TBC1D4 antagonizes RAB2A-mediated autophagic and endocytic pathways

ABSTRACT

Macroautophagic/autophagic and endocytic pathways play essential roles in maintaining homeostasis at different levels. It remains poorly understood how both pathways are coordinated and fine-tuned for proper lysosomal degradation of diverse cargoes. We and others recently identified a Golgi-resident RAB GTPase, RAB2A, as a positive regulator that controls both autophagic and endocytic pathways. In the current study, we report that TBC1D4 (TBC1 domain family member 4), a TBC domain-containing protein that plays essential roles in glucose homeostasis, suppresses RAB2A-mediated autophagic and endocytic pathways. TBC1D4 bound to RAB2A through its N-terminal PTB2 domain, which impaired RAB2A-mediated autophagy at the early stage by preventing ULK1 complex activation. During the late stage of autophagy, TBC1D4 impeded the association of RUBCNL/PACER and RAB2A with STX17 on autophagosomes by direct interaction with RUBCNL via its N-terminal PTB1 domain. Disruption of the autophagosomal trimeric complex containing RAB2A, RUBCNL and STX17 resulted in defective HOPS recruitment and eventually abortive autophagosome-lysosome fusion. Furthermore, TBC1D4 inhibited RAB2A-mediated endocytic degradation independent of RUBCNL. Therefore, TBC1D4 and RAB2A form a dual molecular switch to modulate autophagic and endocytic pathways. Importantly, hepatocyte- or adipocyte-specific tbc1d4 knockout in mice led to elevated autophagic flux and endocytic degradation and tissue damage. Together, this work establishes TBC1D4 as a critical molecular brake in autophagic and endocytic pathways, providing further mechanistic insights into how these pathways are intertwined both in vitro and in vivo.

Abbreviations: ACTB: actin beta; ATG9: autophagy related 9; ATG14: autophagy related 14; ATG16L1: autophagy related 16 like 1; CLEM: correlative light electron microscopy; Ctrl: control; DMSO: dimethyl sulfoxide; EGF: epidermal growth factor; EGFR: epidermal growth factor receptor; FL: full length; GAP: GTPase-activating protein; GFP: green fluorescent protein; HOPS: homotypic fusion and protein sorting; IP: immunoprecipitation; KD: knockdown; KO: knockout; LAMP1: lysosomal associated membrane protein 1; MAP1LC3B/LC3B: microtubule associated protein 1 light chain 3 beta; OE: overexpression; PG: phagophore; PtdIns3K: class III phosphatidylinositol 3-kinase; SLC2A4/GLUT4: solute carrier family 2 member 4; SQSTM1/p62: sequestosome 1; RUBCNL/PACER: rubicon like autophagy enhancer; STX17: syntaxin 17; TAP: tandem affinity purification; TBA: total bile acid; TBC1D4: TBC1 domain family member 4; TUBA1B: tubulin alpha 1b; ULK1: unc-51 like autophagy activating kinase 1; VPS39: VPS39 subunit of HOPS complex; WB: western blot; WT: wild type.

Introduction

Macroautophagy (hereafter referred to as autophagy) is an evolutionarily conserved lysosomal degradative pathway, which plays fundamental roles in development and homeostasis [1–3]. Dysfunction of autophagy is tightly connected with major human diseases [4–6]. Morphologically, autophagy is initiated by the formation of a phagophore (PG). After nucleation, this double-membrane structure further expands and ultimately seals to generate an autophagosome, which then fuses with a lysosome leading to the degradation of autophagy cargoes [7–10]. How the autophagosome forms and matures into an autolysosome remain to be the fundamental unresolved questions in the autophagy field.

The de novo synthesis of autophagosomes acquires membranes from multiple sources, including ER [11,12], the Golgi network [13–16], mitochondria [17], the plasma membrane [9], endosomes [18–20] and ER-Golgi intermediate [21–24]. We recently uncovered that RAB2A links the Golgi apparatus to autophagosome formation. First, RAB2A directionally transports Golgi-derived ATG9 (autophagy related 9)-positive vesicles to the PG formation sites for the construction of early autophagic structures [25]. This route may exist in parallel with previously characterized pathways in which plasma membrane- and recycling endosome-derived ATG9⁺ vesicles participate in autophagosome biogenesis [18,19,26]. Second, RAB2A facilitates ULK1 (unc-51 like autophagy activating kinase 1) activation to propagate signals for autophagy initiation. The acquisition of ULK1 to the RAB2A⁺ early autophagic membrane structures may enhance the clustering of the ULK1 complex, which inevitably leads to ULK1 activation [27–30]. Next, activated ULK1 phosphorylates ATG9 to enable these Golgi-derived vesicles to fuse into PGs [31–34], and phosphorylates the ATG14 (autophagy related 14)-containing class III phosphatidylinositol 3-kinase (PtdIns3K) complex to activate it for PG nucleation [35–37].

Independent studies from several groups discovered that RAB2A regulates the formation of autolysosomes in both Drosophila and mammals [25,38–40]. The consensus of these works is that the autophagosomal RAB2A recruits the homotypic fusion and protein sorting (HOPS) complex to facilitate the tethering of autophagosomes with lysosomes, a pivotal step prior to soluble N-ethylmaleimide-sensitive factor attachment protein receptor/SNARE complex-mediated fusion of these two organelles. We provide further insights into this key regulatory process by proposing an autophagosomal trimeric protein complex containing RAB2A, RUBCNL and STX17 for specifying the HOPS recruitment [25,41–43]. Moreover, a series of studies, including ours, demonstrated that RAB2A is essential for endocytic and phagocytic pathways [38,39,44–46]. It remains unclear how RAB2A coordinates these processes and how its functions are modulated.

As a member of the RAB GTPases family, RAB2A also requires guanine nucleotide exchange factor/GEF to promote the release of GDP and subsequent binding to GTP. The GTPase-activating protein (GAP) stimulates the intrinsic GTPase activity of RAB GTPases to hydrolyze GTP, inactivating RAB GTPases. RAB GTPases switch back and forth between these two forms in response to different stimulus signals [47–49]. TBC1D4, a TBC domain-containing protein that can inactivate many RAB GTPases, including RAB2A [50]. While TBC1D4 is well-characterized as a GAP, previous research has predominantly focused on its negative regulation of SLC2A4/GLUT4 translocation to the plasma membrane [51–55]. Notably, there is a significant gap in the literature regarding whether TBC1D4 exerts any influence on the autophagy pathway. Here, we propose an updated model in which TBC1D4, a protein that plays critical roles in glucose homeostasis, inhibits RAB2A-mediated autophagic and endocytic pathways.

Results

Tandem affinity purification (TAP)-coupled mass spectrometry identifies TBC1D4 as an interactor of RUBCNL

Our previous mass spectrometry (MS) analysis of the proteins that co-immunoprecipitated (co-IP) with RUBCNL in TAP resulted in the identification of TBC1D4 as a candidate RUBCNL interactor (Figure S1) [41]. Indeed, co-IP experiments indicated that endogenous RUBCNL and RAB2A GTPase were associated with TBC1D4 under unstressed conditions (Figure 1A). In contrast, no interaction was observed for other RUBCNL-interacting proteins such as VPS39, the key component of HOPS complex, BECN1 and UVRAG, subunits of the PtdIns3K, and the autophagosomal soluble N-ethylmaleimide-sensitive factor attachment protein receptor, STX17. Next, we dissected the interaction between RUBCNL and TBC1D4 by co-IP experiments using different RUBCNL mutants and we observed that the RH domain of RUBCNL is dispensable for TBC1D4 interaction (Figure 1B,E). RUBCNL amino acids (aa) 1–400 but not RUBCNL aa1–300 appeared to interact with TBC1D4, so we postulated that the region ranging from aa 301 to 400 of RUBCNL might be important for their association. Indeed, RUBCNL aa 301–400 was required and sufficient for its binding with TBC1D4 (Figure 1C–E). Using the same strategy, we uncovered that TBC1D4 utilizes its PTB1 domain to interact with RUBCNL (Figure 1F,G). The association of RUBCNL and TBC1D4 appeared to be direct, as purified recombinant proteins for TBC1D4 PTB1 and RUBCNL [301–400] exhibited robust interaction in an affinity-isolation assay (Figure 1H). Together, these results showed that TBC1D4 directly interacts with RUBCNL.

