derived skin fibroblasts, we found that caspase-8 proteolytically trimmed p62 into a stable protein, which we called p62TRM. Our studies revealed that proteolysis of p62 into p62TRM was regulated and that p62TRM activated mTORC1. We found two rare natural polymorphisms in SQSTM1 that abrogated its cleavage by caspase-8 and specifically impaired mTORC1 functions. These findings provide new molecular insights into the detrimental effects of these mutations in SQSTM1 and the exclusive and opposing roles of full-length p62 in autophagy and antimicrobial xenophagy and of p62TRM in mTORC1 activation.


Caspase-8 can cleave p62 at Asp329

We were interested in assessing the autophagic turnover of p62 induced by microbial TLR ligands in healthy donor–derived skin fibroblasts (called C12 cells here). We noticed that treatment with the TLR3 ligand poly(I:C), but not with the TLR2/6 ligand PAM2CSK4, led to the appearance of a smaller fragment of p62 on immunoblots (Fig. 1A). The dependence of the appearance of a shorter (~40 kDa) p62 protein (p62TRM henceforth) on poly(I:C) concentration and treatment time was consistent with its generation by proteolytic cleavage (Fig. 1A and fig. S1, B and C). Consistent with our previous work with donor-derived skin fibroblasts, the viability of C12 cells treated with poly(I:C) was similar to that of untreated cells as measured by lactate dehydrogenase (LDH) release, propidium iodide (PI) uptake, and live time-lapse imaging (Fig. 1B and fig. S1D). These findings ruled out that these TLR ligands induced necrosis in our experiments. Although TLR3 can induce apoptosis in some contexts, lysates from poly(I:C)-treated C12 cells did not show evidence for apoptosis because we did not observe proteolytic activation of apoptotic caspases (caspase-8, caspase-3, and caspase-7) or the apoptosis substrate PARP1 [poly(adenosine 5′-diphosphate–ribose) polymerase 1] (Fig. 1C). In contrast, apoptosis-inducing drugs such as staurosporine or actinomycin D induced the processing of these caspases and PARP1, and cell death in C12 cells (Fig. 1, B and C). However, processing of p62 into p62TRM was only observed with poly(I:C) treatment. This suggested that p62TRM production is uncoupled from cell death and proteolytic activation of caspases (Fig. 1, B and C).

Fig. 1 Caspase-8 cleaves p62 to generate p62TRM.

(A) Immunoblots of healthy donor–derived skin fibroblast (C12 cells) left untreated (UT) or treated with PAM2CSK4 (5 μg/ml) or poly(I:C) (10 μg/ml) for 20 hours. (B and C) Representative two-channel images (B) and immunoblots (C) of C12 cells left untreated (UT) or treated with poly(I:C) (10 μg/ml, 20 hours), staurosporine (STS; 1 μM, 6 hours), or actinomycin D (ActD; 2 μM, 20 hours). In (B), cells were treated in the presence of PI (red). Scale bars, 100 μm. (D) Immunoblots from C12 cells transfected with nontargeting (CTRL) or the indicated small interfering RNA (siRNA) for 72 hours and left untreated or treated with poly(I:C). (E) Immunoblots from HeLa cells transfected with nontargeting (CTRL) or caspase-8 siRNA for 72 hours and then left untreated or treated with poly(I:C) as indicated. (F) Top: Schematic of GST-p62297–440 showing the region around the caspase-8 cleavage site (Asp329) and expected sizes of fragments after proteolysis. Bottom: Coomassie-stained gel of assay with increasing amounts of caspase-8 and WT or D329A GST-p62297–440; IETD, caspase-8 inhibitor Ac-IETD-fmk; *, caspase-8 p20 from self-processing; ter, terminus. (G) Immunoblots from C12 cells transfected with indicated Flagp62 variants and treated with poly(I:C). Data represent n = 2 (D) or n = 3 (A to C and E to G) biologically independent experiments.

Because previous in vitro work implicated caspase-1, caspase-6, and caspase-8 in processing p62 (30, 31), we investigated their role further using inhibitors and RNA interference (RNAi). The appearance of p62TRM was blocked by pan-caspase inhibition with zVAD-fmk and caspase-8 inhibition with Ac-IETD-fmk, but not caspase-1 inhibition with Ac-YVAD-fmk (fig. S1E); bafilomycin-A1 (Baf-A1) was effective only when added with poly(I:C) to block lysosomal acidification, which inhibits TLR3 signaling, but not when added 4 hours later to reduce lysosomal protease activities without interfering with signaling (fig. S1, D and E). RNAi against caspase-1, caspase-3, caspase-6, caspase-7, caspase-8, and caspase-10 revealed that caspase-8 was essential for p62TRM generation, but other caspases were dispensable (Fig. 1D and fig. S2A). In agreement with this finding, a small but measurable increase in caspase-8 activity was detected in poly(I:C)-treated cells (fig. S2B). Similarly, caspase-8 silencing in HeLa cells impaired p62TRM production by poly(I:C) (Fig. 1E). Together, these experiments established that poly(I:C)-TLR3 signaling induces caspase-8–dependent, cell death–independent proteolysis of p62 into p62TRM in cells.

We therefore asked whether p62 was a direct substrate of caspase-8. Although caspase-1 can cleave p62 at Asp329 in vitro (30), chemical inhibitor and RNAi results ruled out the involvement of caspase-1 in intact cells (Fig. 1D and fig. S1E). We hypothesized that caspase-8 might cleave p62 at Asp329 because this would be consistent with the ~40-kDa p62TRM detected on immunoblots (amino acids 1 to 329; Fig. 1A and figs. S1A and S2C). Recombinant caspase-8 proteolyzed glutathione S-transferase (GST)–p62, but not the D329A cleavage site mutant (Fig. 1F). Similarly, wild-type (WT) but not D329A p62 was cleaved into p62TRM by transfected caspase-8 in human embryonic kidney (HEK) 293E cells (fig. S2D). Furthermore, poly(I:C) stimulation of C12 fibroblasts and HeLa cells resulted in processing of WT but not D329A p62 into p62TRM (Fig. 1G and fig. S2E). Human p62 expressed in mouse L929 fibroblasts was cleaved in response to poly(I:C) treatment, pointing to a similar TLR3–RIPK1–caspase-8 pathway in mouse fibroblasts, whereas mouse p62 lacks the equivalent aspartic acid residue (Asp331) (fig. S2F). Thus, we concluded that TLR3 signaling triggers p62 processing at Asp329 by caspase-8 and generates the p62TRM protein species in cells.

Caspase-8–driven generation of p62TRM requires RIPK1 but is independent of TRAF3, TBK1, or ATG7

To investigate TLR3 signaling genes required for p62TRM production, we used skin fibroblasts from pediatric patients with previously characterized natural loss-of-function or dominant-negative mutations in UNC93B1, TLR3, TRIF, TRAF3, and TBK1 (Fig. 2A) (3237). Poly(I:C)-induced production of p62TRM was abrogated by inactivating mutations in UNC93B1, TLR3, or TRIF, but not by those in TRAF3 or TBK1 (Fig. 2A). We concluded that TLR3, its trafficking by UNC93B, and its signaling adaptor TRIF were essential for p62 processing induced by poly(I:C). Because TBK1-deficient cells are severely defective in TLR3-driven type I IFN production (33, 36), these results also ruled out a role for IFNs in the cleavage of p62 (fig. S2G). TRIF interacts with RIPK1, which recruits caspase-8 to alter gene expression and promote cell survival in a context-dependent manner (12). Stable silencing of RIPK1 in C12 and HeLa cells with an optimized miRNA30E (miR)–based plasmid impaired poly(I:C)-induced cleavage of p62 (Fig. 2B and fig. S3A). Notably, necrostatin-1 did not affect p62 cleavage in either cell type (Fig. 2, C and D), suggesting that the kinase activity of RIPK1, which is essential for RIPK1-dependent necroptosis (fig. S3B), was dispensable for p62TRM production and was consistent with the lack of cell death under these conditions. Furthermore, p62TRM production was not affected by silencing the core autophagy genes ATG7 or RB1CC1, which encodes FIP200, suggesting that autophagy was also dispensable (Fig. 2E and fig. S3C). Together, the RIPK1 branch of the bipartite TLR3 signaling pathway stimulates p62 proteolysis independently of necroptosis and autophagy.

