GORASP2 promotes phagophore closure and autophagosome maturation into autolysosomes

ABSTRACT

As the central hub of the secretory pathway, the Golgi apparatus plays a crucial role in maintaining cellular homeostasis in response to stresses. Recent studies have revealed the involvement of the Golgi tether, GORASP2, in facilitating autophagosome-lysosome fusion by connecting LC3-II and LAMP2 during nutrient starvation. However, the precise mechanism remains elusive. In this study, utilizing super-resolution microscopy, we observed GORASP2 localization on the surface of autophagosomes during glucose starvation. Depletion of GORASP2 hindered phagophore closure by regulating the association between VPS4A and the ESCRT-III component, CHMP2A. Furthermore, we found that GORASP2 controls RAB7A activity by modulating its GEF complex, MON1A-CCZ1, thereby impacting RAB7A’s interaction with the HOPS complex. The assembly of both STX17-SNAP29-VAMP8 and YKT6-SNAP29-STX7 SNARE complexes was also attenuated without GORASP2. These findings suggest that GORASP2 helps seal autophagosomes and activate the RAB7A-HOPS-SNAREs membrane fusion machinery for autophagosome maturation, highlighting its membrane tethering function in response to stresses.Abbreviations: BafA1: bafilomycin A₁; ESCRT: endosomal sorting complex required for transport; FPP: fluorescence protease protection; GEF: guanine nucleotide exchange factor; GFP: green fluorescent protein; GORASP2: golgi reassembly stacking protein 2; GSB: glucose starvation along with bafilomycin A₁; HOPS: homotypic fusion and protein sorting; LAMP2: lysosomal associated membrane protein 2; MAP1LC3B: microtubule associated protein 1 light chain 3 beta; PBS: phosphate-buffered saline; PtdIns3K: phosphatidylinositol 3-kinase; PtdIns3P: phosphatidylinositol-3-phosphate; PK: proteinase K; SNARE: soluble N-ethylmaleimide-sensitive factor attachment protein receptor; SIM: structured illumination microscopy; UVRAG: UV radiation resistance associated.

Introduction

Macroautophagy, referred to as autophagy herein, is a fundamental intracellular degradation process that responds to diverse stresses such as nutrient starvation to maintain cellular and tissue homeostasis [1–3]. Autophagy involves several crucial steps, including the initiation and nucleation of a crescent-shaped phagophore, the expansion and closure of the phagophore into autophagosomes, and ultimately, the delivery and fusion of autophagosomes with lysosomes [2,4]. The completion of each step is pivotal for the overall success of autophagy.

The process of phagophore closure remains incompletely understood, with several ATG (autophagy related) proteins, including ATG2, ATG3, ATG4, ATG5, and GABARAP, implicated in this process [5]. Recent studies have revealed a direct involvement of the endosomal sorting complex required for transport (ESCRT) complex and its accessory proteins in phagophore closure [6–9]. With a distinct role in remodeling membranes, the ESCRT complex serves as a conserved membrane scissor in various intracellular membrane processes, including multivesicular endosomes (MVEs) formation, cytokinetic abscission, and virus budding. The ESCRT machinery consists of four multimeric protein complexes, ESCRT-0, ESCRT-I, ESCRT-II, and ESCRT-III, among which ESCRT-III functions as the major membrane-remodeling sub-complex [10].

Components of ESCRT-III, such as CHMP2A and CHMP4B, are recruited to the unsealed autophagosomes during autophagy [8,9]. The ESCRT accessory protein, AAA-ATPase VPS4, aids in the disassembly and recovery of ESCRT-III polymers, facilitating membrane remodeling and scission [11]. VPS4 contributes to autophagy by cooperating with the ESCRT-III component CHMP2A at the phagophore closure site for membrane fission [8]. ESCRT-I components VPS37A and VPS28 have also been found to target the phagophore and facilitate the recruitment of CHMP2A, CHMP4B, and VPS4 [7,8]. Despite these findings, our understanding of how cells recruit and regulate the ESCRT complex on the unsealed autophagosome site is still limited.

The final stage of autophagy involves the fusion of the autophagosome and lysosome to degrade their contents. This step is influenced by several factors, including multiple phosphoinositides, RAB7A, MAP1LC3B/LC3 (microtubule associated protein 1 light chain 3 beta), GABARAP, the homotypic fusion and protein sorting (HOPS) complex, EPG5, RILP (Rab interacting lysosomal protein), and soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complexes [12–14]. The small GTPase RAB7A plays a crucial role in this step by cycling between GDP- or GTP-bound states. Its guanine nucleotide exchange factor (GEF), the MON1A-CCZ1 complex, recruits it to autophagosome and lysosome membranes and stimulates GDP dissociation to allow its replacement by GTP, activating it [15,16]. Phosphatidylinositol-3-phosphate (PtdIns3P) is also required for the recruitment of RAB7A. Possible mechanisms are that PtdIns3P promotes RAB7A GEF association with membranes, enhancing its activity and resulting in efficient RAB7A recruitment and activity [17,18]. A recent study showed that the class III phosphatidylinositol 3-kinase (PtdIns3K) core protein PIK3C3 interacts with MON1A-CCZ1, crucial for maintaining GEF and RAB7A activity [19].

In its active GTP-bound form, RAB7A recruits various effector proteins, including the HOPS complex, to the surfaces of autophagosomes and lysosomes, mediating the final step of membrane fusion [20,21]. The HOPS complex comprises six subunits, and its unique subunits, VPS39 and VPS41, interact with RAB7A at opposite ends of the autophagosome and lysosome [22]. Two SNARE complexes, STX17-SNAP29-VAMP7/VAMP8 and STX7-SNAP29-YKT6 [23], mediate the final fusion step of autophagosome and lysosome. Despite the identification of numerous molecules regulating the autophagosome maturation into autolysosome, the exact mechanism of their cooperation remains largely unclear.

