UBE3C tunes autophagy via ATG4B ubiquitination

ABSTRACT

ATG4B is a core protein and essential for cleaving precursor MAP1LC3/LC3 or deconjugating lipidated LC3-II to drive the formation of autophagosomes. The protein stability and activity of ATG4B regulated by post-translational modification (ubiquitination) will directly affect macroautophagy/autophagy. However, the mechanism involved in ATG4B ubiquitination is largely unclear. In this study, a new E3 ligase of ATG4B, UBE3C, was identified by mass spectra. UBE3C mainly assembles K33-branched ubiquitin chains on ATG4B at Lys119 without causing ATG4B degradation. In addition, the increased ubiquitination of ATG4B caused by UBE3C overexpression inhibits autophagy flux in both normal and starvation conditions, which might be due to the reduced activity of ATG4B and ATG4B-LC3 interaction. This reduction could be reversed once the lysine 119 of ATG4B was mutated to arginine. More important, under starvation conditions the interaction between ATG4B and UBE3C apparently decreased followed by the removal of the K33-branched ubiquitin chain of ATG4B. Thus, starvation-induced autophagy could be partially suppressed by an increased ubiquitination level of ATG4B. In conclusion, our research reveals a novel modification mode of ATG4B in which UBE3C can fine tune ATG4B activity by specific ubiquitination regulating autophagy without causing ATG4B degradation.

Abbreviation: ATG: autophagy-related; Baf: bafilomycin A₁; CBB: Coomassie Brilliant Blue; CM: complete medium; CQ: chloroquine; GFP: green fluorescent protein; HA-Ub: HA-tagged ubiquitin; IF: immunofluorescence; IP: immunoprecipitation; K: lysine; KO: knockout; K0: all K-to-R mutant; MAP1LC3/LC3: microtubule associated protein 1 light chain 3; MS: mass spectrometry; NC: negative control; R: arginine; WCL: whole cell lysate; WT: wild-type.

Introduction

Autophagy is a conserved protein and organelle degradation system, which is regulated by multiple autophagy-related (ATG) genes [1]. Atg4 is the only specific cysteine protease of Atg8-family proteins to play a crucial role in the autophagy cascade [2,3]. There are four Atg4 homologs in mammalian cells, in which ATG4B has the strongest enzyme activity and broadest spectrum against all substrates such as the MAP1LC3/LC3 and GABARAP subfamilies (henceforth we refer only to LC3) [2,4,5]. During autophagy biogenesis, ATG4B cleaves the precursor form of LC3 to generate LC3-I, exposing the active glycine residue, which is conjugated to the lipid phosphatidylethanolamine (PE) on the membrane to form LC3-II, thereby allowing expansion of the phagosome membrane [6]. ATG4B additionally induces the delipidation of LC3-II from the membrane once the biogenesis of the autophagosome is completed, forming LC3-I again to participate in new autophagosome formation.

The protein stability and activity of ATG4B could be regulated by various post-translational modifications. ULK1 phosphorylates ATG4B on Ser316 to inhibit the ATG4B activity and the LC3 processing [7]. In glioblastoma, the kinase STK26/MST4 phosphorylates ATG4B on Ser383 of ATG4B, thereby stimulating ATG4B activity and increasing autophagic flux. In hepatoma cells, phosphorylation of ATG4B by AKT can inhibit the mitochondrial activity and enhance the Warburg effect [8]. The DDIT4/REDD1-TXNIP prooxidant complex regulates ROS generation, and then regulate ATG4B activity to control pressure-induced autophagy and maintain exercise ability [9]. In addition, we have found that Cys292 and Cys361 are the key sites to regulate the redox form and oligomer formation of ATG4B, and reversible redox modification can regulate the protease activity of human ATG4B, thereby affecting autophagy [10,11]. ATG4B enzyme activity could also be reduced once ATG4B was nitrosylated at Cys189 and Cys292 thus mediating neurotoxicity [12]. When ATG4B is oxygen glycosylated, it can promote autophagy by increasing its hydroxylase activity [13]. In addition, the ubiquitination modification mediated by E3 ubiquitin ligase RNF5 impaired the stability of ATG4B, thereby affecting the level of autophagy and the sensitivity to bacterial infection [14]. However, whether ATG4B could be ubiquitinated by any other E3 ligases in normal or starvation conditions and whether ubiquitination of ATG4B could lead to protein degradation merit further clarification.

Protein ubiquitination is under the catalysis of a series of enzymes, mainly E1 ubiquitin activating enzyme, E2 ubiquitin binding enzyme and E3 ubiquitin ligase [15]. Polyubiquitin chains are usually linked to the lysine residues or N-terminal methionine of the substrate via one of ubiquitin’s seven lysine residues (Lys6, Lys11, Lys27, Lys29, Lys33, Lys48, and Lys63) or ubiquitin’s N-terminal methionine, which may be the basic degradation signal of proteins, and may also change the location, function or activity of proteins [16]. The Lys48-linked chain targets its substrate for degradation by the 26S proteasome [17], the Lys63-linked chain plays a role in NFKB/NFκB signaling [18], DNA repair and receptor endocytosis [19], the Lys29-linked chain can target proteins for degradation by lysosomes [20], Lys29/Lys33 can be assembled on AMP-activated protein kinase (AMPK)-related kinase family substrates to participate in cell polarity and proliferation [21], Lys6 May regulate DNA repair [22], and Lys27 May function in stress response pathways [23], K11 and K48-linked ubiquitin chains are involved in endoplasmic reticulum-associated degradation [24].

E3 ligase plays a key role in ubiquitination, which can be classified into three major structurally distinct types: E3s containing the Homology to E6AP C Terminus (HECT) domain, N-end rule E3s, and E3s with the Really Interesting New Gene/RING finger, including its derivatives, the U-Box and the Plant Homeo-Domain/PHD [25]. Studies have found that human HECT E3 ligases UBE3C can mediate the polyubiquitin chains of K29 and K48, preferentially K48 link [26]. Whether UBE3C can also assemble other Ub chains on substrates is still unknown.

In this study, UBE3C was demonstrated to interact with ATG4B by mass spectrometry. As an E3 ligase, UBE3C was found to assemble K33-branched ubiquitin chain to ATG4B without causing protein degradation. ATG4B ubiquitination by UBE3C therefore tunes autophagy function through downregulating the activity of ATG4B and the interaction between ATG4B and LC3 in both normal and starvation conditions.