Fig. 2 RIPK1 is required for caspase-8–driven generation of p62TRM.

(A) Schematic (left) of TLR3 signaling and immunoblots (right) from skin fibroblasts of individuals with dominant loss of function in the indicated genes that were untreated (−) or treated (+) with poly(I:C) as indicated. (B) Immunoblots from C12 cells stably expressing nontargeting control (CTRL) or RIPK1-targeting miR treated with poly(I:C). (C) Immunoblots from C12 cells treated with poly(I:C) in the presence of the indicated inhibitors or vehicle [dimethyl sulfoxide (DMSO)]. (D) Immunoblots from HeLa cells left untreated or treated with poly(I:C) in the presence of DMSO or the indicated inhibitors. (E) Immunoblots from HeLa cells stably expressing CTRL or ATG7 miR left untreated or treated with poly(I:C). Blots on the right show ATG7 expression. IETD, Ac-IETD-fmk (caspase-8 inhibitor); NEC1, necrostatin-1 (RIPK1 inhibitor). Data represent n = 2 (D) or n = 3 (A to C and E) biologically independent experiments.

p62TRM cannot participate in autophagy and antibacterial xenophagy

The commercial antibodies we used to detect p62 recognized epitopes within the p62TRM region and could not detect the C-terminal p62330–440 fragment. A monoclonal antibody with an epitope within p62330–440 also did not detect endogenous p62330–440 upon poly(I:C) treatment, which suggested that this fragment may be less stable than p62TRM and was being degraded (fig. S3D). Some p62330–440 was detected in the combined presence of inhibitors of the proteasome (epoxomicin) and vacuole acidification (Baf-A1) (fig. S3D). We reasoned that the presence of the LC3-interacting region (LIR) and UBA might contribute to the turnover of p62330–440. Supporting this notion, mutational deletion of the LIR (p62ΔLIR) resulted in detectable levels of the p62330–440 fragment in cells (fig. S3E). These findings led us to conclude that p62330–440 was turned over and p62TRM was the only stable species produced upon cleavage of p62 by caspase-8.

We therefore wanted to identify the role of p62TRM. As expected, p62TRM did not interact with LC3B when endogenous p62 was silenced; the presence of WT p62 enabled p62TRM co-immunoprecipitation (co-IP) with LC3B through PB1 domain–dependent oligomerization (Fig. 3A). We also tested p62TRM in four autophagy assays. First, we assessed the rate of autophagic turnover of p62 and p62TRM. To avoid the confounding effects of endogenous p62, we reconstituted p62-silenced HEK293p62miR cells with Flag-tagged miRNA-resistant p62 variants (Fig. 3A). Torin 1, which induces autophagy by inhibiting mTORC1, stimulated the turnover of WT p62 in a Baf-A1–sensitive manner (Fig. 3B), verifying the degradation of p62 in autophagosomes. However, p62TRM levels were not affected by Torin 1 (Fig. 3B), which was likely due to its failure to be recruited to autophagosomes. On the other hand, in cells with intact expression of endogenous p62, transfected Flagp62 was similarly turned over upon Torin 1 treatment, but Flagp62TRM levels did not decrease appreciably (Fig. 3C). This finding indicated that although a portion of cellular Flagp62TRM undergoes autophagic degradation, probably by interacting with full-length p62, a large fraction remains resistant to autophagy. Second, we quantified the formation of ubiquitylated aggregates containing p62 or p62TRM by microscopy. Here, we used CRISPR-Cas9–targeted SQSTM1-deficient (Δp62) HeLa cells that were stably transduced with Flag-tagged p62 or p62TRM. The percentage of cells with large (>1 μm in size) ubiquitin and WT p62 (p62+Ub+) aggregates increased fourfold after poly(I:C) stimulation due to the induction of autophagy (10), but such aggregates were not present in p62TRM-expressing cells without or with stimulation (Fig. 3D). This finding indicated that p62TRM failed to aggregate ubiquitylated cargo. Third, we quantified p62 and p62TRM trafficking to Listeria monocytogenes ΔactA. L. monocytogenes use the ActA virulence protein for actin-based cytosolic motility, evasion of xenophagic capture, and enhanced replication; the ΔactA mutant is ubiquitylated and cleared by p62-dependent autophagy (38). At 3 hours after infection, targeting of L. monocytogenes ΔactA by p62TRM was ~20-fold less than WT p62, further highlighting the trafficking defects of p62TRM (Fig. 3E). Fourth, we quantified the inhibition of cytosolic replication of L. monocytogenes ΔactA by p62 and p62TRM. Whereas WT p62 restricted L. monocytogenes ΔactA replication between 2 and 10 hours after infection, p62TRM was completely defective (Fig. 3F), indicating a failure in bacterial clearance through xenophagy. Together, these experiments established that caspase-8–processed p62 lacks the C-terminal fragment and p62TRM cannot execute autophagy-related functions in cells.

Fig. 3 p62TRM cannot execute autophagy-related functions.

(A) Top: Schematic representation of p62 and p62TRM. Bottom: Immunoblots (IBs) from co-IP experiments in HEK293E cells expressing nontargeting (CTRL) or p62-targeting miR. Cells were transiently transfected with MycLC3B and Flagp62TRM. (B) Immunoblots from HEK293E cells showing p62 turnover upon treatment with Torin 1 (250 nM) in a Baf-A1 (100 nM)–sensitive manner. Indicated Flag-tagged p62 variants were transiently transfected in HEK293E cells, and endogenous p62 was stably silenced with a p62 3′ untranslated region (3′UTR)–specific miR (HEKp62miR cells). (C) Immunoblots from HEK293E cells transiently transfected with either Flagp62 or Flagp62TRM and left untreated or treated with Torin 1 (250 nM) without or with Baf-A1 (100 nM) for 4 hours. (D and E) Representative images and quantification from immunofluorescence analyses of Δp62 HeLa cells stably expressing the indicated Flag-tagged p62 variants showing trafficking defects of p62TRM. In (D), cells were treated with poly(I:C) for 20 hours, and Flagp62 variants and ubiquitin (Ub) were stained. In (E), cells were infected with L. monocytogenes ΔactA for 3 hours, and Flagp62 variants and bacteria were stained. Regions in white box are shown in insets. Scale bars, 10 μm. Plots show means ± SEM. n = 3 independent experiments. nd, not detected. ***P < 0.005 by paired Student’s t test. (F) Impaired restriction of L. monocytogenes ΔactA by p62TRM. Plots show fold replication between 2 and 10 hours after infection of indicated Δp62 HeLa cells (matched means; n = 6 independent experiments). ns, not significant; ***P ≤ 0.005 by paired Student’s t test. Data in (A) to (E) represent n = 3 biologically independent experiments.

p62TRM is generated during leucine starvation and is required for leucine sensing by mTORC1

We asked how the processing of p62 might affect its scaffolding role in NF-κB activation and nutrient sensing. p62 regulates NF-κB activation through aPKCs, especially in T cells, osteoclasts, and osteoblasts (1, 2). However, NF-κB–dependent production of the inflammatory IL-6 cytokine upon poly(I:C) or TNF treatment of skin fibroblasts was not affected by p62 silencing (fig. S3F). This finding ruled out a role for p62 in NF-κB activation downstream of TLR3 or TNF signaling in fibroblasts.

