The HECT E3 ubiquitin ligase NEDD4 interacts with and ubiquitylates SQSTM1 for inclusion body autophagy

SQSTM1 (p62) is an autophagic cargo receptor that plays a key role in selective autophagy (Kirkin et al., 2009a; Rogov et al., 2014). Early studies have shown that SQSTM1 is associated with protein inclusions and aggregates, such as Mallory–Denk bodies and Lewy bodies (Stumptner et al., 1999, 2007; Nakaso et al., 2004; Tanji et al., 2015), and is considered to be a universal component of protein inclusions (Zatloukal et al., 2002). Further studies found that SQSTM1 functions as a receptor for ubiquitylated protein inclusion bodies or aggregates and recruits them to autophagosomes for degradation via interactions with LC3-II (Komatsu et al., 2007; Pankiv et al., 2007). Now, we know that SQSTM1 is a universal autophagic cargo receptor involved in multiple types of selective autophagy, such as mitophagy, pexophagy, xenophagy and aggrephagy (Zheng et al., 2009; Geisler et al., 2010; Bartlett et al., 2011; Ishimura et al., 2014; Zhang et al., 2015). As a key cargo receptor in selective autophagy of protein inclusions and aggregates, malfunction of SQSTM1 is associated with multiple diseases, such as Parkinson's disease, Huntington's disease, Alzheimer's disease, alcoholic hepatitis and cirrhosis (Kuusisto et al., 2002; Stumptner et al., 2002; Zatloukal et al., 2002; Nakaso et al., 2004; Du et al., 2009; Geisler et al., 2010; Cuyvers et al., 2015).

Ubiquitylation is an important biochemical process in SQSTM1-mediated selective autophagy. Multiple studies have shown that the ubiquitylation of protein inclusion bodies, aggregates or other autophagic cargos is pivotal for recognition by SQSTM1 in the autophagic degradation process (Kim et al., 2008; Johansen and Lamark, 2011; Rogov et al., 2014). In addition, SQSTM1 is capable of recruiting E3 ubiquitin ligases, such as TRAF6 and KEAP1, for ubiquitylating autophagic cargos or autophagic proteins during initiation, formation or transportation of selective autophagosomes (Kirkin et al., 2009a; Fan et al., 2010; Fusco et al., 2012; Isakson et al., 2013; Stolz et al., 2014). As an autophagic cargo receptor, SQSTM1 is transported along with the autophagic cargos in autophagosomes to lysosomes for degradation (Bjørkøy et al., 2005, 2006; Ichimura et al., 2008). Therefore, degradation of SQSTM1 sometimes is used as a molecular marker for activation of autophagy (Bjørkøy et al., 2009).

While the role of SQSTM1 in selective autophagy is well established, it remains poorly understood how the receptor activity of SQSTM1 is regulated during selective autophagy. A recent study found that casein kinase 2 (CK2) phosphorylates S403 in the Uba domain of SQSTM1 and enhances the binding capacity of SQSTM1 to the polyubiquitin chain (Matsumoto et al., 2011). This phosphorylation promotes SQSTM1 to target polyubiquitylated proteins and recruit ubiquitylated cargos to autophagosomes (Matsumoto et al., 2011). Recent studies indicate that ubiquitylation also regulates the autophagy receptor function of SQSTM1 for recognition of autophagic cargos. It has been found that SQSTM1 is ubiquitylated by the ring family E3 ubiquitin ligases TRIM21 (Pan et al., 2016), KEAP1–CULLIN3 (Lee et al., 2017), PARKIN (Song et al., 2016), and the E2 conjugating enzymes UBE2D2/3 (Peng et al., 2017). The ubiquitylation produces diversified effects on SQSTM function, including suppression and activation of the autophagic receptor activity (Pan et al., 2016; Lee et al., 2017; Peng et al., 2017) and promotion of the proteasomal degradation of SQSTM1 (Song et al., 2016). However, how ubiquitylation of SQSTM1 regulates cellular inclusion body autophagy remains unknown. Furthermore, SQSTM1 also participates in other cellular signaling pathways, such as atypical PKC and NF-κB signaling pathways (Puls et al., 1997; Sanchez et al., 1998; Sanz et al., 1999). Whether these pathways are regulated independently or are connected to autophagy has not been clarified.