GORASP2, a prominent member of the Golgi reassembly stacking proteins localizing to the medial- and trans-cisternae of the Golgi, collaborates with the cis-Golgi located GORASP1 (golgi reassembly stacking protein 1) to physically tether Golgi membranes into stacks and ribbon [24–26]. GORASP2 contains a highly conserved N-terminal GRASP domain and a flexible C-terminal serine/proline-rich (SPR) domain. The N-terminal GRASP domain forms trans-oligomers that hold adjacent membranes together, while the C-terminal SPR domain regulates the tethering process [27].

Our recent research has expanded the understanding of GORASP2’s membrane tethering role to the autophagosome-lysosome fusion under nutrient starvation by interacting with LC3-II, LAMP2 (lysosomal associated membrane protein 2), and the UVRAG (UV radiation resistance associated) complex [28–30], suggesting that the Golgi can sense stresses and contribute to cellular homeostasis. However, the exact mechanism of how GORASP2 contributes to the autophagic flux remains unclear.

In this study, we explored the details of GORASP2’s role in glucose starvation-induced autophagy and revealed its multifunctional involvement in autophagosome maturation. GORASP2 not only aids in the association of VPS4 and ESCRT-III components to promote phagophore closure but also regulates the activity of RAB7A, influences the assembly of SNAREs, and thereby facilitates autophagosome-lysosome fusion. These new findings underscore the crucial role of GORASP2 as a tether that responds to cellular stresses.

Results

GORASP2 localizes on the surface of autophagosomes and contributes to phagophore closure

In recent studies, we discovered that glucose starvation induced de-O-GlcNAcylation of the Golgi stacking protein GORASP2, resulting in GORASP2 localization on the autophagosomes, where it promotes autophagosome-lysosome fusion by tethering LC3 and LAMP2 [28]. To investigate in detail how the Golgi tether GORASP2 facilitates autophagosome maturation, three-color structured illumination microscopy (SIM) was used to detect the relative position of GORASP2-GFP, autophagosome marker LC3, and lysosome marker LAMP2 in glucose-deprived U-2 OS cells. 3D-SIM reconstruction images showed that GORASP2 overlapped with LC3 and approached LAMP2 (Figure 1A).

Figure 1.

GORASP2 depletion leads to accumulation of unclosed autophagosomes/phagophores. (A) 3D-SIM images showed that GORASP2-GFP colocalized with LC3 and approached LAMP2. U-2 OS cells were transfected with GORASP2-GFP (GORASP2-GFP hereafter), treated with glucose starvation medium with BafA1 for 4 h, and stained with LC3 and LAMP2 antibodies. Overlapping GORASP2 and LC3 signals were indicated by arrows, while adjacent GORASP2 and LAMP2 were indicated by arrowheads. Scale bar: 5 μm (left), 1 μm (right zoomed-in panel). (B) GORASP2 was localized on the surface of the phagophores. HeLa cells were transfected with GORASP2-MYC, treated with glucose starvation medium with BafA1 for 4 h, and stained with MYC and LC3 antibodies. Scale bar is indicated in the figure. (C) A line profile that acrossed an autophagosome and GORASP2-MYC on its surface in the STED data (yellow solid line in B). (D) GORASP2 depletion caused defects in phagophore closure. HeLa WT and GORASP2 KO cells expressing mCherry-LC3 were treated with glucose starvation medium with BafA1 (GSB) for 4 h, followed by digitonin and proteinase K (PK) treatment. Time-lapse images displayed the remaining mCherry-LC3 puncta that were resistant to PK treatment (FPP assay). Scale bar: 20 μm. (E) Quantification of D for the percentage of LC3 puncta that remained over time in the FPP assay. Values were shown as means ± SEM (n = 15 cells in each group). (F) GORASP2 depletion resulted in an increase in unclosed phagophores. HeLa WT or GORASP2 knockdown (GORASP2 KD) cells expressing HaloTag-LC3 were treated with GSB for 4 h, followed by labeling them with a membrane-impermeable Halo Tag ligand, MIL, and then a membrane-permeable Halo ligand, MPL. Scale bar: 10 μm, 1 μm (inserted panel). (G) Quantification of the ratio of the indicated LC3 puncta to total LC3 in F. The HaloTag-LC3 puncta were categorized into three groups: MIL⁺ MPL⁻ (unclosed phagophores), MIL⁻ MPL⁺ (mature autophagosomes/amphisomes/autolysosomes), and MIL⁺ MPL⁺ (nascent autophagosomes). Values were shown as means ± SEM (n = 60 cells in each group). Significance was determined using unpaired two-tailed Student’s t test. **p < 0.01.

For detailed investigation of GORASP2 on autophagosomes, we utilized stimulated emission depletion (STED) microscopy, which allows for a resolution of approximately 40 nm. Through this method, the refined position between GORASP2 and LC3 was observed, revealing that GORASP2 was present in multiple spots around the surface of autophagosomes (Figure 1B). Further analysis using line profiles also indicated that GORASP2 was located on the surface of LC3-labeled autophagosomes (Figure 1C). These findings were consistent with previous findings that GORASP2 plays a role in linking autophagosomes and lysosomes. Moreover, these observations prompted us to explore the specific role of GORASP2 on the surface of autophagosomes.

The final step of autophagosome formation is known as phagophore closure, which involves the separation of the inner and outer phagophore membranes to form double-membrane autophagosomes [31]. To study the role of GORASP2 in this process, we carried out a fluorescence protease protection (FPP) assay [32]. In brief, wild type (WT) or GORASP2 knockout (KO) HeLa cells expressing mCherry-LC3 were subjected to glucose starvation along with bafilomycin A₁ (BafA1) treatment (GSB). Cells were first permeabilized with digitonin and then treated with proteinase K (PK), allowing PK to hydrolyze LC3 located on the outer autophagosome membrane or the unclosed autophagosomes, leading to fluorescence quenching. We found that the fluorescence of mCherry-LC3 puncta was diminished faster and more in GORASP2 KO cells, indicating that GORASP2 depletion may contribute to the defect in phagophore closure (Figure 1D,E).