In contrast, basal IL6 expression by p62-deficient C12 and HeLa cells was greater (fig. S3G). Because mTORC1-p70S6K1 controls MYC abundance and IL-6 production (25, 39), we investigated the interplay between p62 and mTORC1 in human fibroblasts. Reduced p62 expression by stable silencing in C12 cells (C12p62miR) or loss of p62 expression by CRISPR-Cas9 in HeLa cells (Δp62 HeLa cells) impaired mTORC1 activation by amino acids (fig. S4A) but, more specifically, by leucine (Fig. 4, A and B). mTORC1 activation by arginine or glucose was p62 independent (Fig. 4, A and B). Further, the phosphorylation of p70S6K1 and the abundance of MYC were reduced in C12p62miR and Δp62 HeLa cells (fig. S4B); both cell types also produced more IL-6 (Fig. 4C and fig. S4C). Direct inhibition of mTORC1 with Torin 1 also increased IL-6 production (fig. S4D). Thus, results from RNAi, CRISPR-Cas9 targeting, and chemical inhibition independently verified the p62–mTORC1–MYC–IL-6 axis in human fibroblasts. Δp62 cells also showed reduced proliferation (fig. S4E), further validating reduced mTORC1 function (27).

Fig. 4 p62TRM promotes leucine sensing by mTORC1.

(A and B) Immunoblots from C12 cells (A) or HeLa cells (B) starved of either leucine (Leu), arginine (Arg), or glucose (Glc) and then reactivated with these substances for 30 min. P, phosphorylated. (C) IL-6 enzyme-linked immunosorbent assay (ELISA) from Δp62 HeLa cells expressing the indicated p62 variants 20 hours after plating. Means ± SEM are plotted; Δp62 compared to +WT (n = 33 independent experiments), +TRM (n = 17 independent experiments), +iso2 (n = 14 independent experiments), or yellow fluorescent protein (YFP) (n = 9 independent experiments). (D) Immunoblots from Δp62 or Δp62 HeLa cells expressing the indicated p62 variants that were leucine-starved for 4 hours and left untreated or reactivated with leucine for 30 min. Intervening irrelevant lanes were removed, and pictures from the same blot developed with antibodies to p62 or β-actin (below) are shown. (E) Immunoblots from C12 cells transfected with nontargeting (CTRL) or caspase-8 siRNA for 72 hours and then left for 1 hour in serum-free RPMI with or without (w/o) leucine. (F) Immunoblots from C12 cells transfected with the indicated siRNA for 72 hours, then starved for either leucine or glucose, and left untreated (−) or treated with leucine or glucose, respectively, for 30 min. ***P ≤ 0.005, ****P < 0.0001 by paired Student’s t test. Data in (A), (B), and (D) to (F) represent n = 3 biologically independent experiments.

Could p62TRM support leucine sensing and mTORC1 activation? Reconstitution of Δp62 HeLa cells with either WT p62 or p62TRM restored mTORC1 activation by leucine and corrected defects in basal mTORC1 activity as measured by p70S6K1 phosphorylation, MYC levels, IL-6 production, and cell proliferation (Fig. 4, C and D, and figs. S4, E and F, and S5H). As a negative control, expression of YFP in Δp62 cells did not correct these defects, ruling out nonspecific effects of stable protein expression (Fig. 4C and fig. S4G). Therefore, as compared to full-length p62, the shorter p62TRM protein was sufficient for basal activity of mTORC1. Consistent with this conclusion, we found that phosphorylation of p70S6K1 during poly(I:C) treatment correlated with p62TRM production (fig. S4H).

Our findings implied that full-length p62 and p62TRM had mutually exclusive roles in autophagy (Fig. 3, B to F) and mTORC1 activation. To independently validate this idea, we reconstituted Δp62 cells with the autophagy-deficient p62 isoform 2 (p62iso2; which lacks the N-terminal PB1 domain; fig. S1A), which cannot oligomerize into filaments (40, 41). p62iso2, which was also cleaved upon leucine starvation, restored basal mTORC1 functions and leucine sensing in Δp62 cells (Fig. 4, C and D, and fig. S4F), which further pointed to the functional distinction between p62 proteins produced through splicing and/or posttranslational processing. Furthermore, the cargo receptor NBR1, which can cooperate with p62 (42, 43), interacted with full-length p62 and p62TRM, but not p62iso2 (fig. S4I). These findings ruled out a role for p62-NBR1 interactions in mTORC1 regulation.

We hypothesized that conversion of a portion of p62 to p62TRM during leucine starvation could help retain an autophagy-resistant fraction for mTORC1 regulation. In support of this hypothesis, leucine starvation stimulated p62 conversion to p62TRM in a manner dependent on caspase-8, thereby implicating this protease in mTORC1 activation (Fig. 4E). RNAi silencing of caspase-8 attenuated mTORC1 reactivation by leucine, but not by glucose, which confirmed the link between p62 and caspase-8 in relaying leucine sensing to mTORC1 (Fig. 4F). Caspase-8 silencing also increased the abundance of the lipidated form of LC3B (LC3BII), which indicated increased autophagic activity in cells (fig. S5A). Furthermore, treatment with Ac-IETD-fmk reduced the levels of phosphorylated p70S6K1 and increased basal IL-6 production, thereby mimicking Torin 1 treatment or p62 loss of function and further implicating caspase-8 in mTORC1 activation (fig. S5B). In addition, during TLR3 signaling, caspase-8 silencing reduced the phosphorylation of p70S6K1, which correlated with loss of p62TRM production (fig. S5C). Thus, we concluded that caspase-8–directed p62TRM production promotes basal mTORC1 activity and leucine sensing, and autophagy and mTORC1 regulation are carried out by distinct p62 protein species.

Natural D329G and D329H polymorphisms in p62 abrogate cleavage by caspase-8 and leucine sensing by mTORC1

Our identification of a potential link to inflammation through mTORC1 inactivation led us to ask whether SQSTM1 polymorphisms could disrupt mTORC1 function. A D329G polymorphism in SQSTM1 (rs148294622, 5:179260603 A/G, frequency < 1/10,000) has been reported in a Spanish male with behavioral variant FTD, a form of FTD that results in the deterioration in socially appropriate behavior (44); however, its functional effect was not biochemically characterized. We independently found a rare D329H polymorphism (5:179260602 G/C, singleton) in the Genome Aggregation Database (gnomAD) of genome and exome sequences from >130,000 individuals (45). We predicted that both nonsynonymous changes at Asp329 should ablate caspase-8–mediated p62 processing but would not markedly affect autophagy functions. FTD-linked p62 D329G and the D329H variant were resistant to caspase-8–mediated cleavage in cells and in vitro (Fig. 5, A and B). Autophagy assays revealed that the p62 D329G and D329H variants interacted with LC3B were normally turned over by autophagy, formed ubiquitin aggregates (fig. S5, D to F), trafficked to L. monocytogenes ΔactA, and restricted their intracellular replication through xenophagy (Fig. 5, C and D). These results indicate that the p62 D329G and D329H variants are as capable of autophagy functions as WT p62. In contrast, neither variant supported leucine sensing nor corrected mTORC1 defects as shown by reduced cell proliferation, reduced phosphorylation of p70S6K1, reduced MYC abundance, and increased IL-6 release (Fig. 5, E and F, and figs. S4E and S5, G and H); as expected, glucose sensing was not affected (fig. S5I). Thus, a single amino acid change in p62 attenuated the mTORC1–MYC–IL-6 nutrient-sensing axis without affecting autophagy functions. This highlighted not only the exquisitely modular nature of the p62 scaffold but also the involvement of full-length p62 and p62TRM in physiologically opposing pathways.