Our recent studies found that NEDD4 (also known as NEDD4-1), a member of the HECT E3 ubiquitin ligase family, interacts with the autophagic protein LC3 through an LIR domain and is essential for starvation or rapamycin-induced activation of autophagy (Sun et al., 2017). Knockdown of NEDD4 by shRNA caused aggregation of GFP–LC3 puncta in the ER and deformation of mitochondria. It appears that interaction of NEDD4 with LC3 is not only necessary for association with autophagosomes, but also for activation of the E3 ubiquitin ligase. Our preliminary data also demonstrate that NEDD4 ubiquitylates SQSTM1, but not LC3 (Sun et al., 2017). These results clearly indicate that NEDD4 is an important E3 ubiquitin ligase involved in autophagic activation. In this report, we continue investigating the role of NEDD4 in autophagy by characterizing interaction and ubiquitylation of SQSTM1 and defining the function of SQSTM1 ubiquitylation. We found that NEDD4 interacts with SQSTM1 through the HECT (homologous to E6-AP carboxyl terminus) domain. The PB1 domain in SQSTM1 appears to be the NEDD4 interactive and ubiquitylating region. The polyubiquitylation of SQSTM1 by NEDD4 is mainly through K63 conjugation, which is important for the SQSTM1-mediated inclusion body autophagy, rather than proteasomal degradation. Our studies demonstrate NEDD4 as a key E3 ubiquitin ligase in selective autophagy that interacts with and ubiquitylates the autophagy receptor SQSTM1.

Our previous studies have shown that NEDD4 ubiquitylates SQSTM1 (Sun et al., 2017). To determine whether this ubiquitylation results from interaction between NEDD4 and SQSTM1, we characterized the binding of NEDD4 to SQSTM1 using a co-immunoprecipitation assay. As shown in Fig. 1A, NEDD4 is co-immunoprecipitated with SQSTM1 when both were co-expressed in cells, indicating that SQSTM1 binds to NEDD4. Interestingly, the ligase-dead (LD) mutant NEDD4-C867A bound to SQSTM1 with much higher affinity than wild-type NEDD4 (Fig. 1B). This suggests that the ligase-dead mutant of NEDD4 has a ‘trap’ effect on SQSTM1, which is a typical binding mode for a catalysis-defective enzyme with a substrate, as observed in binding of tyrosine phosphatase-defective mutants with their substrates (Flint et al., 1997).

To determine the SQSTM1-binding region in NEDD4 more specifically, we made a series of truncation mutants of NEDD4 (Fig. 1C), and tested the binding of these mutants to SQSTM1. HA-tagged NEDD4 or its truncation mutants were co-expressed with GFP-tagged SQSTM1 in HEK293 cells and binding was detected by co-immunoprecipitation assay. As shown in Fig. 1D,E, all the truncation mutants are capable of binding to SQSTM1. As the mutant NEDD4-N4Δ contains only the HECT domain (see Fig. 1C), this indicates that NEDD4 interacts with SQSTM1 through the HECT domain. To confirm this, we co-expressed the HECT domain-deletion mutant, NEDD4-HECTΔ, with SQSTM1 in HEK293 cells, and examined interaction of the mutant with SQSTM1. As shown in Fig. 1F, while the NEDD4 ligase-dead mutant NEDD4-C867A (NEDD4-LD) was co-immunoprecipitated with SQSTM1, NEDD4-HECTΔ showed little co-precipitation with SQSTM1, confirming that the HECT domain interacts with SQSTM1. We also examined the binding of NEDD4 to another autophagy receptor NBR1 using a co-immunoprecipitation assay, and found no detectable binding (Fig. 1G).

SQSTM1 is a key autophagic protein that interacts with LC3 and it plays an important role in selective autophagy, including mitophagy (Kirkin et al., 2009a; Johansen and Lamark, 2011; Stolz et al., 2014). Here, we characterized the ubiquitylation of SQSTM1 by NEDD4 using both immunoprecipitation and GST–Uba pulldown assays. GST–Uba pulldown assay has been successfully used for detection of ubiquitylated proteins in our previous studies (Lin et al., 2010). As shown in Fig. 2A, GFP-tagged SQSTM1 was ectopically expressed with or without NEDD4 in HEK293 cells. Without NEDD4, SQSTM1 had a low level of ubiquitination (lane 4). When co-expressed with NEDD4, SQSTM1 was heavily polyubiquitylated (lane 2). This result confirms that SQSTM1 is a ubiquitylation substrate of NEDD4. We further examined whether endogenous SQSTM1 is the substrate of NEDD4 upon activation of autophagy in the lung cancer cell line A549. To elevate the ubiquitylation of SQSTM1, we treated cells with rapamycin to activate autophagy, and with chloroquine to block the autophagic degradation of SQSTM1. As shown in Fig. 2B, upon treatment with rapamycin and chloroquine, endogenous SQSTM1 was significantly polyubiquitylated (lane 2), whereas in the NEDD4 shRNA cell line, SQSTM1 was no longer polyubiquitylated upon treatment with rapamycin and chloroquine, suggesting that endogenous SQSTM1 is ubiquitylated by NEDD4 in response to activation of autophagy.

The carboxyl terminus of SQSTM1 contains an Uba domain that can bind to other ubiquitylated proteins. To exclude the possibility that NEDD4-dependent ubiquitylation detected in SQSTM1 is from the SQSTM1 Uba domain-associated proteins, we made the Uba truncation mutant of SQSTM1, SQSTM1-UbaΔ. Co-expression of SQSTM1-UbaΔ with NEDD4 in HEK293 cells showed polyubiquitylation of SQSTM1-ΔUba (lane 2, Fig. 2C), confirming that the polyubiquitylation of SQSTM1 by NEDD4 is not from the SQSTM1 Uba-associated ubiquitylated proteins. In fact, immunoprecipitation of SQSTM1 or the Uba truncation mutant without NEDD4 co-expression showed no detectable polyubiquitylation (lane 4 in Fig. 2A and lane 3 in Fig. 2C), indicating that the SQSTM1 Uba-associated ubiquitin proteins (if any) produced little interference with NEDD4-mediated polyubiquitylation of SQSTM1.