Results

Identification of E3 ligase UBE3C as a novel protein that binds to ATG4B

In the protein post-translation modification experiment, we found that ATG4B can be ubiquitinated in 293T cells (Figure 1A). Followed by preliminary exploration of the mechanism of ATG4B ubiquitination, we wanted to find any E3 ligase which could regulate ATG4B ubiquitination at its 13 lysines (Figure S1A). Six repeated tandem mass spectrometry analysis were performed to identify the proteins interacting with ATG4B when proteasome activity was inhibited. In 293T cells, Flag-ATG4B and HA-Ub plasmids were co-overexpressed as the experimental group that a total of 1268 proteins were identified from immuno-precipitation with anti-Flag antibodies. In addition, Flag-vector and HA-Ub plasmids were overexpressed as the control group that 301 proteins were identified. Then Venn analysis is performed to identify 1065 proteins specific to the experimental group that associated with ATG4B ubiquitination (Figure S1B). We further removed the proteins that appeared only once and 157 proteins were left (Figure S1C). From these proteins, a total of 79 proteins that appeared at least three times with higher possibility to interact with ATG4B were finally demonstrated in Figure 1B. Next, in order to find the proteins with higher correlation with ubiquitination system, we further annotated these 79 proteins using KEGG and GO:BP analysis that 7 proteins belong to ubiquitin-related processes (hsa04120, GO: 0032434, GO: 2000058, GO: 0030433, GO: 0032435, GO: 2000059, GO: 1904668, GO: 0090085, GO:0032436, GO:0043161), while UBE3C is the only E3 ligase (Figure S1D). Ultimately, comprehensively considered the frequency of protein occurrence and peptide score, UBE3C was chosen for the further tests.

Figure 1.

Identification of E3 ligase UBE3C as a novel protein binds to ATG4B. (A) ubiquitination of ATG4B was detected by immunoprecipitation (IP). 293T cells were transfected with flag-ATG4B or flag-vector respectively together with HA-Ub. After 24 h, cells were incubated with 10 µM of MG132 for 6 h, cell lysates were subjected to IP with anti-flag-gel and then analyzed by SDS-PAGE, and ubiquitinated ATG4B were detected using an anti-HA antibody. (B) a total of 79 probable ATG4B-interacting proteins under the ubiquitination conditions. The heat map showed proteins appeared at least three times in the experimental group (flag-ATG4B and HA-Ub plasmids co-overexpressed in 293T cells) from six repeats of tandem mass spectrometry. (C) HeLa cells transfected with GFP-ATG4B and mCherry-UBE3C plasmids were incubated on a confocal dish, then treated with 0.1% DMSO or MG132 (10 μM) for 6 h. Cells were examined by immunofluorescence microscopy to detect the colocalization of mCherry-UBE3C and GFP-ATG4B. Scale bar: 10 μm. (D) interaction of endogenous UBE3C and endogenous ATG4B were detected without MG132 treatment by western blotting. (E) GST-ATG4B or GST control protein (5 µg) were incubated with GST agarose gel in TBST buffer at room temperature. After 1 h of incubation, HECT segment of UBE3C (5 µg) was added, and thoroughly mixed in vibrating mixer for overnight incubation at 4°C. Then GST-gel was collected and loaded for western blotting. ATG4B and UBE3C levels were detected by GST and UBE3C antibodies. All representative blots are from at least 3 independent experiments. (F) cell lysates of 293T cells transfected with GFP-ATG4B truncations were incubated with GFP-agarose gel overnight at 4°C. Immunoprecipated ATG4B truncations and endogenous UBE3C were detected by western blotting with GFP and UBE3C antibodies.

To further confirm whether UBE3C is a novel interacting protein of ATG4B, we overexpressed GFP-ATG4B and mCherry-UBE3C plasmids with fluorescent labels, the results showed significant colocalization especially when MG132 was added (Figure 1C). Also, we carried out immunostaining of endogenous UBE3C and found that GFP-ATG4B colocalized with UBE3C as well before and after MG132 treatments (Figure S1E). In addition, co-immunoprecipitation analysis in 293T cells demonstrated that there is an obvious interaction between ATG4B and UBE3C (Figure 1D). We further purified GST-fused ATG4B and the HECT segment of UBE3C in vitro [27], purified GST-ATG4B was able to pull down the HECT segment of UBE3C directly, suggesting a direct interaction exists between ATG4B and UBE3C (Figure 1E). To determine which domains of ATG4B interact with UBE3C, we transfected four ATG4B truncations with GFP tag in 293T cells. Only 39–335 fragments of ATG4B was found to strongly bind to UBE3C (Figure 1F).

Overall, we identified an E3 ligase UBE3C as a novel protein that binds to the 39–335 domain of ATG4B.

UBE3C assembles K33 branched ubiquitin chain on ATG4B

Having measured the interaction between UBE3C and ATG4B, we next examined whether UBE3C could ubiquitinate ATG4B. UBE3C and Flag-ATG4B were co-expressed with HA-Ub in 293T cells that ATG4B ubiquitination increased significantly compared to the control group. Conversely, knockdown of UBE3C markedly decreased ATG4B ubiquitination as revealed by western blotting (Figure 2A). Furthermore, in order to verify whether promotion of ATG4B ubiquitination by UBE3C depends on lysine residues, we co-transfected Flag-ATG4B[K0] (13 lysines of ATG4B were all mutated to arginine) and HA-Ub into 293T cells with or without UBE3C. As shown in Figure 2B, UBE3C failed to further promote the ubiquitination of ATG4B[K0]. Similarly, knocking down of UBE3C could not block the ubiquitination of ATG4B[K0] as well (Figure 2B). These data suggested that the ubiquitination of ATG4B by UBE3C relies on lysine residues.

Figure 2.