Figure 1.

Identification TBC1D4 as an interactor of RUBCNL. (A) U2OS stably expressing FLAG-TBC1D4 was applied for co-immunoprecipitation (Co-IP) analysis of the interaction of TBC1D4 with endogenous RUBCNL, RAB2A, UVRAG, VPS39, STX17 and BECN1. (B) FLAG-tagged full-length RUBCNL or mutants were co-expressed with HA-TBC1D4 individually in HEK293T cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (C) WT RUBCNL or RUBCNL [301–400] was co-expressed with HA-TBC1D4 individually in HEK293T cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (D) WT RUBCNL or RUBCNL [Δ301–400] was co-expressed with HA-TBC1D4 individually in HEK293T cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (E) Schematic of RUBCNL mutants. (F) Schematic of TBC1D4 mutants. (G) FLAG-tagged full-length TBC1D4 or mutants were co-expressed with HA-RUBCNL individually in HEK293T cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (H) Recombinant FLAG-RUBCNL [301–400], GST-PTB1, GST-PTB2, and GST-(PTB1+PTB2) were purified as indicated. In vitro GST affinity isolation was performed. Samples were analyzed by anti-FLAG western blot or Coomassie Brilliant Blue staining.

TBC1D4 forms a trimeric complex with RAB2A and RUBCNL

The fact that TBC1D4 was able to pull down RAB2A, but not STX17, BECN1-UVRAG or VPS41-VPS39 (Figure 1A) indicated that TBC1D4 likely forms a complex with RUBCNL and RAB2A instead of with PtdIns3K or HOPS complexes. We then confirmed their colocalization using confocal microscopy imaging analysis under the condition that GFP-TBC1D4 was expressed close to its endogenous levels (Figure 2A). Next, co-IP experiment showed that TBC1D4, similar to RUBCNL [25], interacted with RAB2A WT and mutants, with a higher affinity for the GDP form of RAB2A (RAB2A^N119I) (Figure 2B). As expected, TBC1D4 colocalized with RAB2A WT and mutants (Figure 2C). Further dissection of their interaction by co-IP experiments using different forms of TBC1D4 mutants indicated that the N-terminal PTB2 domain, but not the C-terminal TBC domain of TBC1D4, was required for RAB2A interaction (Figure 1F,D). In vitro affinity-isolation assays using purified recombinant proteins demonstrated that the PTB2 domain of TBC1D4 directly interacted with RAB2A (Figures 2E, S2A and S2B). The TBC1D4 mutants containing the PTB2 colocalized with RAB2A, which further confirmed the importance of PTB2 in the physical interaction (Figure 2F,G). The PTB domain plays a crucial role in facilitating the precise localization of TBC family proteins to specific vesicles [56]. To confirm the specificity of the PTB2-RAB2A interaction, we conducted additional experiments with several other RABs as controls. The findings revealed varying degrees of interaction between the PTB2 domain of TBC1D4 and these control RABs. Notably, the interaction with RAB2A remained the most robust among them (Figure S2C). Given that RUBCNL directly binds to the GDP form of RAB2A [25] and that the PTB1 and PTB2 domains of TBC1D4 directly interact with RUBCNL and RAB2A independently, we postulated that TBC1D4, RAB2A and RUBCNL may form a trimeric complex. To prove this notion, we performed biochemical fractionation assay using cytoplasmic extracts, and we showed that TBC1D4 co-migrated with RAB2A and RUBCNL in size-exclusion chromatography (Figures S2D and S2E). Together, TBC1D4 forms a complex with RAB2A and RUBCNL by excluding HOPS, PtdIns3K and STX17.

Figure 2.

TBC1D4 binds to RAB2A through its PTB2 domain. (A) Inducible GFP-TBC1D4 U2OS was treated with 1 μg/ml doxycycline for 12 h, so that GFP-TBC1D4 expressed in cells close to endogenous levels. Cells fixed and stained with anti-RUBCNL antibody and anti-RAB2A antibody, analyzed by confocal microscopy. Scale bars: 10 µm. (B) FLAG-tagged WT RAB2A, RAB2A^Q65L or RAB2A^N119I was co-expressed with HA-TBC1D4 individually in HEK293T cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (C) TBC1D4 KO U2OS cells co-expressed with GFP-RAB2A mutants and HA-TBC1D4 were fixed and stained with anti-HA antibody, analyzed by confocal microscopy. Scale bars: 10 µm. (D) FLAG-tagged full-length TBC1D4 or mutants were co-expressed with HA-RAB2A^N119I individually in HEK293T cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (E) Recombinant FLAG-RAB2A WT, GST-PTB1, GST-PTB2, and GST-(PTB1+PTB2) were purified from E. Coli. In vitro GST affinity-isolation was performed. Samples were prior to lysis and analyzed by anti-FLAG western blot. (F) TBC1D4 mutants PTB1-HA, PTB2-HA or (PTB1+PTB2)-HA was co-transfected with GFP-RAB2A^N119I in U2OS cells. Cells were then fixed and stained with anti-HA antibody. Cells were analyzed by confocal microscopy and quantified for the co-localization of TBC1D4 mutants and GFP- RAB2A^N119I in (G). Scale bars: 10 µm. Data are shown as mean ± SD (***p < 0.001).

TBC1D4 is an inhibitor of autophagy

Given that RUBCNL and RAB2A are positive regulators and that TBC1D4 was shown to be a putative inhibitor of autophagy in a screen study [18,25,41], we hypothesized that TBC1D4 May have a role in autophagy. To test this idea, we first measured the autophagic degradation of SQSTM1/p62 in TBC1D4 overexpression (OE) cells using a panel of known autophagy regulators as controls. We observed that the overexpression of RUBCNL and WT RAB2A reduced SQSTM1 levels, which was consistent with our previous studies [25]. In contrast, TBC1D4 overexpression resulted in accumulation of SQSTM1 in a manner similar to RAB2A GTP or GDP mutants (Figure 3A–C), indicating that TBC1D4 overexpression impaired autophagy flux under unstressed conditions. Importantly, TBC1D4 overexpression was also able to significantly delay the SQSTM1 degradation stimulated by torin1, a potent MTOR inhibitor (Figure 3D,E). If TBC1D4 overexpression inhibited autophagy, the absence of TBC1D4 would be expected to increase autophagy flux. Indeed, TBC1D4 knockout (KO) by CRISPR-Cas9 or knockdown (KD) by shRNA significantly reduced SQSTM1 levels (Figure 3F,G, S3A and S3B). To further verify that TBC1D4 affects autophagic flux, we used the GFP-LC3B-RFP-LC3B∆G reporter [3,57]. The decrease in GFP:RFP ratio represents an increase in autophagic flux. Flow cytometry analysis showed that TBC1D4 KO significantly reduced GFP:RFP ratio upon autophagy induction (Figure 3H). Furthermore, we employed flow cytometry analysis using the Halo-GFP reporter [58,59], a method that provides a more accurate assessment of bulk autophagic flux compared to MAP1LC3B/LC3B-based approaches. Although the degradation amplitude was considerably lower than LC3B-dependent autophagic flux, the statistical results convincingly support the conclusion that TBC1D4 KO promotes autophagic flux (Figure 3I). Given TBC1D4’s role as the GAP for RAB2A, we questioned whether its GAP activity is the sole reason for autophagy inhibition. To explore this hypothesis, we utilized the R972K mutant, which lacks GAP activity [50,51,60], to assess its impact on autophagy. The results revealed that the mutant maintains its interaction with RAB2A, with the strongest interaction observed with the GDP-bound form, albeit with an overall decrease in interaction (Figure S3E). The R972K mutant retains the ability to inhibit autophagy, although the inhibitory effect is diminished, suggesting that TBC1D4’s mechanism for autophagy inhibition is not solely attributed to RAB inactivation (Figure 3J,K). These results demonstrate that TBC1D4 is an autophagy inhibitor under both basal and stress-induced conditions.