We further determined the type of ubiquitin chain linkage of SQSTM1 catalyzed by NEDD4. As shown in Fig. 2D, NEDD4 catalyzed the K63-linked polyubiquitylation of SQSTM1 (top panel), not the K48 polyubiquitylation (second panel), suggesting that NEDD4 catalyzed polyubiquitylation of SQSTM1 may not be involved in proteasomal degradation. However, we detected a minor K63 and K48 polyubiquitylation of SQSTM1 with expression of the ligase-dead mutant NEDD4-LD or without exogenous NEDD4 (lanes 3 and 4, the top two panels) that might be produced by endogenous ubiquitylation.

To confirm that NEDD4-dependent polyubiquitylation of SQSTM1 does not lead to proteasomal degradation, we examined levels of ubiquitylation and SQSTM1 protein upon treating the cells with the specific proteasomal inhibitor bortezomib. As shown in Fig. 2E, treatment with bortezomib did not induce accumulation of either NEDD4-dependent ubiquitylation or protein of SQSTM1. In fact, bortezomib eliminated the NEDD4-dependent ubiquitylation of SQSTM1 (compare lane 7 with lane 8 in the right top panel in parallel with lane 2 and lane 3 in the left top panel). These data further suggest that NEDD4-dependent polyubiquitylation of SQSTM1 is involved in autophagy, not proteasomal degradation.

NBR1 is another autophagic cargo receptor that contains similar structural domains to SQSTM1 and functions in selective autophagy, particularly in recruiting ubiquitylated protein inclusions to autophagosomes (Kirkin et al., 2009b). We wondered whether NBR1 was ubiquitylated by NEDD4. As shown in Fig. 3A, while SQSTM1 was significantly ubiquitylated when co-expressed with NEDD4 (compare lane 2 with lane 4, second panel), NBR1 showed no increase in ubiquitylation when co-expressed with NEDD4 (compare lane 3 with lane 5, top panel), indicating that NBR1 was not ubiquitylated by NEDD4. Interestingly, endogenous NBR1 was heavily ubiquitylated independent of NEDD4 (top panel). We further confirmed that NBR1 showed NEDD4-independent ubiquitylation when co-expressed with NEDD4 mutants (Fig. 3B). NEDD4-independent ubiquitylation and NBR1 protein level were dramatically enhanced by treatment with proteasomal inhibitor MG-132, but not lysosomal inhibitor chloroquine (Fig. 3B–D), suggesting that NBR1 has an active turnover by proteasomal degradation through ubiquitylation by a non-NEDD4 E3 ubiquitin ligase. NEDD4 seems to antagonize the E3 ubiquitin ligase for NBR1, because co-expression with NEDD4 markedly reduced ubiquitylation of NBR1 (compare lane 9 with lane 8, top panel, Fig. 3B). In addition, co-expression with the N-terminal truncation mutant of NEDD4, NEDD4-N1Δ, which is defective in binding to LC3 (Sun et al., 2017), dramatically enhanced the amount of NBR1 protein through an unknown mechanism (see lane 6, top panel, Fig. 3C,D). These data indicate that NBR1 is not a ubiquitylation substrate of NEDD4.

To identify the region in SQSTM1 that interacts with and is ubiquitylated by NEDD4, we made a series of SQSTM1 mutants in its functional domains and tagged with GFP (Fig. 4A). These mutants include the PB1 deletion mutant N43Δ, the PB1 domain point mutants K7A, K13E, R21A/R22A (named as 2R2A), K13E/R21A/R22A (named as K13E/2R2A), D69A, the LIR domain point mutant L341A, the LIR deletion mutant [334–342]Δ, and the Uba domain point mutant L417V (Seibenhener et al., 2004). Ubiquitylation of the SQSTM1 domain mutants by NEDD4 was first examined by the GST–Uba pulldown assay (Fig. 4B). Deletion of PB1 (N43Δ) diminished the ubiquitylation of SQSTM1 (lane 5). Mutations in the LIR and the Uba domain produced an insignificant effect on the ubiquitylation. These results suggest that the PB1 domain is essential for ubiquitylation by NEDD4, while interaction with LC3 or ubiquitin is dispensable.