UBE3C assembles K33-branched ubiquitin chain on ATG4B. (A) analysis of the effect of UBE3C on the ubiquitination level of ATG4B. The ubiquitinated proteins were pulled down by anti-flag-gel from 293T cells transfected with flag-vector, or flag-ATG4B, HA-Ub, UBE3C, and sh-UBE3C and analyzed by western blotting with indicated antibodies. The transfection lasted 24 h, following cells were incubated with MG132 (10 µM) for 6 h. Ratios of ubiquitinated ATG4B:Input flag-ATG4B were quantified. n = 3, ***P < 0.001, ****P < 0.0001. (B) analysis of the effect of UBE3C on the ubiquitination level of ATG4B[K0]. 293T cells were transfected with flag-vector, flag-ATG4B[K0], HA-Ub, UBE3C, and sh-UBE3C as indicated, after 24 h, cells were incubated with MG132 (10 µM) for 6 h, cell lysates were subjected to IP with anti-flag-gel and then analyzed by western blotting. Ratios of ubiquitinated ATG4B:Input flag-ATG4B were quantified. n = 3, ns: no significance. (C) analysis of the types of ATG4B ubiquitin chain that affected by UBE3C. The ubiquitinated proteins were pulled down by anti-flag-gel under denaturing conditions from 293T cells transfected with indicated constructs and analyzed by western blotting with indicated antibodies. After 24 h transfection, cells were incubated with MG132 (10 µM) for another 6 h. The amounts of different ubiquitin constructs used for transfection were adjusted to allow an equal level of ATG4B ubiquitination among the groups without UBE3C transfection. (D) 293T cells were transfected with UBE3C plasmids for 24 h, following cells were incubated with MG132 (10 µM) for 6 h, cell lysates were subjected to IP and then analyzed by western blotting. The ubiquitinated endogenous ATG4B was detected using the anti-ub-K33 antibody. Ratios of ATG4B with K33 ub chain:endogenous ATG4B were quantified. n = 3, *P < 0.05.

Ubiquitin contains seven lysine residues which were used to link to another ubiquitin moiety to form polyubiquitin chains with a highly diverse topology. To verify which ubiquitin chain can be assembled by ATG4B itself, we transfected ubiquitin containing only one lysine residue, namely K6, K11, K27, K29, K33, K48, and K63, respectively, in 293T cells stably expressing Flag-ATG4B. Studies have found that human HECT E3 ligases could assemble atypical Ub chains such as K48/K29- and K11/K33-linked Ub chains [26]. We then explored which ubiquitin chains was assembled by UBE3C. We co-transfected these ubiquitin mutant plasmids containing only one lysine respectively with UBE3C that UBE3C could mainly promote ATG4B to assemble K33 Ub chain, instead of K48 or K63 Ub chain (Figure 2C). To verify this result, we next mutated the lysine 33 of ubiquitin to arginine, namely K33R. The result showed that UBE3C cannot further promote ATG4B to assemble K33R Ub chain (Figure 2C). Furthermore, endogenous ATG4B of cell lysate without overexpression of Ub were found to be significantly ubiquitinated with K33 Ub chain after overexpression of UBE3C (Figure 2D). These data suggest that UBE3C could assemble K33-branched Ub chain on ATG4B.

UBE3C mediates ATG4B ubiquitination at Lys119

To figure out which lysine is responsible for UBE3C-mediated ubiquitination within ATG4B, the individual K-to-R mutants were co-expressed with UBE3C in 293T cells. The result showed that UBE3C failed to strongly increase the ubiquitination level once some lysine residues were mutated to arginines (Figure 3A). To assure which lysine contributed to the ubiquitination of ATG4B by UBE3C, purified ATG4B mutants such as K102R, K119R, K244R, K259R, and K324R were prepared respectively to evaluate their in vitro ubiquitination level by UBE3C. We finally found that compared to other mutants only K119R failed to be efficiently ubiquitinated by UBE3C (Figure 3B). These data suggested that Lys119 of ATG4B might be the potential ubiquitination site responsible for UBE3C-mediated ubiquitination. In addition, we further found that UBE3C does not directly reduce the protein level of ATG4B in both 293T and MEF cells (Figure 3C).

Figure 3.

UBE3C mediates ATG4B ubiquitination at Lys119. (A) analysis of the effect of UBE3C on the ubiquitination level of different K-to-R mutants. 293T cells were transfected with different K-to-R mutants of flag-ATG4B together with HA-Ub and UBE3C with MG132 (10 µM) for another 6 h. Cell lysates were subjected to IP and then analyzed by western blotting. (B) in vitro ubiquitination assay was performed with 5 μg of purified his-ATG4B, His-ATG4B^K119R, HECT and other required reagents as shown in the methods section. The mixture was incubated at 37°C for 120 min. Ubiquitination level was detected by anti-ub antibody and quantified. n = 3, ***P < 0.001. (C) analysis of the effect of UBE3C on the protein level of ATG4B. 293T cells or MEF cells were transfected with plasmids as indicated. Cell lysates were analyzed by western blotting with indicated antibodies.

These data indicated that UBE3C assembled the K33-branched ubiquitin chain at Lys119 on ATG4B without affecting ATG4B degradation.

Overexpression of UBE3C inhibits autophagy under starvation conditions

ATG4B plays a very important role in the process of autophagy. We have found that UBE3C can promote the ubiquitination of ATG4B, so we were committed to verify whether UBE3C could regulate autophagy as well. 293T cells expressing UBE3C were cultured in nutrient-rich conditions or in starvation conditions in the absence or presence of the lysosome inhibitor bafilomycin A₁ (Baf). In nutrient-rich conditions, the basal level of LC3-II was significantly lower than that in the corresponding control group or Baf group once UBE3C was overexpressed, suggesting UBE3C could inhibit autophagy flux (Figure 4A). We then analyzed the effects of UBE3C on starvation-induced autophagy that UBE3C could further reduce LC3-II induced by EBSS or by EBSS plus Baf, suggesting that UBE3C suppressed autophagy flux increased by starvation conditions as well. In addition, the LC3-II level between EBSS plus Baf group and Baf group does not change obviously when UBE3C were overexpressed, suggesting UBE3C could counteract the autophagy induction effect of EBSS (Figure 4A). On the contrary, UBE3C knockdown clearly increased autophagy flux in either normal or starvation conditions (Figure 4B). We further verified the effect of UBE3C on endogenous autophagosome formation using HeLa cells. Similar to the above findings, UBE3C overexpression decreased the amount of autophagosomes cultured in either complete medium or EBSS while sh-UBE3C strongly enhanced the number of autophagosomes indicated by LC3 immunostaining (Figure 4C). Besides EBSS, another autophagy inducer torin1 was applied for dissecting the role of UBE3C on autophagy. Intriguingly, UBE3C can still inhibit autophagy flux under the treatment of torin1 illustrated by western blot analysis of LC3-II and immunostaining analysis of endogenous LC3 puncta or GFP-LC3 puncta (Figure S2A-C).