Figure 3.

TBC1D4 is an inhibitor of autophagy. (A) HEK293T cell lines was transfected with vector, FLAG-RUBCNL, FLAG-RUBCN, FLAG-RAB2A, FLAG-RAB2A^Q65L,FLAG-RAB2A^N119I and HA-TBC1D4 were analyzed for their autophagy activity by measuring the relative levels of SQSTM1. (B) TBC1D4 WT and TBC1D4 OE cell lines were fixed and analyzed by confocal microscopy, and quantified for the number of endogenous SQSTM1 puncta in (C). Scale bars: 10 µm. Data are shown as mean ± SD (**p < 0.01). (D) TBC1D4 WT and TBC1D4 OE cell lines treated with torin1 as indicated. Autophagy activity was measured by analyzing SQSTM1 level. (E) TBC1D4 WT and TBC1D4 OE cell lines treated with torin1 or bafilomycin A₁ as indicated. Autophagy activity was measured by analyzing SQSTM1 levels. (F) TBC1D4 WT, TBC1D4 KO (#12) and TBC1D4 KO (#41) U2OS cell lines were harvest, and lysates were analyzed by western blot with anti-SQSTM1 antibody. (G) TBC1D4 WT and TBC1D4 KD HEK293T cell lines were harvest, and lysates were analyzed by western blot with anti-SQSTM1 antibody. (H) Flow cytometry analysis of GFP and RFP intensities in TBC1D4 WT and TBC1D4 KO cells expressing GFP-LC3B-RFP-LC3BΔG. Cells treated with or without torin1 for 6 h. The ratio of GFP:RFP decrease percentage in the treatment group compared to the control group. Data are shown as mean ± SD (** p < 0.01). (I) Flow cytometry analysis of GFP and TMR intensities in TBC1D4 WT and TBC1D4 KO cells expressing Halo-GFP. Cells treated with 80 nM TMR ligand and DMSO or torin1 for 6 h. Statistical analysis of GFP/TMR decrease percentage in the treatment group compared to the control group. Data are shown as mean ± SD, *** p < 0.001. (J) HEK293T cells transfected with vector, TBC1D4 WT and TBC1D4^R972K. After transfection for 24 h, cells treated with torin1 as indicated. Autophagy activity was measured by analyzing SQSTM1 levels. Comparing the grayscale ratio of SQSTM1 to ACTB, and normalized using Vector (0 h) as the standard. Statistical results in (K). Data are shown as mean ± SD (* p < 0.05, ** p < 0.01).

TBC1D4 overexpression induces aberrant structures indispensable of RAB2A

To understand the mechanism underlying TBC1D4-mediated autophagy inhibition, we analyzed which steps of autophagy could be hampered by TBC1D4. We observed that TBC1D4 overexpression induced large membrane structures that were positive for autophagy markers at varying overlaying ratios (Figure 4A,B). The colocalization of TBC1D4 with ATG9, ATG2, ATG14, WIPI2 or ATG5 was significantly higher than that with LC3B. Remarkably, these TBC1D4-marked vesicles largely excluded STX17 and LAMP1 (Figure 4A,B), implying that TBC1D4 expression arrested autophagy progression at different stages prior to fusion with lysosomes. In our previous study, we showed that Golgi-derived RAB2A translocates to early and later autophagic structures, and thereby playing crucial roles in the formation of both autophagosomes and autolysosomes in mammalian cells [25]. Because TBC1D4 colocalized with RAB2A and distributed in a similar pattern (Figure 2C), we reasoned that TBC1D4 overexpression may depend on RAB2A to induce these aberrant membrane structures. Indeed, RAB2A KO abolished the TBC1D4-positive membrane structures (Figure 4C,D). In RAB2A KO cells, TBC1D4 failed to colocalize with autophagy markers (Figure S4). To understand the nature of these aberrant structures caused by overexpression of TBC1D4, we conducted correlative light-electron microscopy (CLEM) analysis, which revealed that a large number of small vesicles were captured within these aberrant structures, while the rest were protein collections (Figure 4E). It has been reported that full-length TBC1D4 can form oligomers of approximately 600 kDa, and its self-interaction requires two PTB domains at the N terminus and 127 amino acids at the C-terminus [61,62]. Moreover, TBC family proteins have been reported to act as scaffolds to promote the assembly of complexes and subsequent functional advancement [63,64]. In summary, we speculate that overexpression of TBC1D4 exacerbates the aggregation of TBC1D4 oligomers, leading to continuous accumulation of RAB2A-positive vesicles. Because RAB2A participates in early autophagy (Figure 4A,B), the early autophagic molecules carried by RAB2A-positive vesicles are also trapped in such aberrant structures, thereby affecting the initiation and progression of autophagy. These results suggest that TBC1D4 relies on RAB2A to associate with early autophagic membrane structures and that TBC1D4 and RAB2A may antagonize each other in autophagy regulation.

Figure 4.

TBC1D4 depends on RAB2A to colocalize with autophagic markers. (A) U2OS cells were co-transfected with HA-TBC1D4 and autophagic markers including (GFP-ATG2 or mCherry-STX17). Endogenous WIPI2, ATG14, ATG5, ATG9, LC3B and LAMP1 were stained in HA-TBC1D4 expressing U2OS cells. Colocalization of TBC1D4 and autophagic markers was analyzed by confocal microscopy, and quantification was shown in (B). Scale bars: 10 µm. Data are shown as mean ± SD (* p < 0.05, ** p < 0.01, ***p < 0.001). (C) HA-TBC1D4 was expressed in RAB2A WT, RAB2A KO and RAB2A OE cells. The percentage of cells with TBC1D4 vesicles was confocal microscopy analyzed and quantified in (D). Scale bars: 10 µm. Data are shown as mean ± SD, ***p < 0.001. (E) Analysis of the aberrant structure caused by overexpression of TBC1D4 by CLEM. The white arrow indicates small vesicles. U2OS cells grown on glass gridded coverslips were transfected with EGFP-TBC1D4 and imaged by Zeiss Airyscan to collect light microscopy images. Samples were then prepared for imaging by FIB-SEM. Light and electron microscope images were superimposed. Scale bars: 10 µm.