To further investigate whether GORASP2 functions in phagophore closure, we conducted a HaloTag-LC3 assay to distinguish between closed or unclosed autophagosomes. HeLa WT and GORASP2-depleted cells expressing HaloTag-LC3 were treated with digitonin to make the plasma membrane permeable. We then added sufficient membrane-impermeable Halo Tag ligand called MIL, followed by a membrane-permeable Halo ligand called MPL. The autophagosomes were distinguished with MIL only, MIL and MPL, or MPL only, depending on whether they were unclosed, nascent, or mature [8]. We found that in cells lacking GORASP2, the ratio of only MIL-labeled LC3 increased significantly after glucose starvation plus BafA1 treatment, indicating an accumulation of unclosed autophagosomes/phagophores (Figure 1F,G). This suggests that GORASP2 was involved in phagophore closure. Additionally, we observed that compared to control WT cells, cells lacking GORASP2 had fewer MPL-only LC3 puncta, supporting previous findings that GORASP2 facilitates autophagosome-lysosome fusion [30].

GORASP2 regulates the interaction between VPS4A and CHMP2A during glucose starvation

Recent research has revealed a crucial role of the ESCRT complex in the process of phagophore closure [7–9]. Among the various ESCRT components, two ESCRT-III proteins, CHMP2A and CHMP4B, move to the phagophore surface to assist in the closure [8,9]. The AAA-ATPase VPS4, comprising VPS4A and VPS4B in mammalian cells, hydrolyzes ATP to depolymerize ESCRT-III assemblies and facilitates membrane fusion [33]. VPS4A has been reported to collaborate with CHMP2A at the phagophore closure site [8]. To investigate the mechanisms of how GORASP2 participates in phagophore closure, a co-immunoprecipitation (co-IP) assay was conducted to determine whether GORASP2 interacts with the components of ESCRT-III and VPS4. The results showed a significant increase in the interaction between GORASP2 and CHMP2A, but not other ESCRT-III proteins, following the treatment with glucose starvation plus BafA1 (GSB) (Figure 2A,B). VPS4A, but not VPS4B and the cofactor VTA1, also showed increased interaction with GORASP2 upon GSB treatment (Figure 2C,D). We further confirmed these specific interactions under different treatments, and the results suggested that GORASP2 interacted significantly more with CHMP2A and VPS4A upon glucose starvation, regardless of whether BafA1 was included in the treatments (Figure S1A-S1D). Furthermore, upon treatment with GSB, immunofluorescence (IF) analysis demonstrated that GORASP2 colocalized well with key ESCRT-III components CHMP2A, CHMP4B, CHMP6, and VPS4A, but not others. Notably, GORASP2 almost did not form puncta outside of the Golgi area under control medium (Figure 2E,F; Figure S1E, S1F). These results suggest that GORASP2 may cooperate with ESCRT-III and VPS4A to close the phagophores during glucose starvation.

Figure 2.

GORASP2 facilitates VPS4A-CHMP2A interaction. (A) GORASP2 interacted with ESCRT-III protein CHMP2A during glucose starvation. GFP or GFP-tagged ESCRT-III proteins as indicated were co-transfected with FLAG-tagged GORASP2 (GORASP2-FLAG) in HeLa cells. Cells were then treated with control medium (Con) or glucose starvation medium with BafA1 (GSB), immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. Whole-cell lysates were blotted as input to demonstrate the expression of ESCRT-III proteins and GORASP2-FLAG. (B) Quantification of the interaction between GORASP2 and ESCRT-III proteins in (A). The interaction efficiency was normalized by calculating the ratio of precipitated FLAG protein to corresponding precipitated GFP protein (set as one under control conditions). n = 3 independent experiments. (C) GORASP2 interacted with VPS4A upon glucose starvation. GFP or GFP-tagged VPS4A, VPS4B and its cofactor VTA1 were co-transfected with GORASP2-FLAG in HeLa cells. The same treatment and co-immunoprecipitation procedure were performed as in (A) to determine the interaction between GORASP2-FLAG and indicated proteins. (D) Quantification of the interaction between GORASP2 and the proteins in (C). n = 4 independent experiments. (E) GORASP2 puncta colocalized with CHMP2A, VPS4A and CHMP4B during glucose starvation. HeLa cells were transfected with GFP or FLAG-tagged proteins as indicated, treated with control medium or glucose starvation medium with BafA1 for 4 h, and stained for GORASP2 antibody (or GORASP2 and FLAG antibodies for CHMP2A-FLAG transfected cells). Scale bar: 20 μm, 2 μm (inserted panel). (F) Quantification of (E) for the colocalization of GORASP2 puncta with the indicated proteins. GFP, n = 25 cells in each group; CHMP2A, n = 30 cells; VPS4A, n = 36 cells; CHMP4B, n = 32 cells. (G) GORASP2 directly bound to VPS4A. GST or GST-tagged CHMP2A, CHMP4B, VPS4A, CHMP3, and CHMP6 recombinant proteins were incubated with His-GORASP2, affinity isolated with glutathione beads, and blotted for His antibody. (H) GORASP2 depletion reduced VPS4A colocalization with LC3 puncta. HeLa WT or GORASP2 knockout (KO) cells were transfected with VPS4A-FLAG, treated with glucose starvation medium with BafA1 for 4 h, and stained for LC3 and FLAG antibodies. Scale bar: 20 μm, 2 μm (inserted panel). (I) Quantification of (H) for the colocalization of VPS4A and LC3. n = 40 cells in each group. (J) GORASP2 depletion decreased the interaction between CHMP2A and VPS4A. GFP (as negative control) or GFP-tagged CHMP2A was co-transfected with FLAG-tagged VPS4A (VPS4A-FLAG) in HeLa WT or GORASP2 KO cells as indicated. Cells were treated with glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. “*” indicated unspecific bands. (K) Quantification of the interaction between CHMP2A and VPS4A in (J). n = 3 independent experiments. Statistical analyses values were shown as means ± SEM. Significance was determined using unpaired two-tailed Student’s t test in this figure. “ns” indicated not significant, *p < 0.05, **p < 0.01, ***p < 0.001.