Figure 4.

Overexpression of UBE3C inhibits autophagy under starvation conditions. (A-B) analysis of the effect of UBE3C on starvation induced autophagy. 293T cells transfected with his-UBE3C for 48 h or stably expressing sh-UBE3C were incubated with EBSS for 1 h with or without 0.5 μM of bafilomycin A₁ (baf) for 3 h, and then analyzed by immunoblotting with indicated antibodies. Quantification of LC3-II:GAPDH ratios were performed. n = 3, *P < 0.05,**P < 0.01. (C) the amount of endogenous LC3 puncta was affected by UBE3C. HeLa cells were transfected with his-UBE3C or sh-UBE3C plasmids as indicated for 48 h, then incubated with EBSS for 1 h. More than 50 cells were counted for the quantification of LC3 puncta per experimental condition. n = 4, **P < 0.01, ***P < 0.001. Scale bar: 10 μm. (D) analysis of the effect of UBE3C on nonradioactive quantification of autophagic protein degradation with l-azidohomoalanine labeling. HeLa cells were transfected with or without sh-UBE3C or his-UBE3C plasmids for 4 h and labeled with AHA for 20 h, then cells transfected with his-UBE3C were starved in EBSS for another 12 h. Fluorescence tagged of AHA-labeled proteins with the click reaction and then relative fluorescence intensity of the cells were analyzed by flow cytometry. In this chart, the lower intensity means the higher protein degradation activity. n = 5, ***P < 0.001.

Moreover, to further verify the roles of UBE3C in the regulation of autophagy, we measured the long-lived protein degradation before and after knockdown or overexpression of UBE3C at normal or starvation conditions (Figure 4D). Consist with other assays we applied, knockdown of UBE3C could significantly promote the basal level degradation of long-lived proteins, while overexpression of UBE3C strongly suppressed the EBSS-induced long-lived protein degradation, suggesting that UBE3C could negatively regulate starvation-induced autophagy.

Taken together, these results suggest that overexpression of UBE3C may inhibit the biogenesis of autophagy and inhibit autophagy under starvation conditions.

Starvation reduced the interaction between ATG4B and UBE3C

As studied before that UBE3C could interact with ATG4B to facilitate the ubiquitination of ATG4B and inhibit starvation-induced autophagy flux. We then compared the interaction of UBE3C and ATG4B in detail and found endogenous UBE3C and ATG4B could interact weaker in starvation conditions compared to that in normal conditions (Figure 5A). We wondered whether the interaction change could have any effects on the ubiquitination of ATG4B in different culture medium. In EBSS the K33 chain ubiquitination of ATG4B was obviously reduced and such reduction could be further abolished by sh-UBE3C (Figure 5B), suggesting that the UBE3C-specific K33 chain ubiquitination of ATG4B was downregulated in starvation conditions. Similarly, the Ub-K33 modification of endogenous ATG4B was reduced after the treatment with torin1 as well (Figure S2D). These data suggested that reduced interaction of UBE3C and ATG4B in starvation conditions weakened the ubiquitination level of Ub-K33 chain of ATG4B.

Figure 5.

Starvation reduced the interaction between ATG4B and UBE3C. (A) 293T cells were starved with EBSS for 2 h, and the interactions between ATG4B and UBE3C were analyzed by IP and western blotting. Quantification of ATG4B:UBE3C and UBE3C:ATG4B in IP fraction were performed. n = 3, **P < 0.01, ****P < 0.0001. (B) WT and stably expressing sh-UBE3C 293T cells were incubated with MG132 (10 µM) for 6 h with EBSS for 2 h or not, cell lysates were collected and subjected to IP and then analyzed by western blotting. Ubiquitinated ATG4B was detected using the anti-ub-K33 antibody. n = 3, ***P < 0.001, ns: no significance.

UBE3C downregulated ATG4B activity and ATG4B-LC3 interaction

Since UBE3C could bind to ATG4B and regulate autophagy without altering the protein level of ATG4B (Figure S3A), we next explored whether UBE3C could affect the cleavage activity of ATG4B or the interaction between ATG4B and LC3 which might be important for autophagy regulation. We have previously reported that LC3-GST and FRET-GABARAPL2/GATE-16 (CFP-GABARAPL2-YFP) could be specific ATG4B substrates [2,28]. Here we firstly measured the activity of ATG4B using cells overexpressing or knocking down of UBE3C. The results indicated that higher level of UBE3C can partially inhibit the enzymatic activity of ATG4B, instead lower level of UBE3C enhanced the activity (Figure 6A). Importantly, activity inhibition by UBE3C overexpression could be eliminated by the mutation of K119 of ATG4B (Figure 6B), suggesting UBE3C-mediated ATG4B activity inhibition depends on K119. To verify the effect of UBE3C on ATG4B activity, we also measured the activity of ubiquitinated ATG4B enriched by HA-Ub beads from cell lysates before or after the transfection of UBE3C. As we see the cleavage activity of UBE3C group were 20% less than that of vector group (Figure S3B), suggesting the ubiquitinated ATG4B might have negative effects on ATG4B activity. Another substrate FRET-GABARAPL2, a more sensitive substrate with stronger cleavage efficiency by ATG4B, were applied that both CBB and western blotting results showed that total cell lysates with overexpressed UBE3C could also significantly inhibit the activity of endogenous ATG4B (Figure S3C). Taken together, higher UBE3C level could weaken the cleavage activity of ATG4B.

Figure 6.