Fig. 5 Natural D329H and D329G mutations in p62 abolish caspase-8–mediated processing and mTORC1 activation.

(A) Immunoblots from C12 cells transfected with the indicated Flag-tagged p62 variants and left untreated or treated with poly(I:C). (B) Coomassie-stained gel from assay with recombinant GST-p62297–440 variants and caspase-8. Also see schematic in Fig. 1F. (C) Representative images from immunofluorescence staining of Flagp62 variants and L. monocytogenes ΔactA in Δp62 HeLa cells stably expressing the indicated Flag-p62 proteins 3 hours after infection and quantification of p62-positive bacteria on the right (means ± SEM; n = 3 independent experiments). (D) Fold replication of L. monocytogenes ΔactA between 2 and 10 hours after infection in indicated ΔHeLa cells. Matched means are shown (n = 5 independent experiments). (E) ELISA quantification of IL-6 secreted by Δp62 HeLa cells expressing the indicated p62 variants. Means ± SEM are plotted; Δp62 compared to +WT (n = 33 independent experiments), +D329H (n = 16 independent experiments), or +D329G (n = 27 independent experiments). (F) Immunoblots from indicated Δp62 HeLa cells leucine-starved for 4 hours and then left untreated or reactivated with leucine (Leu) for 30 min. ***P ≤ 0.005 by one-way analysis of variance (ANOVA) (C) or paired Student’s t test (D and E). Data represent n = 3 (A and B) or n = 4 (F) biologically independent experiments.

Synthetic cleavage of p62 by TEV bypasses the requirement for caspase-8

Caspase-8 has many roles in cells in cell death and signaling; we therefore asked whether it contributed to mTORC1 regulation in addition to controlling p62 proteolysis. To address this question, we uncoupled p62TRM generation from caspase-8 by replacing the caspase-8 recognition region in p62 with the tobacco etch virus (TEV) protease site to generate p62TEV (Fig. 6A). p62TEV was not cleaved by caspase-8 in HEK cells or by poly(I:C) when stably expressed in Δp62 HeLa cells (fig. S6, A and B). We expressed the TEV protease using a doxycycline-inducible promoter, which led to robust p62TEV cleavage into p62TRM upon doxycycline treatment (Fig. 6B and fig. S6, A to C). As expected, in nutrient-deprived cells, mTORC1 was reactivated by leucine only upon doxycycline treatment, which resulted in the generation of p62TRM (Fig. 6C), whereas mTORC1 reactivation by glucose, which is p62 independent, was similar without or with doxycycline treatment (Fig. 6C). In doxycycline-treated cells, silencing of caspase-8 or treatment with a pan-caspase inhibitor did not impair mTORC1 activation by leucine (Fig. 6D and fig. S6C). We therefore concluded that caspase-8 can be dispensable if p62TRM can be generated through a different proteolytic event, thereby underscoring the link between mTORC1 signaling and caspase-8.

Fig. 6 Synthetic generation of p62TRM by TEV protease–mediated cleavage bypasses the requirement for caspase-8.

(A) Schematic of the strategy for synthetic generation of p62TRM. The Asp329 region in p62 was replaced with a TEV recognition sequence to generate the Flagp62TEV plasmid for stable constitutive expression in Δp62 HeLa cells. Doxycycline (DOX) was used to control expression of TEV-T2A-GFP (green fluorescent protein); T2A is a self-cleaving peptide sequence. (B) Immunoblots showing doxycycline-inducible expression of GFP and cleavage of p62TEV into p62TRM in Δp62/Flagp62TEV+TEV-T2A-GFP cells. (C) Representative immunoblots from Δp62/p62TEV+TEV-T2A-GFP cells either leucine-starved for 4 hours or glucose-starved for 6 hours, followed by reactivation with Leu or Glc for 30 min as indicated. Cells were either untreated (w/o) or treated with doxycycline for 24 hours and recovered without doxycycline for 18 hours before starvation/reactivation treatments. (D) Δp62/p62TEV/TEV-T2A-GFP cells were transfected with nontargeting (CTRL) or caspase-8 siRNA for 72 hours and treated with doxycycline before leucine starvation/reactivation as in (C). Graph below shows ratio of P-p70-S6K1/p70-S6K1 after leucine reactivation relative to CTRL siRNA-transfected cells. Means ± SEM (n = 3 independent experiments) are plotted. ns, not significant, by paired Student’s t test. Data in (B) and (C) represent n = 3 biologically independent experiments.

RIPK1-p62 interaction promotes p62TRM generation by caspase-8

We next used biochemical approaches to investigate the molecular basis of p62TRM generation in cells. We noted that transfection of HEK293E cells with TRIF triggered caspase-8–dependent cleavage of endogenous and transfected p62, but not of the p62 D329A, D329G, or D329H variants (fig. S7, A and B). RIPK1 silencing blocked TRIF transfection-induced p62TRM production, validating its role downstream of TRIF in our assays (fig. S7C). We reconstituted HEK293p62miR cells with truncation and deletion mutants of p62, including those lacking the zinc finger (ZZ) domain that contributes to the p62-RIPK1 interaction (fig. S7D). p62ΔUBA and p62ΔZZ were processed less efficiently than WT p62, whereas the double-deletion p62ΔZZ/UBA mutant was not proteolyzed (fig. S7, D and E). Inactivation or deletion of the PB1 domain (p62iso2), the TRAF6-binding site, the LIR, or KEAP-interacting region (KIR) did not affect p62 proteolysis (fig. S7, D and E). Because p62 is a RIPK1-binding partner, we postulated that this interaction was required for its proteolytic processing. Co-IP experiments revealed a direct correlation between RIPK1 binding and proteolysis, in that RIPK1 did not interact with the cleavage-resistant p62ΔZZ/UBA but did interact with p62 and the mutants that were processed by caspase-8 (Fig. 7A and fig. S7, E and F). This finding suggested that the ZZ and UBA domains promote bipartite RIPK1-p62 interactions and facilitate p62 processing.

Fig. 7 RIPK1-dependent p62TRM generation and leucine sensing through mTORC1.

(A) Immunoblots from IP experiments in HEK293p62miR cells transfected with Flagp62 variants and YFP or MycRIPK1 as indicated. (B) ELISA quantification of IL-6 secreted by Δp62 HeLa cells stably reconstituted with the indicated Flagp62 variants. +D329G was compared to +WT (n = 22 independent experiments), +ΔZZ/UBA (n = 17 independent experiments), or +D329GΔUBA (n = 13 independent experiments). (C) Immunoblots from Δp62 HeLa cells expressing the indicated p62 constructs leucine-starved for 4 hours and then left untreated or treated with leucine for 30 min. (D) IL-6 ELISA from C12 cells stably expressing nontargeting (CTRL) or RIPK1 miR 20 hours after plating. n = 9 independent experiments. (E) Immunoblots from C12 cells transfected with nontargeting (CTRL) or RIPK1 siRNA for 72 hours and then treated with serum-free RPMI containing the indicated concentrations of leucine (Leu) for 1 hour. (F) Immunoblots from C12 cells stably expressing nontargeting (CTRL) or RIPK1 miRNA starved for either leucine or glucose for 1 hour and then untreated (−) or treated with leucine or glucose, respectively, for 30 min. (G) Model showing RIPK1–caspase-8–driven cleavage (showed by green scissor) that generates p62TRM. Full-length p62 and D329H/G variants are competent in autophagy and antimicrobial xenophagy. p62TRM generated from p62 (or p62iso2; not shown) can regulate mTORC1. Natural D329H/G mutants cannot be processed by caspase-8 or activate mTORC1. RIPK1–caspase-8 contributes to regulation of mTORC1 through p62TRM production. Means ± SEM are plotted in graphs. ***P < 0.005, ****P < 0.0001 by paired Student’s t tests. Data represent n = 3 (A, C, and E) or n = 4 (F) biologically independent experiments.