Although phagophore closure can be regulated by a noncanonical ESCRT pathway, where only a few proteins in ESCRT, such as VPS37A, CHMP2A, CHMP4B, and VPS4, have been identified to function in this process, other ESCRT components still have the potential to play a role in autophagy [7–9]. Therefore, we investigated whether GORASP2 works with them for phagophore closure. We found that compared to the control condition, glucose starvation and BafA1 treatment significantly enhanced GORASP2’s association with HGS, STAM and STAM2 of ESCRT-0 (Figure S2A and S2B). Furthermore, the interaction between GORASP2 and ESCRT-I component MVB12A was intensified following GSB treatment, whereas VPS28 exhibited interaction exclusively under GSB treatment (Figure S2C, SD). Additionally, upon glucose starvation, GORASP2 showed a significant increase in association with ESCRT-II components SNF8 and VPS36 (Figure S2E and S2F). However, despite the significant associations between GORASP2 and multiple ESCRT-0, I, and II components, these proteins did not seem to colocalize with GORASP2 (Figure S2G-S2I).

Combining the above co-IP and IF colocalization analysis, we aimed to understand how GORASP2 interacts with CHMP2A and VPS4A to facilitate phagophore closure. Results from the GST affinity-isolation assay revealed that GORASP2 directly interacted with VPS4A but not with the ESCRT-III components (Figure 2G). Since VPS4A recruitment to phagophores is crucial for its function, we investigated the effect of GORASP2 depletion on the VPS4A localization on phagophores. Our results showed that GORASP2 depletion impaired the colocalization of VPS4A with LC3 (Figure 2H,I). Significantly, the interaction between CHMP2A and VPS4A was reduced in the absence of GORASP2 (Figure 2J,K). Based on these findings, we conclude that GORASP2 facilitates the localization of VPS4A on phagophores, promoting the assembly of VPS4A and ESCRT-III and thus modulating the ESCRT capacity of phagophore closure.

GORASP2 depletion affects the activity and localization of RAB7A on autophagosomes

The small GTPase RAB7A is widely distributed on the autophagosome and late endosome/lysosome compartments, serving as a key regulator of autophagosome-lysosome fusion by recruiting various effectors [34–36]. In a previous study, we observed that GORASP2 puncta overlapped with RAB7A-labeled late endosome/lysosomes [28]. In the current study, further investigations by co-IP revealed an enhanced interaction between GORASP2 and RAB7A during GSB condition (Figure 3A,B). We further confirmed this conclusion under different treatments, the interaction between GORASP2 and RAB7A was slightly increased upon BafA1 treatment, whereas it exhibited a notable enhancement with glucose starvation treatment (Figure S3A and S3B). This prompted us to explore how GORASP2 works with the classical membrane fusion regulator in autophagosome maturation.

Figure 3.

GORASP2 depletion reduces RAB7A activity and autophagosome localization. (A) GORASP2 interacted with RAB7A upon glucose starvation. GFP or GFP-tagged RAB7A was co-transfected with GORASP2-MYC in HeLa cells, cells were then treated with control medium (Con) or glucose starvation medium with BafA1 (GSB), immunoprecipitated with a GFP antibody, and blotted for GFP and MYC. Whole-cell lysates were blotted to show the expression of GFP-RAB7A and GORASP2-MYC. (B) Quantification of the interaction efficiency between GORASP2 and RAB7A in (A). n = 3 independent experiments. (C) GORASP2 interacted more with RAB7A^T22N upon glucose starvation. GFP or GFP-tagged different forms of RAB7A was co-transfected with GORASP2-MYC in HeLa cells, cells were then treated with glucose starvation medium (GS), immunoprecipitated with a GFP antibody, and blotted for GFP and MYC. (D) Quantification of the interaction efficiency between GORASP2 and different forms of RAB7A in (C). n = 3 independent experiments. (E) GORASP2 depletion impaired RAB7A activity. HeLa cells were first transfected with either negative control (NC) or GORASP2 RNAi oligo (GORASP2 KD) for 72 h and then treated with glucose starvation medium for another 6 h. Cell lysates were collected and incubated with MBP-RILP protein. Active RAB7A bound to MBP-RILP, and the indicated protein expression levels were detected. (F) Quantification of the activated RAB7A in (C). n = 4 independent experiments. (G) GORASP2 depletion reduced RAB7A localization on autophagosomes. HeLa WT or GORASP2 knocked-out (GORASP2 KO) cells as indicated were transfected with GFP-RAB7A, treated with glucose starvation medium with BafA1 for 4 h, and stained for LC3 antibody. Scale bar: 10 μm, 1 μm (inserted panel). (H) Quantification of (G) for the colocalization between RAB7A puncta and LC3. n = 50 cells in each group. Statistical analyses values were shown as means ± SEM. Significance was determined using unpaired two-tailed Student’s t test in this figure. *p < 0.05, **p < 0.01, ***p < 0.001.