TBC1D4 inhibits autophagy initiation by antagonizing ULK1 complex activation

The colocalization of TBC1D4 with early autophagy markers suggested a potential function in autophagy initiation. Indeed, we observed that LC3B puncta were significantly increased in TBC1D4 KO cells under both autophagy-stimulating and non-stimulating conditions, but this phenomenon was reversed by RAB2A KD (Figure 5A,B), indicating that TBC1D4 impeded LC3B lipidation. To further strengthen these notions, we performed GFP-LC3B-RFP-LC3BΔG cleavage assays (Figure 5C,D) [3,57], and we showed that TBC1D4 KD significantly increased the levels of GFP-LC3B-II and accordingly enhanced the release of free GFP under both unstressed and torin1-stimulated conditions. LC3B lipidation is mainly catalyzed by ATG12–ATG5-ATG16L1 on the expanding PG membrane [65]. Indeed, membrane recruitment of endogenous ATG16L1 (autophagy related 16 like 1) was increased in TBC1D4 KO cells (Figure 5E,F). Similarly, TBC1D4 KO enhanced the formation of the earliest autophagic structures labeled by ULK1 (Figure 5G,H). But both puncta of ATG16L1 and ULK1 were reduced by RAB2A KD (Figure 5E–H). We previously demonstrated that RAB2A interacts with ULK1 to promote the clustering of ULK1-ATG13 complexes, a process that is essential for ULK1 activation [25]. Conversely, the interaction of ULK1 and ATG13 was suppressed by TBC1D4 OE or enhanced by TBC1D4 KO (Figure 5I). As a result, TBC1D4 was capable of modulating ULK1 activity, which was evidenced by the altered phosphorylation of ULK1 and ATG14 (substrate of ULK1) in TBC1D4 OE or KO cells (Figure 5J). Consistently, TBC1D4 disrupted the interaction between RAB2A and ULK1 (Figure 5K), and TBC1D4-deficiency induced ULK1 activation was ablated by RAB2A KD (Figure 5L). The phosphorylation of various autophagy machineries, including ATG14, by ULK1 is critical for autophagy initiation, we conclude that TBC1D4 blocks RAB2A-mediated ULK1 activation and signal propagation to impede autophagy initiation.

Figure 5.

TBC1D4 impedes autophagy initiation by blocking ULK1 activation. (A) WT, TBC1D4 KO, RAB2A KO and TBC1D4 KO RAB2A KD U2OS cell lines were treated with DMSO or torin1 for 2 h. Cells were fixed and analyzed by confocal microscopy, and quantified for the number of endogenous LC3B puncta in (B). Scale bars: 10 µm. Data are shown as mean ± SD (****p < 0.0001). (C)Western blot analysis of lipidated GFP-LC3B and free GFP in TBC1D4 WT, TBC1D4 KD and TBC1D4 OE cells expressing GFP-LC3B-RFP-LC3BΔG. Cells were treated with torin1 as indicated. Measure the grayscale value and statistically analyze the relative ratio of free GFP:total GFP in (D). Data are shown as mean ± SD (* p < 0.05, ** p < 0.01). (E) Endogenous ATG16L1 puncta was analyzed in WT, TBC1D4 KO, RAB2A KO and TBC1D4 KO RAB2A KD U2OS cell lines and the number of ATG16L1 puncta was quantified in (F). Scale bars: 10 µm. Data are shown as mean ± SD (****p < 0.0001). (G) HA-ULK1 was expressed in WT, TBC1D4 KO, RAB2A KO and TBC1D4 KO RAB2A KD U2OS cell lines. Cells were treated with DMSO or torin1 for 2 h and the number of ULK1 puncta was quantified in (H). Scale bars: 10 µm. Data are shown as mean ± SD (** p < 0.01). (I) FLAG-ATG13 was transfected in WT, TBC1D4 OE and TBC1D4 KO cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (J) Phosphorylation of ULK1 and ATG14 were detected in TBC1D4 WT, TBC1D4 OE and TBC1D4 KO cell lines by anti-p-ULK1 (S555) and pATG14 (S29) antibodies. (K) FLAG-RAB2A was transfected in TBC1D4 WT, TBC1D4 OE and KO cells, and anti-FLAG immunoprecipitation was performed and analyzed by western blot. (L) SQSTM1, phosphorylation of ULK1 and ATG14 were detected in WT, TBC1D4 KO, RAB2A KO and TBC1D4 KO RAB2A KD U2OS cell lines by western blot.

TBC1D4 impairs autolysosome formation by impeding HOPS recruitment.

We have shown that the autophagosomal trimeric complex containing RAB2A, STX17 and RUBCNL specifies the recruitment of HOPS complex to facilitate autophagosome maturation [25]. Therefore, we investigated whether TBC1D4 has a role in autolysosome formation. To do this, we first performed autophagosome maturation assays using the tandem mCherry-GFP-LC3B construct [3] (Figure 6A), and we observed that TBC1D4 OE significantly reduced autophagosome maturation, as measured by the percentage of mCherry-positive GFP-negative (mCherry⁺ GFP⁻) puncta in U2OS cells (Figure 6B,C). Furthermore, the colocalization of LC3B and LAMP1 puncta, the indicator for autophagosome-lysosome fusion, was dramatically reduced in TBC1D4 OE cells (Figure 6D,E). The defect in autolysosome formation was due to inefficient HOPS recruitment, which was proved by reduced colocalization of RAB2A and VPS39, a critical component of HOPS complex, under TBC1D4-overexpressing conditions (Figure 6F,G). Consistently, the interaction of RAB2A and VPS39 or STX17 was increased in TBC1D4 knockdown cells (Figure 6H). Similarly, the colocalization of RUBCNL and VPS39 were reduced under TBC1D4 overexpressing situation (Figure 6I,J), but TBC1D4 KD facilitated the interaction between RUBCNL and VPS39 or STX17 (Figure 6K). Because TBC1D4 interacts with RUBCNL and RAB2A, but not with STX17, and TBC1D4 overexpression reduced the colocalization of RAB2A with HOPS complex and STX17, we conclude that TBC1D4 abrogates formation of the tripartite complex of RUBCNL, RAB2A and STX17 (Figure 6L), leading to the reduced recruitment of HOPS complex and consequentially the formation of autolysosomes.

Figure 6.

TBC1D4 prevents autolysosome biogenesis by disrupting the autophagosome recruitment of HOPS complex. (A) mCherry-GFP-LC3B was expressed in TBC1D4 WT and TBC1D4 OE U2OS cells, respectively. Cells were treated with DMSO or torin1 for 2 h. LC3B was monitored by confocal fluorescence microscope. mCherry⁺ GFP⁻ puncta, which indicate autolysosomes, were quantified and were summarized in (B) and (C). Scale bars: 10 µm. Data are shown as mean ± SD (*p < 0.05, **p < 0.01). (D) Endogenous LAMP1 and LC3B in TBC1D4 WT and TBC1D4 OE U2OS cells were monitored by fluorescence microscope and the colocalization percentage was quantified in (E). Scale bars: 10 µm. Data are shown as mean ± SD (***p < 0.001). (F) mCherry-RAB2A and FLAG-VPS39 were co-transfected in TBC1D4 WT and TBC1D4 OE U2OS cells. Colocalization of RAB2A and VPS39 was monitored by fluorescence microscope, and quantification was shown in (G). Scale bars: 10 µm. Data are shown as mean ± SD (***p < 0.001). (H) FLAG-RAB2A was expressed in TBC1D4 WT and TBC1D4 KD HEK293T cells, anti-FLAG IP was performed and endogenous VPS39, STX17 and RUBCNL were detected by western blot. (I) GFP-RUBCNL and FLAG-VPS39 were co-transfected in TBC1D4 WT and TBC1D4 OE U2OS cells. Colocalization of RUBCNL and VPS39 was monitored by fluorescence microscope, and quantification was shown in (J). Scale bars: 10 µm. Data are shown as mean ± SD (* p < 0.05). (K) FLAG-RUBCNL was expressed in TBC1D4 WT and TBC1D4 KD HEK293T cells, anti-FLAG IP was performed and endogenous VPS39, STX17, RAB2A were detected by western blot. (L) a working model indicating TBC1D4 inhibits the recruitment of HOPS complex to autophagosome.