Removal of the UBA domain restores the ability of cleavage-resistant p62 D329G to activate mTORC1

We next addressed why p62 processing was required for mTORC1 activity. Unlike the cleavage-resistant D329G or D329H variants, the expression of the noncleavable p62ΔZZ/UBA variant corrected the aberrant IL-6 production, MYC levels, and leucine sensing by mTORC1 in Δp62 cells (Fig. 7, B and C, and fig. S8A). Both p62TRM and p62ΔZZ/UBA proteins lack the UBA domain (residues 391 to 440; fig. S1A). We reasoned that removal of the UBA domain might promote mTORC1 regulatory functions of p62. Deletion of the UBA in the FTD-associated p62 D329G variant restored leucine sensing and basal mTORC1–MYC–IL-6 signaling (Fig. 7, B and C, and fig. S8A). We therefore concluded that the loss of the UBA domain promoted optimal mTORC1 activation.

RIPK1 is required for p62TRM production during leucine starvation and promotes leucine sensing by mTORC1

Our results suggested that like caspase-8, RIPK1 was involved in mTORC1 activation. RIPK1 silencing impaired mTORC1 functions as seen from reduced phosphorylation of p70S6K1 and MYC abundance and increased IL-6 secretion (Fig. 7D and fig. S8, B and C). Moreover, p62 cleavage into p62TRM during leucine starvation was RIPK1 dependent (Fig. 7E). Similarly, RIPK1 knockdown blocked mTORC1 activation by leucine but not by glucose (Fig. 7F). These results establish RIPK1 and caspase-8 as key activators of p62TRM production and cellular homeostasis through the mTORC1 signaling hub.


Context-specific regulation of ubiquitous signaling proteins underlies their functional diversification. Posttranslation regulation of p62 by ubiquitylation, phosphorylation, and differential splicing enables functional separation and fine-tuning (1). For example, p62iso2 that lacks the PB1 region does not participate in autophagy; phosphorylation of the UBA domain by TBK1 increases ubiquitin binding; and cargo receptor functions and ubiquitylation regulate redox homeostasis and autophagy (4649). Similarly, during antimicrobial immunity, trafficking of p62 to microbial vesicles is tightly regulated. For instance, p62 trafficking to the bacterial pathogens Salmonella and Listeria can occur in naïve cells, whereas trafficking to the parasitophorous vacuoles of Toxoplasma requires IFN-γ treatment of cells (4, 50). We established that p62TRM was generated during TLR3 signaling and leucine starvation in a manner dependent on RIPK1 and caspase-8 and independent of cell death. We thus add RIPK1–caspase-8–dependent proteolysis as a regulatory mechanism that abrogates the autophagy functions of p62 and allows p62TRM to participate in the mTORC1 pathway.

Caspases not only regulate cell death but also contribute to immunity and homeostasis, for example, by controlling NF-κB activation and inflammation, vesicle trafficking, keratinocyte differentiation, and dendritic cell maturation (51, 52). We have reported that caspase-1 targets the conjugating enzyme UBE2L3, which regulates pro–IL-1β production by inflammasomes (53). Here, we identified and characterized p62 as a physiological substrate of caspase-8 in intact cells. TLR3-induced apoptosis typically requires cotreatment with SMAC mimetics, and necroptosis requires pan-caspase inhibitors (20, 54). The catalytic activity and substrate selectivity of caspase-8 can be modified by the regulatory protein cFLIP (55). cFLIP suppresses cell death during TLR4-TRIF signaling (56) and might play a similar role downstream of TLR3-TRIF. RIPK1 kinase activity was dispensable for p62TRM generation, which further supported the uncoupling of this process from necroptosis.

In addition to serving as an essential amino acid precursor, leucine has important signaling roles in regulating mTORC1 and anabolism, glutaminolysis, insulin secretion, and food intake (5759). We propose that the regulated proteolysis of p62 enables homeostasis in cells. Reduced leucine availability, for example, 85 μM or lower (25% of its concentration in RPMI medium), induced p62 proteolysis into p62TRM. Cells expressing noncleavable p62D392G or D329H showed impaired mTORC1 function comparable to Δp62 cells, which indicated that p62 processing under basal conditions is critical. Further, cells that cannot generate p62TRM, for example, due to silencing of RIPK1 or caspase-8 or treatment with caspase inhibitors, also displayed deficiencies in mTORC1–MYC–IL-6 signaling. Thus, p62TRM helps cells cope with reducing leucine availability over time and prevents spontaneous IL-6 production.

mTORC1 activation by p62 required the removal of the C-terminal UBA domain from p62. We propose that the UBA domain has autoinhibitory activity, which prevents p62 from participating in mTORC1 activation, and that RIPK1–caspase-8 relieves this block by proteolytically trimming p62. Because p62330–440 that lacks the C terminus was rapidly degraded, these regions of the protein were not required for p62TRM function. Further, the requirement for RIPK1–caspase-8 for mTORC1 activation could be bypassed by the p62ΔZZ/ΔUBA variant, which suggested that the main role for the RIPK1-p62 interaction was to promote p62 cleavage by caspase-8. This notion was also supported by the p62TEV-based synthetic strategy, which revealed that caspase-8 was dispensable as long as p62TRM was present. Our finding that RIPK1 was involved in leucine sensing and mTORC1 activity is consistent with previous work that identified RIPK1 as a regulator of autophagy (60, 61). Further studies should focus on how the leucine sensors of the Sestrin family, the lysosomal amino acid transporter SLC38A9, or leucyl–transfer RNA synthetase may regulate p62TRM generation (57). Prolonged mTORC1 activation in cancer cells mediated by glutaminolysis (starvation and then activation with both leucine and glutamine for 72 hours) triggers p62- and caspase-8–dependent apoptosis (62). In light of our findings, it is plausible that long treatments with leucine and glutamine stimulate alternative RIPK1–caspase-8 signaling, leading to cell death.

Our work also offers clues on how p62 can be co-opted for autophagy and mTORC1 in the same cells. Although p62 levels can drop upon starvation-induced autophagy, transcriptional increases and proteasomal turnover also maintain p62 levels (63). The transcription of genes encoding p62 and core autophagy factors such as ATG10 and ATG12 increases upon leucine starvation (64). We observed that leucine starvation did not reduce p62 abundance and a fraction was processed into autophagy-resistant p62TRM. Leucine starvation–induced p62 proteolysis would allow cells to be ready for mTORC1 reactivation when leucine becomes available. A model emerges in which p62 has mutually exclusive and opposing tasks depending on its processing by RIPK1–caspase-8 (Fig. 7G). Full-length p62 and the D329G or D329H variants execute autophagy and antibacterial xenophagy, whereas p62TRM is involved in leucine sensing by mTORC1. On the basis of our work, the FTD-linked p62 D329G and the rare p62 D329H mutations specifically affect leucine sensing and not autophagy. Full-length p62 undergoes phase separation in a PB1- and UBA-dependent manner upon binding polyubiquitin chains (65, 66). Phase separation and crowding in the cytosol can be regulated by mTORC1 (67). Future studies should investigate the contributions of p62TRM and p62iso2, which are unlikely to undergo phase separation themselves, in these processes.

In addition to human p62, the Asp329 site is conserved in primate, bovine, equine, and porcine p62, which are therefore likely to be cleaved by caspase-8 (fig. S2C). However, among rodents, we noticed that although Asp329 is conserved in rat p62, there is a glycine at this position in mouse p62 in almost all mouse genomes we analyzed (mouse genomes available through Wellcome Trust Sanger Institute). Therefore, we could not use mouse cells or knockout embryonic fibroblasts for our experiments.