Using a co-IP assay, we found that GORASP2 exhibited a higher tendency to interact with the dominant inhibitory RAB7A^T22N mutant than with the constitutively active RAB7A^Q67L mutant (Figure 3C,D). This result suggests that GORASP2 may function as an upstream regulator of RAB7A, rather than serving as a downstream effector of RAB7A. To validate this hypothesis, we employed a RILP assay, which is based on the concept that the active, GTP-bound RAB7A efficiently interacts with its lysosomal effector RILP [37]. In essence, we used purified MBP-RILP to isolate intracellular activated RAB7A, and the results showed a decrease in the amount of activated RAB7A in GORASP2 knocked-down cells (Figure 3E,F). As activated RAB7A tends to localize on the membranes, we further investigated whether GORASP2 affected RAB7A localization on autophagosomes. The results demonstrated that GORASP2 depletion significantly reduced the colocalization between RAB7A and LC3 (Figure 3G,H), underscoring the regulatory role of GORASP2 in RAB7A activation and autophagosome maturation, particularly during glucose starvation.

GORASP2 enhances MON1A-CCZ1 activity on RAB7A

In our efforts to understand how GORASP2 influences RAB7A activity, we discovered that GORASP2 did not directly interact with RAB7A, regardless of its form (Figure S3C). However, considering the enhanced interaction of GORASP2 with RAB7A^T22N (Figure 3C,D), we investigated whether GORASP2 modulates the GEF activity of RAB7A. The MON1A-CCZ1 complex is a well-established GEF for RAB7A. Our co-IP results showed that GORASP2 interacted with both CCZ1 and MON1A during glucose starvation (Figure 4A–C; Figure S4A-4D). Furthermore, our findings suggested that MON1A formed fewer puncta and exhibited reduced colocalization with LC3 in GORASP2-depleted cells (Figure 4D,E). This implies that GORASP2 regulates MON1A recruitment to autophagosomes, indicating a potential decrease in the association of MON1A-CCZ1 with RAB7A in the absence of GORASP2. The co-IP assay confirmed that the CCZ1 interaction with RAB7A was reduced in GORASP2 knockdown cells during glucose starvation (Figure 4F,G). These changes were not due to altered expression levels of RAB7A, MON1A, and CCZ1 in GORASP2-depleted cells (Figure S3D). Collectively, these data demonstrate that GORASP2 helps recruit MON1A-CCZ1 and thereby facilitates RAB7A activation on autophagosomes.

Figure 4.

GORASP2 promotes MON1A-CCZ1 activity on RAB7A. (A) GORASP2 interacted with CCZ1 upon glucose starvation. GFP or GFP-tagged CCZ1 was co-transfected with GORASP2-FLAG in HeLa cells. Cells were then treated with control medium (Con) or glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. Whole-cell lysates were blotted to show the expression of GFP-CCZ1 and GORASP2-FLAG. (B) GORASP2 interacted with MON1A upon glucose starvation. GFP or GFP-tagged GORASP2 (GORASP2-GFP) was co-transfected with FLAG-MON1A in HeLa cells. The same treatment and co-immunoprecipitation procedure were performed as in (A) to determine the interaction between GORASP2-GFP and FLAG-MON1A. (C) Quantification of the interaction efficiency between GORASP2 and CCZ1, MON1A in (A and B). n = 3 independent experiments. (D) GORASP2 depletion reduced MON1A localization on the autophagosomes. HeLa WT or GORASP2 knocked-out (GORASP2 KO) cells as indicated were transfected with MON1A-mCherry, treated with glucose starvation medium for 6 h, and stained for LC3 antibody. Scale bar: 10 μm, 1 μm (inserted panel). (E) Quantification of (D) for the colocalization of MON1A puncta with LC3. n = 46 cells in each group). (F) GORASP2 depletion decreased the interaction between RAB7A and CCZ1. HeLa cells were first transfected with negative control (NC) or GORASP2 RNAi oligo (GORASP2 KD) for 48 h, and then with GFP-RAB7A and FLAG-CCZ1 for 24 h. Cells were treated with glucose starvation medium for another 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. Whole-cell lysates were blotted to show the expression of GFP-RAB7A, FLAG-CCZ1 and endogenous GORASP2. (G) Quantification of the interaction efficiency between RAB7A and CCZ1 in (F). n = 3 independent experiments. (H) GORASP2 depletion decreased the interaction between CCZ1 and PIK3C3. HeLa cells were first transfected with control negative (NC) or GORASP2 RNAi oligo (GORASP2 KD) for 48 h, and then with CCZ1-GFP and FLAG-PIK3C3 for 24 h. The cells were treated with glucose starvation medium for another 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. Whole-cell lysates were blotted to show the expression of CCZ1-GFP, FLAG-PIK3C3 and endogenous GORASP2. (I) Quantification of the interaction efficiency between CCZ1 and PIK3C3 in (H). n = 3 independent experiments. Statistical analyses values were shown as means ± SEM. Significance was determined using unpaired two-tailed Student’s t test in this figure. *p < 0.05, **p < 0.01.

It has been previously shown that MON1A directly interacts with PtdIns3P and that the GEF activity of the MON1A-CCZ1 complex on RAB7A is highly stimulated on membranes rich in PtdIns3P [18,19]. It has also been demonstrated that GORASP2 interacts with BECN1 to help assemble the PtdIns3K UVRAG complex during autophagosome maturation [29]. We hypothesized that GORASP2 deletion could cause defects in the assembly of UVRAG on autophagosomes, subsequently reducing the amount of PtdIns3P on the autophagosome membrane. This, in turn, might decrease GEF recruitment and activity. To validate this hypothesis, we assessed the interaction between CCZ1 and PIK3C3, which is primarily responsible for producing PtdIns3P in the UVRAG complex. As anticipated, knockdown of GORASP2 impaired the interaction between CCZ1 and PIK3C3 (Figure 4H,I). In summary, these results demonstrate that GORASP2 is required for the MON1A-CCZ1 GEF activity for RAB7A activation by regulating the assembly of PtdIns3K on autophagosomes.