TBC1D4 is an inhibitor of the endocytic pathway

Given the essential role of RAB2A in endocytic degradation [38,39,46], we anticipated a negative role for TBC1D4 in endocytic pathway. We monitored the endocytic degradation of EGFR (epidermal growth factor receptor) in TBC1D4-depleted or overexpressed cells. EGF (epidermal growth factor) was used to stimulate EGFR activation and subsequent degradation. In TBC1D4 overexpression cells, EGFR degradation was delayed compared with that in WT TBC1D4 cells (Figure 7A,B). In contrast, the basal level of EGFR was significantly lower in TBC1D4 KO cells (Figures 7A–D). The aberrant degradation of TBC1D4 was rescued by simultaneous RAB2A KD (Figure 7C,D). These data indicated that TBC1D4 negatively regulates RAB2A-controlled endocytic degradation in mammalian cells. Indeed, in TBC1D4 overexpression cells, EGFR accumulated in the large vacuoles (Figure 7E,F), which were absent in adjacent control cells without exogenous TBC1D4 expression. Furthermore, RUBCNL overexpression failed to alter the EGFR degradation kinetics (Figure 7G,H). We employed the TBC1D4^R972K mutant to probe whether TBC1D4’s negative regulation of endocytic degradation is solely attributed to its GAP activity. The results unequivocally demonstrated that while GAP activity contributes to TBC1D4’s inhibition of endocytic degradation (Figure 7I,J), it represents only a partial explanation. Notably, TBC1D4 had no discernible impact on the uptake efficiency of dextran (Figure 7K). Intriguingly, it did impede the delivery of dextran to late endosomes or lysosomes (Figure 7L,M), aligning with the established localization of RAB2A in late endosomes and its involvement in fusion with lysosomes. These findings suggest a multifaceted mechanism wherein TBC1D4, through its diverse activities, modulates distinct facets of endocytic processes [38,39,66]. Together, TBC1D4 inhibits RAB2A-mediated endocytic degradation, which is independent of RUBCNL.

Figure 7.

TBC1D4 negatively regulates endocytic degradation. (A) TBC1D4 WT, TBC1D4 KO and TBC1D4 OE U2OS cell lines were treated with EGF as described, and cell lysates were analyzed by western blot by anti-EGFR antibody. Comparing the grayscale ratio of EGFR to ACTB, and normalized using WT (0 h) as the standard. Statistical results in (B). Data are shown as mean ± SD (*p < 0.05, **p < 0.01). (C) WT, TBC1D4 KO, RAB2A KO and TBC1D4 KO RAB2A KD U2OS cell lines were stimulated with EGF as described. Cell lysates were analyzed by western blot. Comparing the grayscale ratio of EGFR to ACTB, and normalized using WT (0 h) as the standard. Statistical results in (D). Data are shown as mean ± SD (****p < 0.0001). (E) U2OS cells were transfected with HA-TBC1D4. Cells were stained with anti-EGFR and HA antibody. EGFR phenotype was monitored by confocal fluorescence microscope, and quantification of number of EGFR-positive vacuoles1.0 μm in diameter was shown in (F). Scale bars: 10 µm. Data are shown as mean ± SD (**p < 0.01). (G) RUBCNL WT and RUBCNL OE HEK293T cell lines were treated with EGF as described. Cell lysates were analyzed by western blot by anti-EGFR antibody. Comparing the grayscale ratio of EGFR to ACTB, and normalized using WT (0 h) as the standard. Statistical results in (H). Data are shown as mean ± SD. (K) Flow cytometry was analyzed the dextran uptake efficiency of TBC1D4 WT and KO cells. The mean fluorescence intensity was quantified and statistically analyzed. Data are shown as mean ± SD. (L) Dextran uptake assay was performed in TBC1D4 WT and TBC1D4 OE U2OS cell lines. Cells were monitored by fluorescence microscope and the colocalization percentage of LAMP1 and dextran was quantified in (M). Scale bars: 10 µm. Data are shown as mean ± SD (**p < 0.01).

TBC1D4 is a molecular brake of both endocytic and autophagic pathways in vivo and is required for liver homeostasis

To investigate TBC1D4’s physiological function in vivo, we generated a liver-specific knockout model for tbc1d4 by crossing mice carrying the floxed- Tbc1d4 allele (Tbc1d4^f/f) with the Alb-Cre mice (Figures 8A and S4A). Backcrossing of tbc1d4^f/+; Alb-Cre mice resulted in homozygous depletion of TBC1D4 in the mouse liver (tbc1d4^f/f; Alb-Cre). Western blot analysis further confirmed the loss of TBC1D4 expression (Figures 8A). Consistent with the observations from the in vitro assays, autophagic flux and endocytic degradation were significantly elevated in the knockout mice (tbc1d4^f/f; Alb-Cre) when compared with the wild-type control (Tbc1d4^f/f) (Figure 8A–D). The result demonstrated that TBC1D4 is a negative regulator of both autophagic and endocytic pathways in vivo. Notably, TBC1D4 ablation resulted in liver damage, which was indicated by the increased infiltration of immune cells and aberrant upregulation of serum GOT1/AST (glutamic-oxaloacetic transaminase 1), total bile acid (TBA) and GPT/ALT (glutamic – pyruvic transaminase) levels (Figure 8E–H). However, there was no significant difference serum in ALB (albumin), ALP (alkaline phosphatase), body weight, liver:body weight ratio between the wild-type and mutant mice (Figures S5B, S5C, S5D and S5E). We assessed changes in autophagic flux by comparing the colocalization of LC3B and LAMP1 in the liver tissues of tbc1d4^f/f and tbc1d4^f/f; Alb-Cre mice. The results unequivocally reveal that the knockout of Tbc1d4 significantly enhances autophagic flux (Figure 8I,J). To reinforce these in vivo findings, we performed tbc1d4 knockout in adipocyte tissues, using atg5 conditional KO mice as a positive control to examine alterations in autophagy-related proteins. Remarkably, the absence of TBC1D4 led to increased autophagic flux in adipocyte tissues, all while maintaining body weight and the epididymal fat/body weight ratio unchanged (Figures 8K, S5F, S5G, and S5H). Collectively, these observations underscore the role of TBC1D4 in maintaining liver homeostasis and establish it as a key regulator of autophagic and endocytic pathways in vivo.

Figure 8.

TBC1D4 is a regulator of autophagic and endocytic pathways in vivo. (A) the protein levels of TBC1D4, EGFR, RUBCNL, SQSTM1, RAB2A, LC3B and TUBA1B in the livers of six pairs of Tbc1d4^f/f and tbc1d4^f/f; Alb-Cre mice for 2 months old were determined by western blot. [(B) and (C)] Immunofluorescence showing SQSTM1 (B) or LC3B (C) protein levels in liver tissues of 2 mouths Tbc1d4^f/f and tbc1d4^f/f; Alb-Cre mice. Scale bars: 50 µm. 600×, n = 6. Puncta were quantified and analyzed in (D). Data are shown as mean ± SD, **p < 0.01. (E) H&E staining showing liver histopathologic changes in liver tissues of 2 mouths Tbc1d4^f/f and tbc1d4^f/f; Alb-Cre mice. Scale bars: 100 µm. 200×, n = 6. Irregular structures were seen in the cytoplasm of hepatocytes (red arrow); Hepatocyte necrosis with scattered inflammatory cell infiltration (black arrow). [(F) to (H)] Serum AST (F), TBA (G), ALT (H); n = 6. Data are shown as mean ± SD, **p < 0.01, ***p < 0.001, Student’s unpaired t-test. (I) Immunofluorescence showing colocalization of LC3B and LAMP1 in the liver tissues of Tbc1d4^f/f and tbc1d4^f/f; Alb-Cre mice in liver tissues of 4 mouths Tbc1d4^f/f and tbc1d4^f/f; Alb-Cre mice. Scale bars: 50 µm. 600×, n = 3. Puncta were quantified and analyzed in (J). Data are shown as mean ± SD (**p < 0.01). (K)The protein levels of TBC1D4, EGFR, RUBCNL, SQSTM1,ATG5, RAB2A, EEA1 and GAPDH in white adipose tissue (WAT) of Tbc1d4^f/f, atg5^f/f; Adipoq-Cre, and tbc1d4^f/f; Adipoq-Cre mice for 3–4 months old were determined by western blot.