IL-6 has pro- and anti-inflammatory effects, and its therapeutic neutralization is effective for the treatment of some inflammatory conditions (68). Increased IL-6 has been reported in patients with SQSTM1-linked conditions (6972). Our results suggest that cell-intrinsic loss in p62TRM-mTORC1 function could contribute to greater IL-6 production. Natural mutations in SQSTM1 often have paradoxical effects on cells, and our studies clarify how two p62 polymorphisms specifically compromise p62TRM-dependent functions. Whether other p62 polymorphisms or genes linked to FTD or ALS also affect mTORC1 warrants future investigation. In summary, p62TRM is the functional link between leucine sensing, RIPK1–caspase-8, and mTORC1 with important implications for immune homeostasis, antimicrobial responses, and hereditary human diseases.


Cell culture, treatments, and immunoblotting

Human donor–derived skin fibroblasts were obtained with informed consent, approved by ethics committee, and have been described previously (3237). Antibodies and key reagents used in the study are listed in tables S1 and S2. Cell lines (healthy donor–derived control, which we call C12 cells here, HeLa, HEK293E, HEK293T, and L929) were grown in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) plus penicillin and streptomycin, sodium pyruvate, and 10% heat-inactivated fetal bovine serum (all from Sigma-Aldrich). Retrovirally and lentivirally transduced cell lines were cultured in the same medium plus puromycin (2 μg/ml; Sigma-Aldrich). All cell lines were maintained at 37°C and 5% CO2 and were tested to be mycoplasma-negative (LookOut Mycoplasma PCR Detection Kit, Sigma-Aldrich).

Unless otherwise specified, cells were treated with poly(I:C) (25 μg/ml), PAM2CSK4 (5 μg/ml), or ultrapure Escherichia coli O111:B4 lipopolysaccharide (5 μg/ml) for 20 hours or as indicated. The following inhibitors were used: Ac-YVAD-fmk (50 μM), Ac-IETD-fmk (50 μM), Baf-A1 (100 nM), epoxomicin (5 μM), necrostatin-1 (25 μM), Torin 1 (250 nM), and z-VAD-fmk (50 μM). Cells were passaged by trypsinization and usually cultured for ~4 to 10 weeks for experiments. HeLa cells were plated at a density of 7.5 × 104 per well in 96-well plates for ELISA and 2 × 105 per well in 24-well plates for immunoblots. C12 cells were plated at a density of 1.75 × 104 per well in 96-well plates for ELISA, LDH, and PI assays in phenol red–free media and 7 × 104 per well in 24-well plates for immunoblots.

For starvation of all amino acids, cells were washed with warm phosphate-buffered saline (PBS) before incubation in serum- and amino acid–free Earle’s balanced salt solution (Gibco) for 45 min (skin fibroblasts) or 2 hours (HeLa) and stimulated for 15 min by addition of 1× amino acid mixture containing all amino acids at a final concentration that matched RPMI (from a 50× solution plus glutamine, both from Sigma-Aldrich). For single amino acid starvation and restimulation experiments, cells were washed with PBS and incubated in serum-free RPMI lacking leucine or arginine for 60 min (skin fibroblasts) or 4 hours (HeLa) and stimulated for 30 min by adding leucine or arginine, respectively. After stimulation, the final concentration of amino acids in the media was the same as in RPMI. For glucose starvation and restimulation, cells were starved of glucose for 80 min (skin fibroblasts) or 6 hours (HeLa) and stimulated with 25 mM glucose for 30 min. For starvation experiments in Δp62/Flagp62TEV+TEV-T2A-GFP cells, doxycycline was added for 24 hours (1 μg/ml), and then cells were left to recover in the absence of doxycycline overnight. Starvation experiments were performed the following day as described above. Where indicated, Ac-IETD-fmk (50 μM) was added 30 min before starvation and replenished during starvation. Immunoblotting, ELISAs, and cell death assays were described before (53). Staurosporine (1 μM) was used for 6 hours and actinomycin D (2 μM) for 20 hours. Live time-lapse images were taken using a Cytation 1 cell imaging multi-mode reader (BioTek Instruments). Necroptotic cell death was induced by treatment with TNF (10 ng/ml) combined with cycloheximide (5 μg/ml), birinapant (100 nM), and zVAD-fmk (50 μM) for 20 hours. When used, necrostatin-1 (25 μM) was added at the same time.

Cell lysates were prepared in radioimmunoprecipitation assay buffer [120 mM tris (pH 8.0), 300 mM NaCl, 2% NP-40, 1% Na-deoxycholate, 2 mM EDTA] supplemented with complete protease inhibitor cocktail, 1 mM phenylmethylsulfonyl fluoride (PMSF), and phosphatase inhibitor tablets and then mixed with Laemmli buffer containing 5% 2-mercaptoethanol. Proteins were separated by SDS–polyacrylamide gel electrophoresis (PAGE) using tris-glycine buffer systems and transferred to polyvinylidene difluoride membranes (Bio-Rad Laboratories) through semidry transfer. For p62 immunoblots, typically, the primary antibody was incubated in 2% fat-free milk in PBS–Tween 20 (PBST) (4°C overnight) and secondary antibody in 5% fat-free milk in PBST (1 hour at room temperature). Immunoblots were routinely developed with Clarity Western enhanced chemiluminescence (ECL), and ECL prime substrate was used for the detection of poorly expressed proteins or cleaved caspase subunits. Immunoblot quantification used images from independent experiments acquired on a ChemiDoc MP (Bio-Rad) and analyzed using Image Lab software (Bio-Rad Laboratories).

For IL-6 ELISA from resting cells, supernatants were collected 20 hours after plating. When used for 72 hours, Torin 1 and Ac-IETD-fmk were replenished daily. Human IL-6 ELISA kit was from eBioscience (#88-7066).

Cytotoxicity assays and ELISA

Culture supernatants were recovered for LDH assay (CytoTox 96 LDH kit, Promega) and/or ELISA, and PBS containing PI (5 μg/ml; Sigma-Aldrich) was added to cells. LDH assays used untreated cells (0%) and 1% Triton X-100 in PBS (100%) to calculate the percentage of LDH released. Similarly, untreated (0%) and 0.05% Triton X-100 (100%)–treated cells were used to calculate the percentage of PI uptake as described previously (53).

Luminescence assays

Caspase-Glo 8 assays were performed according to the manufacturer’s instructions. Briefly, 1.5 × 104 cells per well (skin fibroblasts) were plated in 96-well plates. Cells were treated as described in the figure legends for indicated time periods. To measure caspase activity, 50 μl of Caspase-Glo reagent was added to each well for 15 min to ensure cell lysis, and the lysates were then transferred to a white opaque plate. Lysates were incubated for 1.5 hours with constant shaking at room temperature, which was in the linear range of reaction kinetics and did not lead to substrate depletion based on pilot experiments. Luminescence was measured using a Cytation 1 multi-mode plate reader (BioTek).