GORASP2 promotes HOPS interaction with RAB7A

The fusion of autophagosomes with lysosomes requires the HOPS tethering complex, which is composed of six subunits: VPS18, VPS33A, VPS11, VPS16, VPS39, and VPS41. The latter two subunits are unique to the HOPS compared to the class C core vacuole/endosome tethering/CORVET complex. The HOPS complex collaborates with Rab GTPase and SNARE complexes to carry out the fusion process [38]. While GORASP2 also promotes autophagosome-lysosome fusion by connecting LC3 and LAMP2, its tethering function differs from that of HOPS. Therefore, we investigated the relationship between GORASP2 and HOPS components VPS39, VPS41, VPS18 and VPS33A. Co-IP analysis revealed that under control conditions, GORASP2 had a slight interaction with HOPS, which was significantly enhanced after GSB treatment (Figure 5A–E). Furthermore, we validated that glucose starvation caused a significant increase in the interactions between GORASP2 and HOPS, regardless of the presence of BafA1 (Figure S5A-S5H). Notably, VPS39 and VPS41, located at opposite poles of the HOPS complex and responsible for linking lysosomes and autophagosomes [39], both interacted with GORASP2, particularly under glucose starvation (Figure 5A,B,E; Figure S5A-S5D). Furthermore, VPS41 directly interacted with GORASP2 in an in vitro GST affinity-isolation assay (Figure 5F; Figure S5I). These results on protein interactions suggest that GORASP2 may regulate the HOPS function during autophagy.

Figure 5.

GORASP2 enhances HOPS interaction with RAB7A. (A-D) GORASP2 interacted with subunits of the HOPS complex upon glucose starvation. GFP or GFP-tagged GORASP2 (GORASP2-GFP) was co-transfected with MYC-tagged HOPS components VPS39 (A), VPS41 (B), VPS18 (C), and VPS33A(D) in HeLa cells. Cells were then treated with control medium (Con) or glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and MYC. Whole-cell lysates were blotted to show the expression of GORASP2-GFP and MYC-tagged HOPS components. “*” indicated unspecific bands. (E) Quantification of the interaction efficiency between GORASP2 and the subunits of HOPS in (A-D). n = 3 independent experiments. (F) GORASP2 directly bound VPS41. GST or GST-tagged VPS39 (indicated by arrowhead) and VPS41 (indicated by arrow) recombinant proteins were incubated with His-GORASP2 (His-GORASP2), affinity isolated with glutathione beads, and blotted for His. (G) GORASP2 depletion did not affect the HOPS assembly. GFP or GFP-tagged VPS41 was co-transfected with MYC-VPS39 in HeLa WT or GORASP2 knocked-out (GORASP2 KO) cells as indicated, treated with glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and MYC. Whole-cell lysates were blotted to show the expression of VPS41-GFP and MYC-VPS39. (H and I) GORASP2 depletion reduced the association between RAB7A and VPS39/VPS41. GFP or GFP-tagged RAB7A was co-transfected with MYC-tagged VPS39 (H) and VPS41(I) respectively in HeLa WT or GORASP2 knocked-out (GORASP2 KO) cells as indicated. Cells were treated with glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and MYC. Whole-cell lysates were blotted to show the expression of GFP-RAB7A, MYC-VPS39 (H), and MYC-VPS41 (I). (J) Quantification of the interaction efficiency between RAB7A and VPS39/VPS41 in (H and I). n = 3 independent experiments. Statistical analyses values were shown as means ± SEM. Significance was determined using unpaired two-tailed Student’s t test in this figure. *p < 0.05, **p < 0.01, ***p < 0.001.

Next, we tested whether GORASP2 modulates the assembly of the HOPS complex. The results showed that VPS39 and VPS41 maintained a strong interaction with or without GORASP2 (Figure 5G), indicating that GORASP2 might affect the association between HOPS and its regulators. The HOPS complex promotes autophagosome-lysosome fusion by binding with RAB7A through VPS41 and VPS39 [40,41]. As expected, when GORASP2 was depleted, the interaction between RAB7A and VPS41, as well as VPS39, was reduced (Figure 5H–J). These results indicate that GORASP2 facilitates autophagosome maturation by promoting the interaction between RAB7A and its effector, the HOPS complex.

GORASP2 binds and colocalizes with the SNARE complexes responsible for autophagosome maturation

In the autophagic process, SNARE complexes mediate the final fusion step between the autophagosome and the lysosome. In mammals, the STX17-SNAP29-VAMP8 and YKT6-SNAP29-STX7 complexes are recognized as key contributors to this fusion [42,43]. Our investigation revealed that GSB treatment significantly increased the interactions between GORASP2 and SNARE complexes, compared to the control conditions. Moreover, SNAP29 interacted with GORASP2 only under GSB conditions (Figure 6A–F). We further confirmed this conclusion under different treatments. The interaction between GORASP2 and STX17 or YKT6 were slightly enhanced with BafA1, while all SNARE proteins significantly improved interaction with GORASP2 upon glucose deprivation (Figure S6A-S6I). Furthermore, in a co-IP assay comparing the binding affinity of GORASP2 with YKT6 or STX17, it was observed that GORASP2 tended to interact more with STX17 than with YKT6 under both control and glucose starvation conditions (Figure 6G,H). This aligns with the hypothesis that the YKT6-SNAP29-STX7 complex likely functions as a second operational SNARE complex alongside to the STX17-SNAP29-VAMP8 complex in mammals [23,42].

Figure 6.