Ubiquitination of ATG4B by UBE3C inhibits ATG4B activity and ATG4B-LC3 interaction. (A) analysis of the effect of UBE3C on the activity of ATG4B. The test operation is as described in the method gel-based assay. LC3-GST was used as ATG4B substrate to measure the activity of ATG4B in cell lysate after UBE3C overexpression or knockdown. Experimental systems were analyzed by Coomassie brilliant blue staining. The rest LC3-GST were quantified. n = 3, **P < 0.01, ***P < 0.001. (B) the activity of ATG4B in ATG4B KO HeLa cells stably expressing GFP-ATG4B^K119R was tested when UBE3C overexpressed. Experimental systems were analyzed by Coomassie brilliant blue staining. The rest LC3-GST were quantified. n = 3, ns: no significance. (C) 293T cells transfected with his-UBE3C for 24 h and stably expressing sh-UBE3C were incubated with EBSS for 1 h. The enzymatic activity of ATG4B were quantified. n = 3, ***P < 0.001. (D) the activity of ATG4B in ATG4B KO HeLa stably expressing GFP-ATG4B and GFP-ATG4B^K119R with or without EBSS treatment for 1 h was detected. Experimental systems were analyzed by Coomassie brilliant blue staining as well. n = 3, ***P < 0.001, ns: no significance. (E) IP analysis of the interaction between flag-ATG4B and GFP-LC3 in 293T cells. 293T cells stably expressing GFP-LC3 were transfected with flag-ATG4B. In the case of UBE3C overexpression or knockdown, the GFP-LC3 were pulled down by anti-flag-gel from 293T cells transfected with indicated constructs. Western blotting with indicated antibodies and quantification of GFP-LC3:Flag-ATG4B in IP fraction was performed. n = 3, *P < 0.05. (F) 293T cells were transfected with Vector, his-UBE3C and GFP-LC3, in the case of EBSS-induced starvation for 1 h or not, the interaction between endogenous ATG4B and GFP-LC3 was analyzed by IP, the GFP-LC3 were pulled down by Magnetic Beads coupled with ATG4B antibody. Western blotting with indicated antibodies and quantification of GFP-LC3:ATG4B in IP fraction was performed. n = 3, ***P < 0.001. (G) ATG4B KO HeLa cells stably expressing GFP-ATG4B and GFP-ATG4B^K119R treated with EBSS for 1 h or not, the interaction between GFP-ATG4B or GFP-ATG4B^K119R and endogenous LC3 was tested. Quantification of LC3:GFP-ATG4B in IP fraction was performed. n = 3, **P < 0.01, ns: no significance.

We further investigated the effect of UBE3C on ATG4B activity under the starvation conditions. We found ATG4B activity was suppressed more severe when UBE3C is overexpressed while less increase for UBE3C knockdown cells under the starvation conditions (Figure 6C). Next, we screened cell lines stably expressing full-length ATG4B and ATG4B^K119R in ATG4B KO HeLa cells to see whether there were any activity difference between these two cell lines under the starvation conditions. As expected, without changing the amount of endogenous UBE3C, ATG4B activity was still increased after EBSS treatment, while we did not detect any change for ATG4B^K119R cell line (Figure 6D), suggesting Lys119 site might be important to regulate ATG4B activity under the starvation conditions.

The ATG4B-LC3 interaction may also reflect the LC3 processing efficiency by ATG4B. We found the level of ATG4B-LC3 interaction decreased after UBE3C overexpression, in contrast more interaction could be detected once UBE3C was knocked down (Figure 6E). Since endogenous LC3 failed to be detected when precipitated endogenous ATG4B, cells stably expressing GFP-LC3 were applied. We further tested the effect of UBE3C overexpression on ATG4B-LC3 interaction under starvation conditions that UBE3C had a more negative impact on the interaction between ATG4B and LC3 (Figure 6F). Importantly, more endogenous LC3 (LC3-I plus LC3-II) were extracted using ATG4B^K119R cell line at normal culture conditions. However, the interaction of ATG4B and LC3 did not change before and after EBSS treatment using ATG4B^K119R cell line (Figure 6G). Interestingly, Flag-ATG4B[K0] could still interact with UBE3C (Figure S3D), but the overexpression of UBE3C cannot interrupt the interaction of ATG4B[K0]-LC3 anymore (Figure S3E).

Taken together, these results indicated that overexpression of UBE3C could downregulate the cleavage activity of ATG4B and ATG4B-LC3 interaction due to the ubiquitination at K119 of ATG4B in both normal and starvation conditions.

Discussion

ATG4B plays a very important role in autophagy, apoptosis, cancer, neurodegenerative diseases and many other physiological and pathological processes. Kuang et al. has reported that E3 ligase RNF5 induces ATG4B ubiquitination and proteasome-mediated degradation [16]. After that, not too much ubiquitination information of ATG4B was elaborated. Therefore, we want to dissect whether the ubiquitination modification of ATG4B also occurred and regulated the function of ATG4B in other physiological conditions in detail.

ATG4B has been reported to interact with various proteins including LC3, TMED10 and SLC27A4, and then autophagy function could be consequently modulated [29]. In this work, we further identified E3 ligase UBE3C as a new ATG4B binding protein. Although Kuang et al. has previously found that RNF5 is the E3 ligase of ATG4B [14]. RNF5 did not appear in our six repeated mass spectrometry results, suggesting ATG4B might be ubiquitinated by different E3 ligase when the experimental conditions and cells are distinct. We conjecture that the ubiquitination of the protein might be regulated by multiple E3 ligases. So, it is reasonable that more E3 ligases would be identified in other conditions.

Although both UBE3C and RNF5 are E3 ligases of ATG4B, RNF5 promotes the ubiquitination and the degradation of ATG4B, instead, UBE3C promotes the ubiquitination but not degradation. Similarly, FUNDC1 is modified by ubiquitin when PRKN/parkin is highly expressed [30], but PRKN cannot induce FUNDC1 ubiquitination degradation [31], suggesting that the occurrence of ubiquitination is not always accompanied by degradation. In addition, yeast CDK inhibitor, Sic1, can be ubiquitinated on multiple lysine residues, but only ubiquitination on a specific single lysine residue can efficiently promote its degradation by the proteasome [32]. We further determined that Lys119 of ATG4B might be responsible for UBE3C-promoted ATG4B ubiquitination, which is the first time to explore a specific ubiquitination site of ATG4B.

Ubiquitin contains seven internal lysine residues, all of which can be ubiquitinated to the formation of polyubiquitin chains. For the seven chain types, K48- and K63-branched chains are mainly reported to mediate protein degradation and signaling transduction [33]. Very little is known for K6, K11, K27, K29, and K33-linked chains. Previous studies have found that the human HECT E3 ligases assemble K48/K29- and K11/K33-linked Ub chains, UBE3C prefers K48/K29-linked Ub chains [26]. Recently, UBE3C has been found to assemble K29/K48-branched ubiquitin chains on VPS34 and inhibit autophagy [34]. Our work clearly indicated that UBE3C assembled the K33-linked Ub chain on ATG4B instead of K48-linked Ub chain, which might respond to why UBE3C can promote ATG4B ubiquitination without degradation. Moreover, we identified lysine 119 of ATG4B is potentially responsible to the UBE3C-induced K33-linked ubiquitin chains.