Molecular cloning and miRNA-based stable silencing

Routine cloning used sequence- and ligation-independent cloning (73). The pcDNA-p62 plasmid (gift from R. Layfield, University of Nottingham) was used for polymerase chain reaction (PCR) (Phusion polymerase, New England Biolabs) to generate the retroviral pMXsIP-p62 plasmid. The cytomegalovirus (CMV) promoter from pEYFP-C1 (Clontech), 3×Flag at N terminus (from p3Tag1 vector, Agilent), and hemagglutinin (HA)–tag at C terminus (YPYDVPDYA introduced by PCR) were added to generate the pMXCMV-flag-p62-HA plasmid. However, as we and others have noted the cleavage of HA-tag by caspases (53, 74), anti-HA Western blots were not used in this study to detect Flagp62HA. The pIRES-Puro2 caspase-8 plasmid was a gift from R. Eils, German Cancer Research Centre (75). pGEX-4T1-p62 (gift from R. Layfield) was used to generate pGEX-6P1-p62297–440. Site-directed mutagenesis used single oligonucleotide-based linear PCR to generate Asp329 (76). Various p62 truncation variants were generated by PCR (regions shown in fig. S1A). The LIR mutant was p62D337A/W338A/L341A, and the KIR mutant was p62T350A. The Myc-RIPK1 was a gift from X. Lin (Addgene plasmid #44159). pMXCMV-YFP-stop, where YFP expression is driven by a CMV promoter (from pEYFP-C1, Clontech), was used as a control plasmid when carrying out retroviral transductions.

A TEV protease site was introduced into the pMXCMV-Flag-p62-HA plasmid by PCR to replace the caspase-8 site: 326MESD↓N330 was replaced by MENLYFQ↓GN (TEV recognition sequence in italics; ↓ indicates protease-cleavage site). An all-in-one lentiviral plasmid pLTREK3-TEV-T2A-GFP for doxycycline-inducible TEV-T2A-GFP and constitutive rtTA3 expression was generated as follows. The TRE-XTight promoter (from pRetroX-Tight-Puro; Clontech) was cloned into the pLentiV2 vector upstream of a TEV-T2A-GFP cassette; the self-cleaving T2A linker peptide generates separate TEV protease and GFP proteins. PGK promoter (from pRetroX-Tight-Puro) and rtTA3 (from pTripZ; Dharmacon) were cloned downstream of TEV-T2A-GFP. Stable pMXCMV-flag-p62TEV-HA transduction was achieved using puromycin selection, and flow sorting of GFP+ cells provided Flagp62TEV- and TEV-T2A-GFP–expressing Δp62 HeLa cells (Δp62/Flagp62TEV+TEV-T2A-GFP cells). All DNA fragments cloned by PCR were validated by sequencing (GATC Biotech).

Retroviral and lentiviral transduction

Virus-like particles were packaged in HEK293E or HEK293T cells using pCMV-MMLV-Gag-Pol or p1266 lentiviral packaging plasmids and pseudotyped with pCMV-VSV-G (gifts from P. Uchil and W. Mothes, Yale University). Packaging and transductions were performed as described previously (53). Puromycin (2 μg/ml; Sigma-Aldrich) was added 48 hours after transduction and replenished until stable pools were obtained (1 to 2 weeks).

RNA interference

For stable gene silencing, 22 base oligonucleotides (first base mismatch plus 21-mer sense and 22-mer antisense without mismatches) were cloned in the optimized miR backbone (77, 78) in pMXCMV-YFP vectors as described previously (53). Antisense 22-mer sequences used were the following: p62 (targeting the 3′UTR), 5′-TTAACACAACTATGAGACAGAA-3′ (from TRCN0000430110); RIPK1, 5′-TTATCCGTCAGACTAGTGGTAT-3′; ATG7, 5′-ATGGAGAGCTCCTCAGCAGGCG -3′ (from TRCN0000007584); nontargeting control LacZ, 5′-TCACGACGTTGTAATACGACGT-3′ (TRCN0000072226).

Transient RNAi used ON-TARGETplus siRNA reagents from Dharmacon. Nontargeting control (Dharmacon) served as negative control. siRNA pools were transfected in cells using Viromer BLUE (Lipocalyx) or TransIT-X2 reagent (Mirus) following the manufacturer’s protocol. Briefly, C12 and HeLa cells were seeded the day before transfection at 2 × 104 per well and 4 × 104 per well, respectively, in 24-well plates. For C12 cells, 50 nM siRNA with 0.5 μl of Viromer was used per well, and 25 nM siRNA and 0.25 μl of Viromer were used for HeLa and HEK293T cells. When using the TransIT-X2 reagent, 20 nM siRNA and 1.5 μl of reagent were used per well. Media were changed 48 hours after transfection, and experiments were performed the following day. Silencing efficiency was assessed by immunoblotting. siRNA sequences are presented in table S3.

Recombinant protein production and caspase-8 assays

Human p62 (amino acids 297 to 440) was cloned in pGEX6P (GST tag), and human caspase-8 p30 (amino acids 216 to 479) was cloned in pProExHT (N-terminal 6×His tag). GST-p62 proteins were purified using standard procedures. Briefly, a 3-hour induction with 100 μM isopropyl-β-d-thiogalactopyranoside (IPTG) at 37°C was followed by sonication in lysis buffer [50 mM tris-HCl (pH 8.0), 100 mM NaCl, 1 mM dithiothreitol (DTT), 2 mM EDTA, 0.1% Triton X-100, and 2 mM PMSF] at 4°C. Lysates were centrifuged (25,000g for 25 min at 4°C), and supernatants were bound to glutathione sepharose beads (GE Healthcare) for 1 hour at 4°C. Beads were then washed with lysis buffer without PMSF and wash buffer [50 mM tris-HCl (pH 8.0), 100 mM NaCl, 1 mM DTT, and 10% glycerol], and proteins were eluted with 100 mM tris-HCl (pH 8.0), 2 mM DTT, 10% glycerol, and 10 mM reduced glutathione. Proteins were desalted into 25 mM tris-HCl (pH 8.0), 100 mM NaCl, 1 mM DTT, and 10% glycerol on MidiTrap G-25 columns (GE Healthcare) and stored in aliquots at −80°C. His6-tagged caspase-8 p30 expression was induced overnight with 100 μM IPTG at 20°C. Ni-NTA beads (Qiagen) packed into columns (Bio-Rad) were used. Buffers were as follows: lysis [50 mM tris-HCl (pH 8.0), 300 mM NaCl, 5 mM 2-mercaptoethanol, 2 mM EDTA, 0.1% Triton X-100, 20 mM imidazole, and 2 mM PMSF], wash buffer [50 mM tris-HCl (pH 8.0), 300 mM NaCl, 5 mM 2-mercaptoethanol, 2 mM EDTA, 0.1% Triton X-100, and 50 mM imidazole], and elution [50 mM tris-HCl (pH 8.0), 300 mM NaCl, 5 mM 2-mercaptoethanol, 2 mM EDTA, 0.1% Triton X-100, and 150 mM imidazole]. Purified caspase-8 p30 was desalted and stored in 100 mM Hepes, 10% sucrose, 4 mM DTT, and 0.1% CHAPS.

Caspase-8 assays were carried out with 100 to 400 ng of caspase-8 and 3 μg of GST-p62297–440 (30 min, 25°C shaking at 300 rpm) in buffer containing 100 mM Hepes, 10% sucrose, 4 mM DTT, and 0.1% CHAPS. Assays were stopped with the addition of Laemmli buffer supplemented with 2-mercaptoethanol (final concentration, 5%) and were analyzed on SDS-PAGE gels stained with Coomassie Brilliant blue dye.

Reverse transcription and quantitative PCR

RNA was harvested according to the manufacturer’s protocol using the E.Z.N.A. Total RNA Kit I (Omega Bio-tek; 1 to 5 μg of RNA) and was reverse transcribed with random hexamer primers using the TaqMan Reverse Transcription Reagents (Thermo Fisher Scientific). Quantitative PCR was performed using SYBR Green PCR mix (Thermo Fisher Scientific) on a StepOnePlus Real-Time PCR System (Thermo Fisher Scientific). GAPDH was used as an internal control to calculate 2−ΔCt. The following primer pairs were used: hGAPDH, 5′-TGCCATCAATGACCCCTTC-3′ and 5′-CTGGAAGATGGTGATGGGATT-3′; hIL-6, 5′-ACTCACCTCTTCAGAACGAATTG-3′ and 5′-CCATCTTTGGAAGGTTCAGGTTG-3′.