Discussion

In this study, we demonstrated that TBC1D4 depends on RAB2A to colocalize with the early autophagy markers. This observation further solidified our previous findings that RAB2A exists on different autophagic membrane structures [25], thereby regulating early and late stages of the autophagy process. Surprisingly, RAB2A GDP failed to form diffused patterns in the presence of TBC1D4 overexpression. This phenotype suggested that TBC1D4 May prevent the GDI-mediated membrane extraction of the RAB2A GDP mutant to enable the latter to stay on membranes. Although RUBCNL has a similar function in this regard [25], they may have opposite roles in recruiting the guanine nucleotide exchange factor for RAB2A.

RAB2A participates in various vesicular trafficking processes [25,38–40,45,46,67–78], but it is unclear how these pathways are coordinately regulated by RAB2A, and it is not known whether TBC1D4 or RUBCNL also has a role in these processes. Although TBC1D4 and RAB2A form a bimolecular switch in endocytic degradation, it is apparent that RUBCNL is not involved in this regulation. How the endocytic complex containing RAB2A and TBC1D4 excludes RUBCNL remains to be studied. It is likely that certain uncharacterized endosome-resident proteins associate with RAB2A on endosomes to ensure the specific recruitment of HOPS complex to facilitate endocytic degradation. If this is the case, this regulatory process will be similar to the model in which RUBCNL, RAB2A and STX17 forms an autophagosomal complex to control specific recruitment of HOPS complex for autophagic degradation.

TBC1D4 plays an essential role in glucose homeostasis by controlling SLC2A4 trafficking. In TBC1D4-depleted mouse adipocytes, SLC2A4 is subjected to lysosomal degradation [79]. Whether RUBCNL and RAB2A are involved in this process requires further investigation. Interestingly, a previous study has indicated that autophagic sequestration of TBC1D5 coordinates glucose uptake by promoting SLC2A1/GLUT1 translocation to the plasma membrane [80]. In autophagy-deficient cells, SLC2A1 was mis-sorted to late endosomes. Autophagy may therefore regulate glucose uptake differently in different tissues or cell types.

Similar to other members of the TBC protein family, TBC1D4 exhibits the ability to regulate the activity of multiple RABs [50]. For instance, TBC1D4 plays a role in SLC2A4 vesicle transport by modulating the activity of RAB8, RAB10, and RAB14 [51,52,54]. Moreover, TBC1D4 is involved in the control of lipid droplet fusion and growth through its interaction with RAB8 [81], and collaborates with RAB19 to coordinate cortical clearing and ciliary membrane growth [82]. Additionally, TBC1D4 regulates glucose-independent eukaryotic cell proliferation by controlling the cyclin-dependent kinase inhibitor p21 [83]. Consequently, the observed phenotype in tbc1d4 knockout mice may be influenced by other factors in addition to RAB2A.

Together with our previous findings, we propose here a model in which TBC1D4 and RAB2A form a dual-molecular switch in regulation of endocytic and autophagic degradation both in vitro and in vivo. These results not only provide further insights into how different autophagy stages were seamlessly connected and properly controlled, but also open up an opportunity to further understand how the autophagy pathway is tightly connected with endocytic processes.

Materials and methods

Antibodies

Anti-ACTB (HuaBio, M1210–2), anti-ATG5 (Abcam, Ab108327), anti-ATG5 (Proteintech 10,181–2-4P), anti-ATG9 (Cell Signaling Technology, 13509S), anti-ATG14 (MBL, PD026), anti-ATG16L1 (MBL, PM040), anti-BECN1(MBL, PD017), anti-EEA1 (Cell Signaling Technology, C45B10), anti-EGFR (Santa Cruz Biotechnology, sc-120), anti-GAPDH (Proteintech 60,004–1-IG), anti-GFP (Santa Cruz Biotechnology, M048–3), anti-LAMP1 (Santa Cruz Biotechnology, sc -20,011), anti-LAMP1 (Cell Signaling Technology, D2D11), anti-LC3B (Sigma, L8918), anti-LC3B (MBL, PM036), anti-LC3B (MBL, M152–3), anti-p-ATG14 (S29; Cell Signaling Technology, 13155S), anti-p-ULK1 (S555; Cell Signaling Technology, 5869S), anti-RAB2A (BBI Life Sciences, D122959–0200), anti-RAB2A (ABGENT, AP52713), anti-RAB2A (Abcam, GR188995–4), anti-RUBCNL/PACER (Homemade), anti-SQSTM1/p62 (MBL, PM045), anti-STX17 (Sigma, HPA001204), anti-TBC1D4 (Sigma, 07–741), anti-TBC1D4 (Millipore, 07–741), anti-TUBA1B (Proteintech 11,224–1-AP), anti-ULK1 (Santa Cruz Biotechnology, sc -390,904), anti-UVRAG (Cell Signaling Technology, D2Q1Z), anti-VPS39 (Santa Cruz Biotechnology,sc -514,762), anti-WIPI2 (Proteintech 28,820–1-AP), anti-HA-Tag-HRP (MBL, M180–7), anti-FLAG-Tag-HRP (MBL, M185–7), anti-HA (Biolegend, 16B12), anti-FLAG (Sigma, F1804), Alexa Fluor 488 (Abcam, GR238847–1), Alexa Fluor 546 (Thermo Fisher Scientific, A11003), Alexa Fluor 546 (Thermo Fisher Scientific, A11010), Alexa Fluor 405 (Thermo Fisher Scientific, A31556), Alexa Fluor 405 (Thermo Fisher Scientific, A81553), Alexa Fluor 488 (Thermo Fisher Scientific, A11008), Alexa Fluor 488 (Thermo Fisher Scientific, A11001), Alexa Fluor 633 (Thermo Fisher Scientific, A21071), Alexa Fluor 633 (Thermo Fisher Scientific, A21050).

Chemicals and reagents

Bafilomycin A₁ (Selleck Chemicals, S1413), bovine serum albumin (BSA; [Sangon Biotech, A500023]), DTT (Sangon Biotech, A100281), doxycycline (Sangon Biotech, A600889), EDTA (Sangon Biotech, A100322), EGF (Peprotech,100–15), ethanol (Sinopharm Chemical Reagent 10,009,218), GST agarose (Probegene, PC014), HEPES (Sangon Biotech, A100511), Lipofectamine 3000 (Thermo Fisher Scientific, L3000015), MG-132 (Selleck, S2619), NaF (Sangon Biotech, A500850), NaCl (Sangon, A610476), NP-40 (Sangon Biotech, A100109), paraformaldehyde (Sangon Biotech, A500684), protease inhibitor cocktail (Bimake, B15001), phosphatase inhibitor cocktail (Bimake, B14001), restriction enzymes (Thermo Fisher Scientific), torin1 (Selleck Chemicals, S2827), Triton X-100 (Sangon Biotech, A110694), tris (hydroxymethyl) aminomethane (Tris; [Sangon Biotech,A501492]), 2× Taq Master Mix (Probegene, ME013), 2× Ultra-Pfu Master Mix (Probegene, ME026), 2× Phanta Master Mix (Vazyme Biotech, P511).