Our work convinced that higher level of ubiquitination of ATG4B induced by enhanced UBE3C inhibits autophagy flux in both normal and starvation conditions. Such conclusion was further confirmed by long-lived protein degradation assay, which was thought as a classic assay for autophagy function [35]. Importantly, we demonstrated the possible mechanism that when ATG4B is strongly ubiquitinated by UBE3C, its function to cleave LC3 and its interaction to LC3 will be weakened, so autophagy will be then suppressed. In addition, the reduced interaction of UBE3C and ATG4B caused by starvation or UBE3C knocking down might further result to the reduction of K33-linked ubiquitination of ATG4B which will finally upregulate autophagy. Nevertheless, how EBSS reduced the interaction of UBE3C and ATG4B, and whether EBSS affected UBE3C activity requires further clarification.

Generally, the interplays between ubiquitin-proteasome system and autophagy is the ubiquitin-dependent regulation of autophagic proteins, which has also been reported before. TRAF6 can promote K48/K63-linked polyubiquitination of ATG9A, and affect the localization of ATG9A on autophagosomes, and thus mediate autophagy under oxidative stress [36]. The ubiquitin ligase UBE3C and deubiquitinating enzyme ZRANB1/TRABID reciprocally regulate K29/K48-branched ubiquitination of PIK3C3/VPS34, enhancing the binding of PIK3C3/VPS34 to proteasomes for degradation, thereby suppressing autophagy [37]. This study added additional evidence to link the ubiquitin-proteasome system and the autophagy-lysosome pathway by demonstrating that E3 ligase UBE3C negatively regulates autophagy by the ubiquitination of ATG4B.

In conclusion, we unraveled a specific molecular mechanism of ATG4B ubiquitination modification that, different from the known RNF5 which impaired the stability of ATG4B under the condition of bacterial infection, a new E3 ligase UBE3C could fine tune autophagy that the increased ubiquitination of ATG4B at Lys119 by UBE3C reduced ATG4B activity, ATG4B-LC3 interaction, and further autophagy function under normal conditions and physiological conditions such as nutrient deprivation.

Materials and methods

Antibodies and reagents

Antibodies used in this study were as follow: anti-ATG4B (15131–1-AP) for western blotting and anti-GAPDH (60004–1) were purchased from Proteintech; anti-ATG4B for immunoprecipitation were prepared by our lab; anti-HA (5299S) was from Cell Signaling Technology; anti-UBE3C (ab226173) was purchased from Abcam; anti-LC3B (L7543) for western blotting was from Sigma Aldrich; anti-LC3 (PM036) for immunostaining and anti-Flag (M185) were from Medical & Biological Laboratories; anti-GFP (sc-9996) and anti-ubiquitin (sc-8017) were from Santa Cruz Biotechnology; anti-ubiquitin K33 (PA5–120623), HRP-conjugated goat anti-mouse IgG (35503), anti-rabbit IgG (35511) secondary antibody, Alexa Fluor 488 (A11008), and Alexa Fluor 594 (A11012) were from Thermo Fisher Scientific; anti-GST (A00097) was from GenScript.

Bafilomycin A₁ (HY-100558) was from MedchemExpress; torin1 (475991) was purchased from Sigma-Aldrich; MG132 (T2154) was from TargetMol Biotech; puromycin (S7417) was from Selleckchem; EBSS (MA0031) was from Meilunbio; E2Select Ubiquitin Conjugation Kit (UBK-982) was from UB Biotechnology.

Plasmid construction and residue mutation

The plasmid encoding Flag-ATG4B was described before [38]. Flag-ATG4B K-to-R mutants were constructed using Fast Mutagenesis System (TransGen Biotech, FM111–01). ATG4B fragment was inserted between XhoI and BamHI residues of modified pLVX-AcGFP1-N1 (Clontech 632,154) to construct GFP-ATG4B. Plasmids encoding HA-ubiquitin and its K-only mutants were purchased from Miaoling Biotech (Wuhan, China). DNA fragment encoding GFP-ATG4B^K119R was from Flag-ATG4B^K119R by PCR and then inserted to pLVX-AcGFP1-N1 vector. Human UBE3C was amplified from HEK-293T cDNA samples through standard PCR methods, and then cloned into pm-Cherry-C1, pLVX-AcGFP-N1-His vectors. Plasmid encoding His-SUMO-HECT (UBE3C) was from Addgene (66711; David Komander lab). A fragment encoding GST-tagged human ATG4B was inserted to pGEX-4T-1 vector (GE Healthcare, 28-9545-49). Fragments encoding His-tagged human ATG4B and K119R mutants were inserted to pET-28a(+) vector (Novagen 69,864). The targeted sequences of UBE3C for RNA interference as following are designed by website, fragments were cloned into the pLKO.1-TRC cloning vector (Addgene 10,878; David Root lab) according to instructions. ShRNA-1: CCGGGTCCTATTTCTATCTCCACTTCTCGAGAAGTGGAGATAGAAATAGGACTTTTTG, ShRNA-2: CCGGGTCCTATTTCTATCTCCACTTCTCGAGAAGTGGAGATAGAAATAGG ACTTTTTG.

ATG4B lysine complete mutant ATG4B[K0] was also constructed using Fast Mutagenesis System (TransGen Biotech, FM111–01). All truncations of ATG4B were amplified by high fidelity PCR using Flag-ATG4B plasmid as template and cloned to pLVX-AcGFP1-N1 for protein overexpression. pLVX-AcGFP1-N1 was a gift from Junjian Wang (Sun Yat-sen University, China). Primer synthesis and sequencing services were provided by Tsingke Biotech (Beijing, China).

Cells were transfected with plasmids using Lipofectamine 2000 (Thermo Fisher 11,668–019) or Hieff TransTM Liposomal Transfection Reagent (Yeasen Biotech, 40802ES03) according to the manufacturer’s instructions.