Listeria infections

The L. monocytogenes ΔactA strain [10403s background; gift from D. Portnoy (79)] was grown overnight in BHI medium in a shaker at 37°C and washed three times in serum- and additive-free DMEM before use. Infections were performed at multiplicities of 5 to 10 bacteria per host cell and synchronized by centrifugation (750g for 10 min). Gentamicin (100 μg/ml) was added 1 hour after infection for 1 hour, after which cells were washed and incubated in complete DMEM containing gentamicin (20 μg/ml) for a further 8 hours. Wells were washed in DMEM and lysed in PBS containing 0.5% Triton X-100, and serial fivefold dilutions were plated (10 μl) on BHI agar plates to enumerate colony-forming units (CFU). Fold replication of bacteria between 10 and 2 hours after infection was calculated from technical triplicates.

Immunofluorescence analyses

Cells were plated on coverslips and treated as described in the figure legends, followed by fixing in 4% paraformaldehyde in PBS for 20 min at room temperature. Coverslips were washed three times with PBS, and 50 mM NH4Cl in PBS was added as a quenching solution for 10 min. After washing two times, cells were solubilized with PBS containing 0.1% saponin for 3 min. Cells were treated with PBS containing 5% bovine serum albumin (BSA) for 60 min, followed by staining with primary antibodies for 3 hours in PBS, 5% BSA, and 0.1% saponin. Mouse anti-FK2 (1:250), rabbit anti-p62 (1:500), and rabbit anti-Listeria (1:200) were used. Secondary donkey anti-rabbit antibody Alexa 647 and donkey anti-mouse antibody Alexa 594 were added after rinsing with PBS and incubated for 1 hour; nuclei were stained with Hoechst 33342 dye. Coverslips were mounted in ProLong Gold Antifade, and images were obtained on a Zeiss inverted microscope using a 100× oil immersion lens. Images were processed using ZEN software (Zeiss).


Cells were harvested in IP buffer [50 mM tris-HCl (pH 8.0), 1% NP-40, and 150 mM NaCl] supplemented with protease inhibitors, phosphatase inhibitors, 1 mM PMSF, 20 mM N-ethylmaleimide (Sigma-Aldrich), and 10 μM MG-132 (Sigma-Aldrich). Lysates were cleared by centrifugation (15,000g for 20 min at 4°C) and incubated for 3 hours or overnight at 4°C with indicated antibodies (1 to 2 μg each). Protein G–magnetic beads were added for 1 hour. Beads were washed five times in IP buffer and resuspended in Laemmli loading buffer for immunoblotting.

Generation of CRISPR-Cas9 knockout cells

To generate Δp62 HeLa cells, a guide DNA sequence targeting exon 1 of the human SQSTM1 gene (5′-GTCATCCTTCACGTAGGACA-3′) was cloned into the Bsm BI site of the LentiCrisprV2 plasmid, which was a gift from F. Zhang (Addgene plasmid #52961), to generate pLentiV2-p62-gRNA (guide RNA). HeLa cells were transiently transfected with the plasmid and selected with puromycin for 1 day. Clonal cells were isolated, and loss of p62 was verified by Western blot and phenotypic analysis. HeLa cells that underwent similar selection but were negative and therefore expressed WT p62 were used as negative controls. Sequencing of p62KO cells confirmed deletion and frameshift mutations that introduced stop codons within the gRNA region. Δp62 cells were stably transduced with retroviral plasmids expressing Flag-tagged p62 variants or YFP as indicated.

gnomAD dataset

The dataset of 123,136 exomes and 15,496 genomes (gnomAD release 27/02/2017) from unrelated individuals from various disease-specific and population genetic studies was used (45). The extremely rare occurrence of D329H suggests its association with disease because disease-specific polymorphisms and those associated with adult and/or late-onset diseases are present within the gnomAD dataset (45).

Cell proliferation

HeLa cells were seeded at 0.4 × 104 cells per well in a 48-well plate. Eight wells were seeded per cell line (two technical replicates × four time points). Viable cells were counted using trypan blue every 24 hours to obtain values for the mean number of cells per day per cell line.

Statistical analyses

No statistical methods were used to determine sample size. Experiments were not randomized, and investigators were allocated without blinding during experimentation and analyses. All experiments were repeated at least twice. For ELISA, quantitative reverse transcription PCR, LDH release, and PI uptake assays, two to three technical replicates were used to estimate experimental mean. Means from three or more biologically independent experiments, as indicated by the values of n in legends, were analyzed by statistical methods. For immunofluorescence assays, typically ~50 to 100 host cells or 100 to 200 L. monocytogenes were counted and mean percentage of cells showing events were obtained, and mean percentage from three biologically independent repeats were compared. Data were normally distributed (based on D’Agostino-Pearson or Shapiro-Wilk normality tests) after logarithm transformation and were then analyzed by statistical methods. Repeated-measures ANOVA was used to analyze means from immunofluorescence assays and immunoblot quantifications, and paired two-tailed Student’s t test was used elsewhere. When more than three comparisons were made, P values were adjusted for multiple comparisons by the Benjamini-Krieger-Yekutieli method (Q = 0.05, false discovery rate) implemented in GraphPad Prism 7. Discoveries at q < 0.05 are reported. Unless otherwise indicated, means ± SEM are plotted. L. monocytogenes ΔactA CFU assays with Δp62 cells were performed six times alongside +WT and +p62TRM cells, and +D329G and +D329H cell lines were also included in five experiments; these data are shown as matched means in Figs. 3F and 5D. In IL-6 ELISAs shown in Figs. 4C, 5E, and 7B, every experiment included Δp62 and typically also +WT cell line, alongside the indicated p62 variants; means ± SEM from all experiments are plotted, and data for Δp62 cells are shown in each figure for comparison. Data plots and statistics used Prism (version 7.04, GraphPad Software Inc.).


Fig. S1. TLR3 induces cell death–independent, caspase-8–dependent processing of p62 at Asp329.

Fig. S2. Caspase-8–dependent processing of p62 at Asp329.

Fig. S3. Role of RIPK1 in p62TRM production and the instability of the p62330–440 fragment.

Fig. S4. Characterization of p62TRM.

Fig. S5. Characterization of p62TRM, p62D329, p62D329H, and other variants.

Fig. S6. Synthetic cleavage of p62 using TEV.

Fig. S7. Role of RIPK1 and caspase-8 in mTORC1 regulation.

Fig. S8. Role of p62 variants and RIPK1 in regulating mTORC1.

Table S1. List of antibodies.

Table S2. List of key reagents and resources.

Table S3. siRNA sequences used in the study.


Acknowledgments: We would like to thank J. L. Casanova (The Rockefeller University) for sharing fibroblasts, M. Kalyan for technical help, F. Gordon (Statistics Advisory Service, Imperial College London) for advice on statistical methods, and the High-Throughput Single-Cell Analysis Facility at the MRC CMBI. Funding: This work was supported by the Wellcome Trust Seed Award 108246/Z/15/Z and Royal Society grant RG130811 to A.R.S. V.S.-S. would like to acknowledge support from the Medical Research Foundation. Author contributions: J.S.-G. designed and performed most experiments, analyzed data, and edited the manuscript. V.S.-S. performed experiments in Fig. 2A, analyzed data, and provided insight and important reagents. A.R.S. designed the study, performed experiments in Figs. 3E and 5D, analyzed data, wrote the paper, obtained funding, and supervised the project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. Plasmid DNA and derivatives of HeLa and HEK293 cell lines are available by material transfer agreement from Imperial College London for noncommercial use.
View Abstract

Navigate This Article