Previous studies have shown that the PB1 domain functions in homo- or hetero-dimerization of SQSTM1 through the interaction between the basic cluster and the OPCA (OPR–PC–AID) motif within the PB1 domain and is required for localization on autophagosomes (Lamark et al., 2003; Itakura and Mizushima, 2011). The results in Fig. 4B indicate that the PB1 domain is required for either binding to or ubiquitylation by NEDD4 or both. Thus, we used the PB1 truncation mutant N43Δ and the dimerization defective mutants K7A and 2R2A for testing the binding to NEDD4. As shown in Fig. 4C, the PB1 deletion mutant N43Δ or the point mutant 2R2A failed to co-immunoprecipitate NEDD4 (lanes 2 and 4), whereas wild-type SQSTM1 or the mutant K7A co-immunoprecipitated NEDD4 (lanes 1 and 3). These data demonstrate that PB1 is the NEDD4-interactive domain and that R21 and R22 in the PB1 domain are the residues essential for binding to NEDD4. In addition, the defect in dimerization in the 2R2A mutant is unlikely to be the cause of loss of NEDD4 binding, because K7A, which is also a dimerization defective mutant (Lamark et al., 2003), retains NEDD4 binding capacity (lane 3).

We further characterized NEDD4-dependent ubiquitylation of the PB1 mutants of SQSTM1 by both immunoprecipitation (Fig. 4D) and GST–Uba pulldown assays (Fig. 4E). The results from both assays were consistent, and showed that mutations on R21/R22 (lane 4 in Figs. 4D and 4E) and mutation on K7 (lane 3 in Fig. 4D, lane 5 in Fig. 4E) significantly reduced the ubiquitylation, whereas the mutation on K13 had no effect (lane 5 in Fig. 4D, lane 3 in Fig. 4E). The results indicate that the PB1 domain and the R21/R22 residues are essential for the binding to and ubiquitylation by NEDD4, and that K7 is one of the major NEDD4 ubiquitylation sites. It has been shown that the RING family E3 ubiquitin ligase TRIM21 also ubiquitylates K7 of SQSTM1, and the ubiquitylation impairs oligomerization of SQSTM1 thus suppressing the SQSTM1-mediated sequestration of KEAP1 (Pan et al., 2016). In future studies, it would be interesting to examine whether the ubiquitylation of K7 by NEDD4 has the same effect on SQSTM1 as that of TRIM21

We also examined whether the ubiquitylation affects binding of SQSTM1 to LC3 by co-expression of the SQSTM1 mutants with NEDD4. The GST–LC3 pulldown assay confirmed that SQSTM1-L341A or [334–342]Δ is defective in LC3 binding (lanes 6–10, Fig. 4F) and other mutants are capable of binding to LC3 (lanes 5,6 and 11–14, Fig. 4F). Co-expression with NEDD4 did not affect binding of SQSTM1 or the mutants to LC3, although the protein level of wild-type SQSTM1 and the PB1 domain truncation mutant was slightly reduced by co-expression with NEDD4 (lanes 4 and 6, second panel, Fig. 4D). This result suggests that ubiquitylation of SQSTM1 by NEDD4 is not involved in regulation of the LC3 binding, which is consistent with our previous studies with immunofluorescence staining (Sun et al., 2017).

Our recent studies have shown that knockdown of NEDD4 impairs rapamycin- and starvation-induced autophagy and autophagosomal biogenesis, and causes aggregation of GFP–LC3 puncta (Sun et al., 2017). Here, we further determined the effect of NEDD4 knockdown on the cellular localization and morphology of SQSTM1-positive fluorescent puncta in response to treatment with rapamycin. Similar to GFP–LC3, SQSTM1 was observed as tiny fluorescent puncta localized at para-nuclei in the vector control cells without rapamycin treatment, while in the NEDD4 knockdown cells, significant accumulation of heterogeneous large SQSTM1 puncta was seen (Fig. 5A). Quantification analysis indicates that the average size of the SQSTM1 puncta increased ∼4-fold up to ∼1 µm, upon knockdown of NEDD4, but the average number of SQSTM1 puncta per cell did not change significantly (Fig. 5B). Furthermore, SQSTM1 puncta in the NEDD4 knockdown cells were randomly distributed in the cells, no para-nuclear localization was observed (Fig. 5A). These large SQSTM1-positive puncta in NEDD4 knockdown cells are likely to be protein inclusion bodies, which are the common autophagic cargos associated with SQSTM1 (Zatloukal et al., 2002). Upon treatment with rapamycin for 18 h, SQSTM1 puncta in the vector control cells was distributed at one side of the nucleus (bottom left panel, Fig. 5A) and average numbers of the SQSTM1 puncta per cell increased significantly (Fig. 5B), indicating that biogenesis of the SQSTM1-positive autophagosomes was induced by rapamycin. In NEDD4 knockdown cells, large SQSTM1 puncta remained randomly distributed, and numbers of the SQSTM1 puncta did not change significantly upon rapamycin treatment (Fig. 5A,B). These results indicate that knockdown of NEDD4 blocks the rapamycin-induced biogenesis of autophagosomes and suggest that knockdown of NEDD4 might impair the SQSTM1-mediated inclusion body autophagy, thus causing accumulation of the large SQSTM1-positive inclusion bodies in cells.