Cell lines and culture conditions

HeLa cells (ATCC, CCL-2), HEK-293T (ATCC, CRL-1573) were routinely cultured in Dulbecco’s modified eagle’s medium (DMEM; Thermo Scientific 11,965,092) with 10% fetal bovine serum (Sigma, F8318) and standard supplements. All cells were maintained in a 37°C, 5% (v:v) CO₂ thermostatic incubator. The ATG4B KO HeLa cell was generated by CRISPR-Cas9 methods before and stored in our lab [41]. HeLa cells stably expressing GFP-ATG4B and GFP-LC3 were achieved by Lentiviral particles. Lentiviral particles were generated by transfection of HEK293T cells with GFP-ATG4B or GFP-LC3 and psPAX2 (Addgene 12,259; Didier Trono Lab) and pMD2.G (Addgene 12,260; Didier Trono lab) at a ratio of 4:3:1 respectively. Viral supernatants were collected 48–72 h after transfection and filtered through a 0.45 μm filter membrane. HeLa cells were transduced with Lentivirus for 24 h with 10 µg/ml of polybrene (Yeasen Biotech, 40804ES76). Then the cells were selected with 5 µg/ml of puromycin for stably expressing cell lines, and half of the concentration of puromycin was continuously kept for long-term cell culture. 293T cells stably expressing sh-UBE3C, ATG4B KO HeLa cells stably expressing GFP-ATG4B and GFP-ATG4B^K119R were also achieved by Lentiviral particles. After 48 h infection, the 293T cells were selected with 2 µg/ml of puromycin, and then the flow cytometry was used to screen HeLa cells with green fluorescence.

Immunoblotting assay

The cells in the petri dish were lysed in RIPA lysis buffer (Beyotime, P0013B) containing protease inhibitor (Bimake, A32961) and placed on ice for 30 min to be fully lysed. The precipitate and lysates were separated by centrifugation at 13, 000×g for 15 min at 4°C. The lysates were quantified by BCA assay and then denatured by incubating on a heat block at 95°C for 10 min with loading buffer. Appropriate amount of lysates were loaded on and separated by SDS-PAGE, then transferred to PVDF membranes (Millipore, ISEQ00010). After being sealed in 5% skim milk at room temperature for 1 h, the PVDF membranes were incubated with primary antibodies for at least 12 h at 4°C and following HRP-conjugated secondary antibodies at room temperature for 1 h. Bands signal was detected by Tanon 5200 Chemiluminescence Image analysis system (Tanon, Shanghai).

Immunoprecipitation assay

The cells were lysed on ice with 300 μl of IP lysis buffer (Beyotime, P0013J) containing protease inhibitor for 30 min. The precipitate and lysates were separated by centrifugation at 13, 000×g for 15 min. The lysates were quantified by BCA assay. There are two ways to do this assay, one is that at least 600 μg of the lysates incubated with 10 μl of anti-Flag-conjugated gel (Bimake, B23101) in gyratory shaker overnight at 4°C. The immunoprecipitates were washed three times with TBST buffer (20 mM Tris, 150 mM NaCl, pH 7.6, 0.1% Tween 20 [Mikxlife, TC279]) and eluted with 10 μl of 2×loading buffer, heated on a heat blocker at 95°C for 5 min. Another is that the antibody was pre-coupled to 20 μl of protein A/G magnetic beads (Bimake, B23202) in a rotating shaker at room temperature for 40 min, and then add at least 600 μg of lysates to incubate for another 40 min. Next the mixture was put on a magnetic rack, after discarding the supernatant, the beads were washed for 3 times with TBST buffer and then transferred to another clean 1.5 ml tube. Finally, 20 μl of 1× loading buffer was added to elute the protein, tubes were put on a heat blocker at 95°C for 5 min. Samples were analyzed by SDS-PAGE and immunoblotting with corresponding antibodies.

For the in cell ubiquitination assay, cells were transiently transfected with plasmids expressing HA-Ub for 24–48 h, or co-transfected with other plasmids. Cells were treated with MG132 for 6 h before collection followed by the operation of immunoprecipitation assay.

GST affinity-isolation assay

Recombinant plasmids encoding GST-tagged ATG4B, GST-vector, His-tagged SUMO-HECT (UBE3C), His-tagged ATG4B and its mutants were transformed into E. coli BL21 (DE3) and a single clone was picked and massively expanded in the shaker (37°C, 3 h, 220 rpm), and then induced by isopropyl β-D-1-thiogalactopyranoside (IPTG [Sangon Biotech, A600168], 0.5 mM, 16°C, 20 h). Then His-SUMO-HECT (UBE3C), His-tagged ATG4B and its mutants were purified with Ni-Charged MagBeads (Genscript, L00295), after purification, the His-tagged SUMO was cleaved with ULP1 enzyme (a gift from Huihao Zhou, Sun Yat-sen University). GST-tagged ATG4B and GST-vector were purified with GST agarose gel (YuanYe, S14059) according to the instruction manual.

GST-ATG4B or GST control protein were incubated with GST agarose gel in TBST buffer at room temperature. After 1 h of incubation, HECT (UBE3C) protein was added, and thoroughly mixed in vibrating mixer for overnight incubation at 4°C. Immunoprecipitates were then washed for five time and analyzed by western blotting.

In vitro ubiquitination assay

One μL of 10×Reaction Buffer, 1 μL of 10×E1 Enzyme, 1 μLof 10×Mg²⁺-ATP, 2 μL of 5×ubiquitin, 0.5 μL of 20×UBE2D1, 1 mM of DTT, 5 μg of His-ATG4B and 5 μg of HECT were mixed in a 1.5-ml centrifuge tube. Then water was added to a total volume of 10 μL. The tube was incubated at 37°C for 120 min. Then 2.5 µL of 5×SDS-PAGE Loading Buffer was added to terminate the reactions. Samples were heated at 95°C for 5 min prior to SDS-PAGE and western blotting analysis using anti-Ub antibody.

Immunofluorescence assay

Cells cultured on a petri dish were washed once with PBS (Servicebio, G0002), fixed with 4% paraformaldehyde for 20 min followed by permeabilization with 0. 1% Triton X-100 (Sangon Biotech, A600198–0500) for 30 min and then blocked in 5% goat serum (Boster Biological Technology, AR0009) for 1 h. The above was done at room temperature. Cells were next incubated with primary antibodies (1: 200) at 4°C overnight and then secondary antibodies (1: 500) at room temperature for 1 h after washed with PBS for 3 times. Fluorescence was visualized by Confocal Laser Scanning Microscope (Olympus, FV3000) or EVOS FL Auto Cell Imaging System (Thermo Fisher Scientific).