Cell lines

U2OS (ATCC, HTB-96), HEK293T (ATCC, CRL-3216), FLAG-TBC1D4 U2OS, EGFP-TBC1D4 U2OS, TBC1D4 KO U2OS, TBC1D4 KD HEK293T, RAB2A KO U2OS and TBC1D4 KO RAB2A KD U2OS (constructed in our lab).

Recombinant DNA

pEGFP-C1 (Clonetech, PT32595), pmCherry-N1 (Clonetech, PT3974–5), pCDNA5/FRT/TO-3× FLAG (Invitrogen, V6010–20), pcDNA3.1-HA (Invitrogen, V709–20), pmCherry-EGFP (Addgene 86,639; deposited by Thomas Leonard and Ivan Yudushkin), pOG44 (Stratagene, 1141), pLKO.1-puro sgRNA (Addgene 50,927; deposited by Rene Maehr and Scot Wolfe), pLVX-AcGFP1-N1-puro (Takara bio,632154) HP138 and HP216 were provided by Hongguang Xia Laboratories.

Software and Algorithms

LSM 800 Browser (ZEISS), SnapGene (Dotmatics), GraphPad Prism (Dotmatics), Canvas X Draw (Canvas GFX), Fiji (ImageJ), FlowJo (BD), IBS (CUCKOO).

Cell culture

U2OS and HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco 11,995,065) supplemented with 10% fetal bovine serum (Gibco, 10100147C), 2 mM L-glutamine, 1% penicillin-streptomycin at 37°C in 5% CO₂.

Stable cell lines construction

FLAG-TBC1D4/U2OS were obtained by Flp-In™ System (Invitrogen, V6005–20). The pOG44 plasmid and the pcDNA5-FRT-TBC1D4 vector were co-transfected into the Flp-In™ HEK293T and Flp-In™ U2OS cells. Cells were divided into two to four 10 cm tissue culture dishes 48 h after transfection. Cells were selected with 100 μg/ml hygromycin (Selleck, S2908) one day after re-plating. About 15 days later, single cell colonies were tripsinized and seeded in 96-well plate. When getting high confluence, cells were transferred to a 12-well tissue culture plate. Once the 12-well tissue culture plate clones have expanded to high confluence, they can be passaged to a 6-well tissue culture plate. A small portion of the cells was assessed for expression of FLAG-TBC1D4 at this time point.

GFP-TBC1D4/U2OS were obtained by TET ON System. The HP216 plasmid and the HP138-GFP-TBC1D4 vector were co-transfected into the U2OS cells. Cells were divided into two to four 10 cm tissue culture dishes 48 h after transfection. GFP-TBC1D4 were induced with 1 μg/ml doxycycline one day after re-plating. Use flow cytometry (Beckman moflo Astrios EO) to sort cells with green fluorescence, and then culture these cells.

Generation knockout cell lines for TBC1D4

pLKO-cas9-TBC1D4 sgRNA vector with designed gDNA sequence 5’ CGACGACCCCGAGTCGCAGA 3’ was cloned. Cells were seeded in a 6-well tissue culture plate with 50% confluency one day before transfection. Cells were transfected with 2 μg pLKO-cas9-TBC1D4 sgRNA vector by Lipofectamine 3000. Twenty-four h later, cells were selected with 1 μg/ml puromycin. After 2 days incubation, cells were diluted and seeded into 15 cm tissue culture dishes. Two weeks later, single cell clones were trypsinized and seeded into 96-well tissue culture plates. A small portion of amplified cells were performed western blot of target protein to screen gene knockout of each clone. Western blot verified knockout cell clones were sent for sequencing verification. The forward sequencing primers: TGCATTCAGGATGAGCCGTT, reverse primer: AGCCAAACCTCAGTCGGTC

ShRNA knockdown

pLVX-TBC1D4-Puro or PLKO-RAB2A-Puro vector with designed shRNA sequence 5’ GAGGTCTTAATAACTTGGGAT3’ or 5’GCTCGAATGATAACTATTGA3’ was cloned. Cell lines stably expressing TBC1D4 or RAB2A shRNA and control shRNA were obtained by lentivirus infection and were selected with 1 µg/ml of puromycin. Recombinant lentiviruses were produced following the lentiviral packaging protocol.

Immunoprecipitation and western blot

Cell pellets were homogenized in TAP buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% NP-40, 1 mM NaF, 1 mM EDTA, 10 nM MG-132, protease inhibitor cocktail, phosphatase inhibitor cocktail) and incubated on ice for 30 min. The cell lysate was cleared by centrifugation at 18,000 g for 30 min at 4°C. The supernatant was incubated with antibody-conjugated beads and rotated for 2 h at 4°C. After incubation, the beads were washed 3 times with TAP buffer. Western blot was performed following standard procedures.

Immunofluorescence

Cells grown on coverslips were transfected with different plasmids, then fixed in 4% paraformaldehyde in PBS (BIOVISTECH, VM-PBS) for 20 min at room temperature and permeabilized with 0.1% Triton X-100 in PBS for 20 min. Following permeabilization, cells were treated with block buffer (5% goat serum [Solarbio Life Sciences, S9070], 0.1% Triton X-100 in PBS) for 1 h at room temperature. Cells were incubated with primary antibodies diluted in block buffer overnight at 4°C. Cells were washed three times with PBS, each for 10 min, followed by incubation with Alexa Fluor-conjugated secondary antibody (Life Technologies) in block buffer for 1 h at room temperature. Slides were examined by using a laser scanning confocal microscope (Zeiss LSM 800).

Autophagy analysis

For LC3B-II degradation assay, TBC1D4 KO U2OS, TBC1D4 KD HEK293T or WT were treated with 250 nM torin1 or DMSO at different time points, and whole cell lysates were briefly sonicated in 1× SDS loading buffer, and incubated at 100°C for 15 min, then subjected to western blot analyses with antibodies against SQSTM1. For autophagosome maturation assays, U2OS cells with TBC1D4 OE or WT were transfected with mCherry-GFP-LC3B, 24 h post-transfection, the cells were treated with torin1 or DMSO at 37°C for 2 h, and analyzed by fluorescence microscopy. U2OS cells with TBC1D4 OE or WT treated with torin1 or DMSO were analyzed by immunostaining with LAMP1 and LC3B antibodies.

Endocytic trafficking of the EGFR

Cells cultured in 6-well tissue culture plates were grown to approximately 80% confluency. Washed with DMEM and serum-starved (culture the cells with DMEM only) overnight. EGFR endocytosis was stimulated by addition of EGF in DMEM containing 20 mM HEPES and 0.2% BSA. Cells were lysed at different time points in 1× SDS loading buffer. Western blot was performed and EGFR expression level was detected using antibody of EGFR.

Dextran assay

For endocytosis uptake assaysTBC1D4 WT and KO cell lines are incubated with dextran (10,000 MW; Thermo Scientific, P10361; 20 μg/ml) for 30 min. After washing cells with PBS, analyzing the average fluorescence intensity (PE) of cells with flow cytometry.

For dextran colocalization experiment, cells grown on coverslips were transfected with different plasmids, were labeled with dextran (10,000 MW; Thermo Scientific, D22914; 1:1000 dilution) for 2 h in cell incubator. After washing ten times with cold PBS (keep the plate on ice during washing). Then immunofluorescence was performed as described above.