Our previous studies showed that knockdown of NEDD4 caused aggregation of GFP–LC3 puncta that was co-localized with ER membrane markers, but not with Golgi marker, and suggested that NEDD4 is required for biogenesis of autophagosomes (Sun et al., 2017). Here, we examined the effect of NEDD4 knockdown on localization of the SQSTM1-positive puncta in cells. As shown in Figs. 5A, 5C and 5D, a portion of the SQSTM1 puncta was aberrantly aggregated into large inclusion bodies in the shNEDD4 cell line, but not in the vector control cell line. Furthermore, the enlarged SQSTM1-positive inclusion bodies in the shNEDD4 cell line were co-localized with the ER marker CANX (Fig. 5D), but not with the Golgi marker GOLGA2/GM-130 (Fig. 5C), and had little change upon treatment with rapamycin (Fig. 5D), suggesting that SQSTM1-positive inclusion bodies are retained in ER membrane vesicles upon NEDD4 knockdown, which is consistent with our previous observation on aggregation of the LC3-positive puncta in ER membrane vesicles upon depletion of NEDD4 (Sun et al., 2017). These results suggest that the defect in inclusion body autophagy caused by NEDD4 depletion might occur in the early stage of the autophagosomal biogenic process in the ER.

As knockdown of NEDD4 caused aberrant aggregation of the SQSTM1-positive inclusion bodies retained in the ER (Fig. 5), we wondered whether the large aggregates of the SQSTM1-positive protein inclusion bodies in NEDD4 knockdown cells were formed upon defective ubiquitylation of SQSTM1 by NEDD4. To investigate this hypothesis, we used HEK293T or HEK293A cells to transiently overexpress GFP–LC3, SQSTM1 and the PB1 defective mutant SQSTM1-2R2A with wild-type NEDD4 or its ligase-dead mutant NEDD4-C867A. As shown in Fig. 6A, co-expression of GFP–LC3 with wild-type SQSTM1 in HEK293T cells led to typical localization of GFP-LC3 and SQSTM1 on autophagosomes. Co-expression of GFP–LC3 with the PB1-defective mutant SQSTM1-2R2A resulted in diffused fluorescence of both GFP–LC3 and SQSTM1-2R2A, confirming that homo-oligomerization, which is defective in SQSTM1-2R2A, is required for SQSTM1 to localize on autophagosomes. However, co-expression of GFP–LC3 and SQSTM1 with NEDD4 ligase-dead mutant C867A resulted in formation of multiple gigantic inclusion bodies (1–10 µm diameter) containing the GFP–LC3 and SQSTM1 fluorescence in cells (the bottom panels, Fig. 6A). These data, together with the data in Fig. 5, suggest that, first, defect in ubiquitylation of SQSTM1 by NEDD4 may impair inclusion body autophagy; and second, ubiquitylation of SQSTM1 is not required for homo-oligomerization, as either knockdown of NEDD4 or overexpression of the ligase-dead mutant of NEDD4 did not result in diffused localization of SQSTM1 and GFP–LC3 in cells (Fig. 5 and Fig. 6A). In addition, overexpression of the ligase-dead mutant induced much larger inclusion bodies than that induced by knockdown of NEDD4 (Fig. 5A-D, Fig. 6A), indicating that the trapping of SQSTM1 by the ligase-dead mutant of NEDD4, as shown in Fig. 1B, which reduces the level of free SQSTM1, produces a much more severe defect in autophagy-mediated removal of cellular inclusion bodies than that by depletion of NEDD4 only, suggesting that SQSTM1 in the NEDD4 knockdown cells may still retain partial function in facilitating the removal of inclusion bodies.

To confirm that ubiquitylation of SQSTM1 is required for inclusion body autophagy, we co-expressed SQSTM1 or SQSTM1-2R2A plus GFP–LC3 with or without NEDD4 or its ligase-dead mutant C867A in HEK293A cells (Fig. 6B). Upon co-expression of SQSTM1 plus GFP–LC3 or NEDD4, the cells showed localization of SQSTM1 and GFP–LC3 on normal autophagosomes (Fig. 6Bi–iii). However, upon co-expression of SQSTM1 with the NEDD4 ligase-dead mutant C867A, the cells formed huge inclusion bodies containing SQSTM1 and GFP–LC3 (Fig. 6B iv–vi, Fig. 6Ci–iii), confirming that lack of SQSTM1 ubiquitylation by NEDD4 impairs inclusion body autophagy. Consistent with Fig. 6A, the PB1-defective mutant SQSTM1-2R2A and the co-expressed GFP–LC3 displayed diffused distribution in cells in either the presence or absence of NEDD4 or the ligase-dead mutant (Fig. 6Bvii–xii, Fig. 6Ci–iii). These data suggest that ubiquitylation of SQSTM1 is required for its function in inclusion body autophagy, but not for its oligomerization or localization on autophagosomes.