GORASP2 binds and colocalizes with SNARE complexes in autophagosome maturation. (A-E) GORASP2 interacted with SNAREs upon glucose starvation. Using GFP protein as a negative control, YFP/GFP-tagged SNARE proteins (A and E) and GFP-tagged GORASP2 (B-D) were co-transfected with FLAG-tagged GORASP2 (A and E) or FLAG-tagged SNARE proteins (B-D). Cells were treated with control medium (Con) or glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. Whole-cell lysates were blotted to show the expression of exogenous GORASP2 and SNARE proteins. (F) Quantification of the interaction efficiency between GORASP2 and SNAREs in (A and C-E). n = 3 independent experiments. (G) GORASP2 interacted with STX17 more than YKT6. GFP-YKT6 or YFP-STX17 was co-transfected with GORASP2-FLAG in HeLa cells. Cells were treated with control medium (Con) or glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. Whole-cell lysates were blotted to show the expression of GFP-YKT6, YFP-STX17 and GORASP2-FLAG. (H) Quantification of the interaction efficiency between GORASP2 and YKT6/SXT17 in (G). n = 3 independent experiments. (I) GORASP2 puncta colocalized with SNAREs during glucose starvation. HeLa cells were transfected with mCherry-tagged SNARE proteins as indicated, treated with control medium (Con) or glucose starvation medium with BafA1 (GSB) for 4 h, and stained for GORASP2 antibody. mCherry-YKT6 was co-transfected with GORASP2-GFP, and cells were pre-permeabilized with digitonin before fixation. Scale bar: 20 μm, 2 μm (inserted panel). (J) Quantification of (I) for the colocalization of GORASP2 puncta and indicated proteins. mCherry, n = 30 cells in each group; STX17, n = 35 cells; SNAP29, n = 37 cells; VAMP8, n = 37 cells; STX7, n = 38 cells; YKT6, n = 40 cells. (K) GORASP2 directly bound VAMP8 and YKT6. GST or GST-tagged STX17, SNAP29, VAMP8, YKT6, and STX7 recombinant proteins were incubated with His-GORASP2, affinity isolated with glutathione beads, and blotted for His. Statistical analyses values were shown as means ± SEM. Significance was determined using unpaired two-tailed Student’s t test in this figure. “ns” indicated not significant, *p < 0.05, **p < 0.01, ***p < 0.001.

Through the induction of autophagy via glucose deprivation and the inhibition of autophagosome-lysosome fusion by BafA1, SNARE was observed to be stuck in the final step of autophagosome-lysosome fusion. The endogenous GORASP2 puncta were found to colocalize with mCherry-tagged SNARE proteins, including STX17, SNAP29, VAMP8, and STX7 (Figure 6I,J). For mCherry-YKT6 co-transfected with GORASP2-GFP, the cells were pre-permeabilized with digitonin before fixation to extract cytosolic YKT6 and reduce the background signal. The results showed that GORASP2-GFP puncta colocalized remarkably well with mCherry-YKT6 (Figure 6I,J). In contrast, GORASP2 barely formed puncta outside of the Golgi area in control medium (Figure 6I,J). To determine whether GORASP2 directly interacts with the SNARE complex, we conducted a GST affinity-isolation assay and the results showed that GORASP2 directly bound to both R-SNAREs VAMP8 and YKT6 (Figure 6K), indicating that GORASP2 may regulate both STX17-SNAP29-VAMP8 and YKT6-SNAP29-STX7 SNARE complexes.

GORASP2 depletion impairs SNARE assembly and localization on autophagosome

To comprehend how GORASP2 regulates SNAREs in autophagy, we determined whether GORASP2 influences the formation of the two SNARE complexes by comparing WT and GORASP2-depleted cells. The levels of endogenous SNAP29, VAMP8, and YKT6 remained consistent with or without GORASP2, indicating that GORASP2 does not affect their expression and stability (Figure S3D). However, FLAG-VAMP8 was noticeably less co-precipitated with YFP-STX17 in cells lacking GORASP2 under glucose starvation (Figure 7A,B). Furthermore, a more significant decrease was observed in the assembly of YKT6-SNAP29-STX7 in GORASP2-depleted cells (Figure 7C,D).

Figure 7.

GORASP2 depletion impairs SNARE assembly and localization on autophagosomes. (A) GORASP2 depletion impaired STX17-SNAP29-VAMP8 assembly. GFP (as negative control) or YFP-STX17 was co-transfected with FLAG-VAMP8 in HeLa WT or GORASP2-depleted cells (GORASP2 KO). Cells were treated with glucose starvation medium with BafA1 (GSB) for 4 h, immunoprecipitated with a GFP antibody, and blotted for GFP and FLAG. Whole-cell lysates were blotted to show the expression of YFP-STX17 and FLAG-VAMP8. (B) Quantification of the interaction efficiency between STX17 and VAMP8 in (A). n = 3 independent experiments. (C) GORASP2 depletion impaired STX7-SNAP29-YKT6 assembly. GFP (as negative control) or GFP-STX7 was co-transfected with FLAG-YKT6 in HeLa WT or GORASP2-depleted cells. The same treatment and co-immunoprecipitation procedure were performed as in (A) to determine the interaction between FLAG-YKT6 and GFP-STX7. (D) Quantification of the interaction efficiency between STX7 and YKT6 in (C). n = 3 independent experiments. (E) GORASP2 depletion reduced the colocalization between YKT6 and LC3. HeLa WT or GORASP2-depleted cells (GORASP2 KO) were transfected with mCherry-YKT6, treated with glucose starvation medium with BafA1 for 4 h, and pre-permeabilized with digitonin before fixation, stained for LC3 antibody. Scale bar: 10 μm, 1 μm (inserted panel). (F) Quantification of (E) for the colocalization of YKT6 puncta and LC3. n = 50 cells in each group. (G) GORASP2 depletion reduced membrane association of SNAP29 and YKT6 during glucose starvation. HeLa WT or GORASP2-depleted (GORASP2 KO) cells were incubated in glucose starvation medium with BafA1 for 4 h, then were subjected to subcellular fractionation to separate the cytosol and membranes and western blotting. (H) Quantification of the ratio of SNAP29 on membranes in (G). n = 3 independent experiments. (I) Quantification of the relative ratio of YKT6 on membrane in (G). n = 3 independent experiments. (J) Schematic model of the role of GORASP2 in autophagosome maturation. In the late stages of autophagy, GORASP2 assists in ESCRT-III-mediated phagophore closure by facilitating VPS4A and CHMP2A interaction and promoting VPS4A recruitment on the autophagosome. After that, GORASP2 assists in the assembly of the UVRAG complex on autophagosomes, affecting the production of PtdIns3P, which is crucial for the GEF activity and membrane localization of MON1A-CCZ1. The activation of RAB7 by MON1A-CCZ1 recruits downstream factors, including HOPS, for autophagosome-lysosome tethering. Finally, GORASP2 promotes the assembly of SNARE complexes STX17-SNAP29-VAMP7 and YKT6-SNAP29-STX7 that mediate the final step of autophagosome-lysosome membrane fusion by directly interacting with R-SNARE VAMP8 and YKT6. Statistical analyses values were shown as means ± SEM. Significance was determined using unpaired two-tailed Student’s t test in this figure. *p < 0.05, **p < 0.01, ***p < 0.001.