Recombinant protein purification

RAB2A (WT, Q65L or N119I) or RUBCNL (∆301–400) was cloned into pGEX-6P1 (Addgene 119,749; deposited by Andrew Jackson and Martin Reijns) and expressed as glutathione-S-transferase (GST) fusion proteins with a TEV protease cleavage site in between. PTB1, PTB2, and (PTB1+PTB2) domain of TBC1D4 was cloned into pGEX-6P1 without the TEV protease cleavage site.

GST fusion proteins were expressed in Escherichia coli BL21 (DE3) at 16°C to achieve maximal soluble expression. Cells were collected by centrifugation and washed three times with cold PBS. The cells were lysed by sonication in lysis buffer (20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 1 mM EDTA, 0.5% Triton X-100, protease inhibitor cocktail) and centrifuged at 12,000 g for 15 min. 0.2 ml GST-Sepharose resin (ProbeGene Life Sciences, J11C) pre-equilibrated with 20 ml TEV protease cleavage buffer (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% NP-40, 1 mM DTT) was added to the supernatant and rotated at 4°C for 2 h. Next, beads were washed three times with TEV protease cleavage buffer, and then, the recombinant protein was eluted from the resin by incubating at 4°C overnight with 10 μg/ml of TEV protease (Homemade) to cleave off the desired protein from the GST tag, which was still bound to the GST-Sepharose resin after the overnight cleavage reaction. Purified untagged recombinant proteins were further fractionated using a Mono-Q column (Amicon, UFC900396, UFC903096), followed by dialysis against PBS. Proteins were quantified by the Bradford method and analyzed by SDS-PAGE and Coomassie Brilliant Blue staining.

Clem

Coat the glass gridded coverslips (Cellvis, D35-14-1.5GI) with 20 ng/ml polylysine (Biosharp Life Sciences, BS198) for 30 min, then rinse it several times with ddH2O. After drying the dish under UV irradiation, inoculate U2OS cells. Cells were transfected with EGFP-TBC1D4 plasmids. Cells were fixed with 4% paraformaldehyde for 15 min after transfected for 24 h and imaged by Zeiss Airyscan to collect light microscopy images (20×) and Z-stacks images (63×). Then cells were fixed with 2.5% glutaraldehyde (TED PELLA 2,171,002) at 4°C overnight. The next day, cells were postfixed in 2% osmium tetroxide (TED PELLA, 4008–160501)–3% potassium ferrocyanide (Sigma 1,445,995–1) in PBS for 1 h followed by 1% thiocarbohydrazide (Sigma, 2231-57-4) dissolved in water for 20 min and incubated in 2% osmium in PBS for 30 min. Samples were then dehydrated with a graded ethanol series (20, 50, 70, 90, and 100%) each for 15 min and processed for EPON (EMS 14,900) embedding. The samples were cut (30 KV and 2.5 nA) and imaged (2 KV and 0.2 nA) by focused ion beam – SEM (Helios UC G3).

Size-exclusion chromatography (SEC)

SEC with Superose 6 10/300 GL was performed at 4°C using an AKTA PURE according to the Handbooks from GE Healthcare Life Sciences (https://www.gehealthcare.com/products/lifesciences). Equilibrating the column with two column volumes of buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, filtration, and ultrasonic defoaming). After collecting cells, resuspend cells in 2 ml lysis buffer (20 mM HEPES, pH 7.2, 100 mM NaCl, 2 mM EDTA, 1 mM DTT, 10 nM MG-132, protease inhibitor cocktail, phosphatase inhibitor cocktail) and grind cells for 6–10 times on ice. Centrifuge at 21,000 g for 15 min. The supernatant was injected into the column. Samples were collected and subjected to SDS-PAGE analyses. The complex size was deduced from the chromatograms of six standard proteins on Superose 6 10/300 GL (Cytiva 17,517,201).

tbc1d4 knockout mouse models

All animal experiments and breeding protocols were approved by the Zhejiang University Ethical Review Committee and conducted under the Office project license (Protocol No. IACUC-ZJU20220256). Mice were maintained in a barrier facility, at normal room temperatures, on a regular 12 h light and 12 h dark cycle. The Tbc1d4^f/f mice were provided by Shuai Chen Laboratories and the atg5^f/f mice were provided by Zhijian Cai Laboratories. We cross Tbc1d4^f/f mice to Alb-Cre or Adipoq-Cre mice (Gempharmatech Co., Ltd, T003814, T052679), and cross atg5^f/f mice to Adipoq-Cre mice. All mice were backcrossed to C57BL/6J background for at least five generations before experiments. For PCR genotyping, genomic DNA was isolated from mouse tails and amplified by standard PCR. For the TBC1D4 conditional knockout mice, forward primer: GAGATTGCTGAGGTGACAAGA, reverse primer: CTGCTGAGCTACCCCATCAT were used to detect both 400 bp and 262 bp product for Tbc1d4 flox and Tbc1d4⁺ allele, respectively. For the atg5 conditional knockout mice, primer 1: GAATATGAAGGCACACCCCTGAAATG, primer 2: ACAACGTCGAGCACAGCTGCGCAAGG and primer3: GTACTGCATAATGGTTTAACTCTTGC were used to detect both 700 bp and 350 bp product for Atg5 flox and Atg5⁺ allele, respectively. Alb-Cre transgene was amplified by forward primer: TGGATGCCACCTCTGATGAAGTC and reverse primer: TCCTGGCATCTGTCAGAGTTCTCC to obtain a specific 424 bp product. Adipoq-Cre transgene was amplified by forward primer: GCCTGCATTACCGGTCGATGC and reverse primer: CAGGGTGTTATAAGCAATCCC to obtain a specific 450 bp product.

Protein extraction from tissue

Liver or epididymal fat pads samples (250 mg) were homogenized in 1 mL or 0.5 mL TAP buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% NP-40, 1 mM NaF, 1 mM EDTA, 10 nM MG-132, supplemented with protease, and phosphatase inhibitors) using a homogenizer at 4°C for 240 s, and left on ice for 30 min. The homogenates were cleared by centrifugation at 21,000 g for 30 min. The supernatants were used for western blot analyses.

Immunofluorescence and hematoxylin and eosin staining

Four % paraformaldehyde fixed tissue embedded in paraffin. For immunofluorescence, sections were deparaffinized in two changes of xylene (Sangon Biotech, A530011) and rehydrated through graded concentrations of ethanol sections, then heated in citrate solution (Sangon Biotech, E673002) to retrieve antigenic activity. After nonspecific reactions had been blocked with 10% goat serum, the sections were incubated with using antibodies specific to SQSTM1 or LC3B overnight at 4°C. Primary antibodies were detected using fluorescent conjugated secondary antibodies. Hematoxylin (HE) staining (Servicebio, GP1031) was performed to analyze histopathological changes of liver according to standard protocols.

Assessment of liver damage

Liver damage was evaluated by standard H&E staining in paraffin-embedded liver sections. Sections were evaluated for inflammatory cells infiltration and tissue vacuolation. Serum levels of GPT/ALT, GOT1/AST, TBA, ALB, ALP were measured using the corresponding kit (Rayto, S03030, S03040, S03074, S03043, S03038) according to the manufacturer’s instruction.

Quantification and statistical analysis

Perform normality test on all data first. Then, t test for 2 normal distribution datasets, u test for 2 where at least one is non-normal, ANOVA for comparing 3 or more normally distributed datasets, and Kruskal-Wallis for 3 or more where at least one is non-Gaussian. Values are expressed as mean ± SD of at least three independent experiments, unless otherwise noted. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001 were considered statistically significant.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Funding Statement

The work was supported by the National Natural Science Foundation of China [32025012, 92254307, 31970695, 31771525, 31900530]; Ministry of Science and Technology of the People’s Republic of China [2021YFC2700901].

Disclosure statement

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

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2024.2367907