Ubiquitylation is an important biochemical process in selective autophagy. Most of the studies on ubiquitylation in selective autophagy have been focused on the role of ubiquitylation in recognition of autophagic cargos (Kirkin et al., 2009a; Johansen and Lamark, 2011; Stolz et al., 2014). Recently, ubiquitylation of autophagy receptors has been studied and found to be involved in a diversified regulatory function in autophagy. Ubiquitylation by RING family E3 ubiqiuitin ligases and E2 conjugating enzymes either regulates the autophagic receptor activity (Pan et al., 2016; Lee et al., 2017; Peng et al., 2017) or promotes the proteasomal degradation of SQSTM1 (Song et al., 2016). Our previous studies have shown that NEDD4, a member of the HECT E3 ubiquitin ligase family, not only directly binds to autophagosomal protein LC3 (Sun et al., 2017), but also interacts with SQSTM1 through the HECT domain and polyubiquitylates SQSTM1. Knockdown of NEDD4 in lung cancer A549 cells impaired both rapamycin- and starvation-induced activation of autophagy and formation of autophagosomes, and caused deformation of mitochondria, as we have shown previously (Sun et al., 2017). In this report, we have demonstrated a new role of ubiquitylation of SQSTM1 by the HECT E3 ubiquitin ligase NEDD4 in regulation of inclusion body autophagy. Depletion of endogenous NEDD4 or ectopic overexpression of the ligase-dead mutant of NEDD4 caused formation of aberrant gigantic aggregates of both LC3- and SQSTM1-positive inclusion bodies, pointing to an important role of NEDD4 in the SQSTM1-mediated inclusion body autophagy.

SQSTM1 is known to be involved in inclusion body autophagy (Zatloukal et al., 2002; Komatsu et al., 2007; Pankiv et al., 2007). Early studies found that SQSTM1 associated with ubiquitylated Mallory–Denk bodies in alcoholic liver and Lewy bodies in Parkinson's disease tissue (Stumptner et al., 1999, 2007; Zatloukal et al., 2002). Thus, SQSTM1 was defined as an inclusion body marker protein (Zatloukal et al., 2002). Subsequently, after interaction of SQSTM1 with the autophagosomal protein LC3 was discovered, this inclusion body association was linked to the function of SQSTM1 in mediating inclusion body autophagy (Komatsu et al., 2007; Pankiv et al., 2007). As both the LC3 (the LIR domain) and the ubiquitin (the Uba domain) binding ability are possessed by SQSTM1, it is possible that SQSTM1 functions as a ubiquitylated inclusion body autophagy receptor by recruiting the inclusion body to autophagosomes through interaction with ubiquitin and LC3 (Kirkin et al., 2009a). In later studies, this functional mode has been extended to the other autophagic cargos, such as invaded bacteria and peroxisomes, whose autophagy is also regulated by SQSTM1 (Zheng et al., 2009; Zhang et al., 2015). Interestingly, NEDD4 was previously found to facilitate the endosome-mediated lysosomal degradation of α-synuclein, a major component of Lewy bodies in Parkinson's disease, by directly interacting with and ubiquitylating α-synuclein (Tofaris et al., 2011; Chung et al., 2013; Sugeno et al., 2014). NAB2, a small chemical that is an activator of NEDD4, reversed a mutated α-synuclein-induced cytotoxicity in neurons derived from Parkinson's disease patients (Chung et al., 2013). Although it has been proposed that NEDD4-facilitated degradation of α-synuclein is through an endosomal/lysosomal route, not autophagy (Sugeno et al., 2014), our work suggest that the role of autophagy in the NEDD4-facilitated α-synuclein degradation might need to be re-examined. Furthermore, our studies also suggest that NEDD4 might be a universal selective autophagic E3 ubiquitin ligase that is involved in other types of SQSTM-mediated selective autophagy, such as xenophagy, pexophagy and mitophagy. In fact, we have observed that knockdown of NEDD4 in lung cancer A549 cells induced aberrant enlargement and deformation of mitochondria (Sun et al., 2017). Thus, NEDD4 might be an effective therapeutic target for SQSTM1-mediated selective autophagy-related diseases, particularly neuronal degenerative diseases.