Mass spectrometry analysis of ATG4B

Mass spectra measurement was performed as reported [39]. Briefly, Flag-ATG4B and HA-Ub plasmids were overexpressed in 293T cells as the experimental group, and Flag-vector and HA-Ub plasmids were overexpressed as the control group. Cells were treated with MG132 for 6 h before collection and total lysates were immunoprecipitated with anti-Flag antibodies. The Flag immunoprecitates were subject to SDS-PAGE analysis followed by Coomassie Brilliant Blue staining. The SDS-PAGE gel was then submitted for tryptic digestion and LC-MS/MS analysis. Gel bands were manually excised and then divided into a number of smaller pieces about 1 mm³ size. The gel pieces were destained and then washed with water and dehydrated in acetonitrile (Macklin, A801556-1 L). The bands were then reduced with DTT (Sangon Biotech, A100281–0005) and alkylated with iodoacetamide (Aladdin, I131590) prior to in-gel digestion. Use trypsin (Thermo Fisher 90,057) to digest all bands in-gel by adding 1–2 μg of trypsin in 50 mM ammonium bicarbonate (Aladdin, A431333) to a total volume of 50 µl and incubating the digestions overnight at 37°C to achieve complete digestion. The formed peptides were extracted from the digestions and dissolved in 200 μl of acetonitrile with 0.1% FA (Energy Chemical, W810042), collect the solution and repeat once again. These extracts were combined and evaporated to dryness in a speedvac. It is then resuspended in 0.1% acetic acid (Aladdin, A116168) to bring the final volume to about 10 μl for LC-MS analysis.

Separation of peptides via reverse chromatographic column (C18-AQ, 1.9 µm, PF360-75-10-N-5, 20 cm), mobile phase A (aqueous) contained 0.1% formic acid in water, and mobile phase B (organic) contained 0.1% formic acid in 80% acetonitrile, gradient was set to 0 ~ 95 min, 3%→32% (B); 95 ~ 105 min, 32%→100% (B); 105 ~ 120 min, 100% (B), the flow rate was 200 nL/min. Then the components were electrosprayed by the NSI ion source and detected by the mass spectrometer. The spray voltage of NSI was 2.1 kV, and the temperature of the ion transfer tube was 250°C, the first order mass spectra were acquired at a resolution of 70,000 and scanned from 355–1700 m/z. The MS spectra were acquired at a resolution of 35,000; CID collision energy, 27%; Loop count, 10; AGC target, 5e4; Isolation window, 1.6 m/z; Fixed first mass, 120 m/z; Exclusion duration, 60 s.

Acquired tandem mass spectra were analyzed by Protein Discovery 2.2.0.388 (Thermo Fisher Scientific) and searched against the SwissProt human protein database (TaxID = 9606). The maximum missed cleavage site, 2; the minimum and maximum peptide lengths, 5 and 50 respectively; the mass error of the precursor ion, 20 ppm; the mass error of the fragment ion, 0.02 Da; the maximum number of peptide modifications, 3. The iodoacetamide modification of cysteine (+57.021 Da) was set as a fixed modification, the false positive rate FDR was set to 1%, and the peptide site positioning probability less than 75% would be excluded.

Bioinformatics analysis

There are 1268 proteins identified in the experimental group. We used R software (R 4.2.0) to screen for E3 ligase which could regulate ATG4B ubiquitination. We used the VennDiagram R package to draw the Venn diagram suggesting nonspecific proteins. The pheatmap R package was used to draw the heat map for the proteins detected by the experimental group at least twice in the six repeated tests. The proteins with higher occurrence frequency (n ≥3) in the experimental group were further analyzed with GO:BP and KEGG analysis by the clusterProfiler R package to perform functional profiles. We focused on ubiquitin-related pathways, so the GOplot R package was used to draw the chord diagram for the enriched proteins and corresponding pathways.

Gel-based assay

Gel-based assay for ATG4B activity was performed as reported [2]. Cells were collected and homogenized in non-denaturing lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA). Cell lysates were incubated with LC3-GST or FRET-GABARAPL2 substrates in reaction buffer (25 mM Tris-HCl, pH 7.4, 50 mM KCl) for 30 min at 37°C. Then the reaction was stopped by addition of reduced loading buffer and separated by SDS-PAGE. Results were analyzed by CBB staining. The remained amounts of LC3-GST or FRET-GABARAPL2, the cleaved GST and LC3, CFP-GABARAPL2 and YFP products were analyzed by ImageJ to determine the ATG4B activity.

Nonradioactive quantification of autophagic protein degradation with L-azidohomoalanine labeling assay

Nonradioactive long-lived protein degradation assay was performed as described [37]. HeLa cells were transfected with sh-UBE3C or His-UBE3C plasmids for 4 h, and methionine free medium was replaced for 30 min and the cells were cultured with AHA (50 µM [MCE, HY-140346A]) in methionine-free DMEM with 10% (v:v) dialyzed FBS for 20 h. Then, cells were incubated in regular DMEM with 10% FBS containing 10X L-methionine (2 mM) for another 2 h. The cells were then incubated in EBSS medium supplemented with 0.1% (w:v) BSA and 10×L-methionine (2 mM) for 12 h, then collect the cells and fix the cells. The cells were permeablized with 0.5 ml of 0.5% (v:v) Triton X-100 in PBS and suspended the cells in 200 µl of reaction master mix (50 µM TAMRA azide [AAT Bioquest, AAT-487], 1 mM TCEP [Calbiochem 580,560], 100 µM TBTA [GLPBIO, GC45003] and 1 mM CuSO₄ in PBS). Incubate the samples in the dark for 2 h at room temperature under constant agitation (Eppendorf mixer, 700 rpm/m). Wash the cells with PBS and pellet them. And then relative fluorescence intensity of the cells were analyzed by flow cytometry. In this method, the lower intensity means the higher protein degradation activity.

Statistical analysis

All data were obtained from at least three independent experiments and expressed as the mean ± SEM of at least 3 independent experiments. Statistical analyses were performed using the Student’s two-tailed t-test, with P-value <0. 05 considered significant. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 versus the corresponding controls were indicated. All statistical calculations were performed using Graphpad prism 8.2.1.

Funding Statement

The author(s) reported there is no funding associated with the work featured in this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2023.2299514