Our studies presented in this report added a new mechanism underlying the SQSTM1-mediated selective autophagy, i.e. the ubiquitylation by NEDD4 via the PB1 domain regulates the cargo receptor activity of SQSTM1. The PB1 domain of SQSTM1 functions in homodimerization and heterodimerization with atypical PKCs (aPKCs), NBR1 and MAP2K5 (MEK5) (Lamark et al., 2003; Moscat et al., 2006, 2009). The PB1 domain of SQSTM1 has two regions that are involved in dimerization: one region at the N-terminus of the PB1 contains several positively charged residues, such as K7, R21 and R22 in SQSTM1, the other region is at the C-terminus of the PB1 the so-called OPCA motif that is conserved in a number of the PB1-containing proteins (Lamark et al., 2003). It has been demonstrated that K7, R21 or R22 in the PB1 of SQSTM1 is essential for either homo-dimerization or heterodimerization through binding to the OPCA motif of the other PB1 (Lamark et al., 2003). Our studies have shown that deletion of the first 43 amino acid residues in the PB1 domain, which does not include the OPCA region, eliminated binding to NEDD4 and the ubiquitylation by NEDD4 (Fig. 4), indicating that the OPCA motif of SQSTM1 is not required for interaction with NEDD4. Mutation of R21/R22 of the PB1 domain, which are critical residues for interaction with the OPCA region in dimerization, significantly reduced interaction with and ubiquitylation by NEDD4 (Fig. 4), suggesting that an OPCA-like region may be present in NEDD4. As the HECT domain of NEDD4 is the region binding to SQSTM1 (Fig. 1), it is possible that the HECT domain contains an OPCA-like region that is capable of binding to the positive residues of the PB1 of SQSTM1. In addition, NEDD4 does not interact and ubiquitylate another selective autophagy receptor NBR1 that also contains a PB1 domain at the N-terminus (Fig. 3) (Kirkin et al., 2009a). Interestingly, the PB1 domain of NBR1 has the OPCA motif but does not have the arginine residues corresponding to R21 and R22 in the PB1 of SQSTM1 (Lamark et al., 2003). This further supports the hypothesis that the HECT domain of NEDD4 contains an OPCA-like region that interacts with the positively charged residues in the N-terminal region of the PB1 of SQSTM1.

The PB1 domain is essential for SQSTM1 to homo- and hetero-dimerize/oligomerize and to localize on autophagosomes (Lamark et al., 2003; Itakura and Mizushima, 2001). However, ubiquitylation by NEDD4 is dispensable for autophagosomal localization of SQSTM1, because knockdown of NEDD4 or expression of the ligase-dead mutant of NEDD4 did not produce a diffused distribution like the dimerization/oligomerization-defective mutants of SQSTM1 (Fig. 6). We currently do not know how the NEDD4-mediated ubiquitylation regulates the exact molecular function of SQSTM1. There are two possible molecular effects for the ubiquitylation. The first effect facilitates the hetero-dimerization of SQSTM1with NBR1. It has been observed that cooperation between SQSTM1 with NBR1 plays an important role in inclusion body autophagy (Kirkin et al., 2009b; Tanji et al., 2015). Ubiquitylation on the PB1 domain by NEDD4 may enhance the binding affinity of SQSTM1 to NBR1, thus enabling SQSTM1 to recruit NBR1 for inclusion body autophagy, while knockdown of endogenous NEDD4 or overexpression of the ligase-dead mutant of NEDD4 may interfere with interaction between SQSTM1 and NBR1 and cause accumulation of inclusion bodies (Fig. 6). The second possible effect is changing conformational structure of SQSTM1 by the ubiquitylation for interaction with downstream effectors, such as TRAF6 and KEAP1, to activate inclusion body autophagy. The ubiquitylated PB1 domain could intramolecularly interact with the Uba domain at the C-terminus to expose the effector interactive regions that sit between, thus enhancing interaction with downstream effectors and activating selective autophagic signaling. Our future studies will follow these questions and determine how the ubiquitylation affects hetero-dimerization of SQSTM1 with NBR1 and interaction with downstream interactive effectors in inclusion body autophagy.

Anti-SQSTM1 (D3; SC-28359) antibody was purchased from Santa Cruz Biotech; anti-NBR1 from Proteintech (16004-1-AP); anti-NEDD4 from Millipore (07-049); anti-LC3 from Abgent (AP1802a); anti-GFP (MMS-118R), anti-ubiquitin (P4G7; MMS-258R) and anti-HA (MMS-101R) from BioLegend; anti-CANX and anti-GOLGA2/GM130 from ECM Biosciences (OK7670); anti-ACTB from Sigma-Aldrich (A5441). The dilution for antibodies was 1:1000 for western blotting; 2 µg antibody/ml lysate for immunoprecipitation; 1:50 for immunofluorescence staining. The DNA mutagenesis kit (QuikChange^® Site-Directed Mutagenesis Kit) was purchased from Strategene (200518). The NEDD4 shRNA (5′-AUUUGAACCGUAUAGUUCAGC-3′) in the lentiviral expression vector pLKO.1 was purchased from Open Biosystems (RHS4533-EG4734). All the cell lines were purchased from ATCC.

HEK293T, HEK293A and A549 cells were maintained in Dulbecco's modified Eagle's medium (Gibco, 11965092) with 10% heat-inactived fetal bovine serum (FBS), 100 units/ml penicillin and streptomycin at 37°C with 5% CO₂. For transfection, the cells were seeded 1 day before transfection. The transfection procedures were the same as described previously (Lin et al., 2010; Sun et al., 2017).

Human SQSTM1 or MAP1LC3B cDNA was subcloned into the mammalian expression vectors pcDNA3-HA, pcDNA3-MYC, lentiviral GFP vector pLVTHM-GFP (a gift from Dr Jihe Zhao at University of Central Florida) or GST fusion vector pGEX4T3 (GE Health Care Life Sciences, 28-9545-52). Human NEDD4 or the mutant cDNA was subcloned into the lentiviral expression vector pFUW (a gift from Dr Jihe Zhao at University of Central Florida) for establishing stable cell lines in A549 cells, and into the mammalian expression vector pcDNA3-HA for transient transfection in HEK293 cells. Point mutations and truncations of NEDD4 or SQSTM1 were created using a mutagenesis kit from Stratagene.

Cells were rinsed once with ice-cold PBS and lysed in ice-cold Mammalian lysis buffer (40 mM Hepes, pH 7.4, 100 mM NaCl, 1% Triton X-100, 25 mM glycerol phosphate, 1 mM sodium orthovanadate, 1 mM EDTA, 10 µg/ml aprotinin, and 10 µg/ml leupeptin) or RIPA buffer (40 mM Hepes, pH 7.4, 1% Triton X-100, 0.5% sodium deoxylcholate, 0.1% SDS, 100 mM NaCl, 1 mM EDTA, 25 mM β-glycerolphosphate, 1 mM sodium orthovanadate, 10 µg/ml leupeptin and aprotinin) as indicated. The cell lysates were cleared by centrifugation at 13,000 rpm for 15 min. For immunoprecipitation, primary antibodies were added to the lysates and incubated with rotation at 4°C for 30 min, followed by adding 20 µl of protein-A–Sephorose bead slurry (1:1) to the lysates and incubating with rotation for an additional 3 h. The immunoprecipitates were washed three times with lysis buffer. The cell lysates or immunoprecipitated proteins were denatured by addition of SDS-PAGE sample buffer and boiled for 5 min, resolved by 8–14% SDS-PAGE. The proteins in the gel were transferred to PVDF membranes (Millipore). Immunoblotting with chemiluminescence was performed as described previously (Lin et al., 2010; Sun et al., 2017).

GST fusion protein expression, purification and affinity precipitation assay were performed as previously described (Lin et al., 2010; Sun et al., 2017).

Cells were cultured in glass coverslip-bottomed culture dishes (MatTek, Ashland, MA) to 50–80% confluence. After the culture medium was aspirated, the cells were rinsed with PBS twice, fixed with 3.7% paraformaldehyde at 25°C for 10 min, and permeabilized with 0.2% Triton X-100 in PBS at 25°C for 10 min. After washing with PBS, the cells were incubated with primary antibody at 8°C overnight. The cells were washed with PBS three times and incubated with secondary antibody conjugated with a fluorescent dye at 37°C for 1–2 h. After washing with PBS three times, fluorescent staining of the cells was visualized under a Zeiss LSM710 confocal microscope or Nikon inverted fluorescent microscope.

The analysis and quantification of fluorescent images were performed using ImageJ. The threshold in detection of the fluorescence was set to cover all the visible fluorescent puncta. Numbers of fluorescent puncta were counted from two randomly selected fluorescence microscopy fields (25 to 47 cells). The size of the fluorescent puncta in each cell was measured and averaged. Statistical analysis was performed based on the numbers and average sizes of puncta from each of the cells.

The viral packaging was performed as described previously (Mi et al., 2015; Sun et al., 2017). Briefly, the lentiviral plasmids were co-transfected with psPAX2 (Addgene) and pMD2.G (Addgene) packaging plasmids into actively growing HEK293KT cells using Lipofectamine 2000 transfection reagent. Viral particle-containing culture medium was collected every 24 h three times. The medium was cleared by centrifugation at 1000 g for 5 min, and used for infecting target cells in the presence of 6 µg/ml polybrene. The infected cells were selected with puromycin.

Autophagy was activated by treatment of cells with the mTOR inhibitor rapamycin (LC Laboratory, R5000) for the indicated time. LC3- or SQSTM1-positive autophagosomes were visualized by either immunofluorescence staining or GFP-tag under Zeiss LSM710 confocal fluorescent microscope or a Nikon inverted fluorescent microscope.

Detection of ubiquitylated proteins was performed using both GST-Uba pulldown and immunoprecipitation assays as described previously (Lin et al., 2010; Wang et al., 2010). Briefly, cells were lysed with RIPA buffer (40 mM Hepes, pH 7.4, 1% Triton X-100, 0.5% sodium deoxylcholate, 0.1% SDS, 100 mM NaCl, 1 mM EDTA, 25 mM β-glycerolphosphate, 1 mM sodium orthovanadate, 10 µg/ml leupeptin and aprotinin) and the ubiquitylated proteins were detected either by immunoprecipitation with the primary antibody followed by immunoblotting with an anti-ubiquitin antibody (BioLegend, 646302), or by affinity precipitation with GST–UBA-conjugated glutathione beads followed by immunoblotting with anti-SQSTM1 antibody.

The Student t-test was used in statistical analysis of experimental data. A P-value less than 0.05 was considered as statistically significant.

We want to thank Dr Jihe Zhao of University of Central Florida for the lentiviral expression vectors pLVTHM-GFP and pFUW.