The recruitment of YKT6 to autophagosomes in mammals is not yet fully understood [42]. Given that GORASP2 directly binds to YKT6 (Figure 6K), we investigated whether GORASP2 affects the recruitment of YKT6 to autophagosomes. Immunofluorescence analysis showed decreased colocalization of YKT6 with LC3 in GORASP2-depleted cells compared to WT cells (Figure 7E,F).

As YKT6 is predominantly distributed in the cytosol, we further examined the impact of GORASP2 on YKT6 localization using a biochemical assay. We separated membranes from the cytosol through ultracentrifugation of the post-nuclear supernatant (PNS) and analyzed YKT6 levels in each fraction by western blot. We found that although the majority of YKT6 remained in the cytoplasm, the association of endogenous YKT6 with membranes was significantly reduced in GORASP2-depleted cells (Figure 7G,I). The fact that GORASP2 depletion did not entirely block the membrane recruitment of YKT6 indicates that there might be other factors responsible for YKT6 localization to autophagosomes. Furthermore, we observed a significant reduction in the recruitment of SNAP29 to the membranes in the absence of GORASP2 (Figure 7G,H). These results indicate that GORASP2 depletion impairs the recruitment of the cytosolic SNAREs to autophagosomes. Taken together, GORASP2 plays a crucial role in facilitating SNARE assembly during glucose starvation.

Discussion

In our previous research, we demonstrated that GORASP2, a Golgi tether protein, acts as a crucial link between autophagosomes and lysosomes during autophagy triggered by nutrient starvation [29,30]. In this study, we delved deeper into the mechanisms by which GORASP2 collaborates with the classical membrane fusion machinery involved in autophagy. Our biochemical and cellular experiments uncovered that GORASP2 is located on the surface of autophagosomes and plays pivotal roles in multiple processes during later stages of autophagy. Firstly, GORASP2 directly binds to and promotes the recruitment of VPS4A to the autophagosomes, facilitating the interaction between VPS4A and the ESCRT-III component CHMP2A. This helps ESCRT-III-mediated phagophore closure. Secondly, GORASP2 facilitates the assembly of the UVRAG complex on autophagosomes [29]. The UVRAG complex is essential for the production of PtdIns3P, which is crucial for the GEF activity of MON1A-CCZ1. Thirdly, once RAB7A is activated by MON1A-CCZ1, downstream effectors such as the HOPS complex are recruited to the autophagosome and lysosome for membrane fusion. Finally, GORASP2 also promotes the assembly of SNARE complexes, including STX17-SNAP29-VAMP8 and YKT6-SNAP29-STX7, which are vital for the final step of autophagosome-lysosome membrane fusion (Figure 7J).

We have previously shown that GORASP2 is located on the surface of autophagosomes through biochemical methods and electron microscopy. In this study, we used a more advanced super-resolution fluorescence microscope to determine the relative position of GORASP2, LC3, and LAMP2 on the surface of autophagosomes (Figure 1A,B). Based on our previous observations that GORASP2 functions in the later stages of autophagy [30], we focused on its roles in phagophore closure. Both the established FPP assay and the HaloTag-LC3 assay indicated impaired autophagosome membrane sealing without GORASP2 (Figure 1D–G), prompting a more in-depth investigation into the novel role of GORASP2 in phagophore closure.

While the function of ESCRT in phagophore closure is well-established [7–9], the exact regulatory mechanism remains incompletely understood. We investigated the relationship between GORASP2 and key components of the ESCRT complex. Surprisingly, components in every sub-complex of ESCRT interacted with GORASP2 during glucose starvation (Figure 2A–D; Figure S1A-S1D, S2A-2F). GORASP2 colocalized with CHMP2A, CHMP4B, CHMP6 in ESCRT-III, and VPS4A in the ESCRT accessory proteins, suggesting that GORASP2 may function together with ESCRT-III and VPS4 in phagophore closure (Figure 2E; Figure S1E). GORASP2 depletion attenuated VPS4A recruitment to the autophagosomes and the interaction between VPS4A and CHMP2A (Figure 2H–K), further confirming its role in autophagosome sealing. While it remains possible that GORASP2 may work together with ESCRT-0, -I, and -II in phagophore closure, the timing of their presence on the autophagosome might be too brief to be observed by microscopy with fixed staining. Further investigation is needed to understand how GORASP2 collaborates with the entire ESCRT complex step by step under stress.

The autophagosome-lysosome fusion event requires activated RAB7A to recruit its effectors. We found that the activity of RAB7A is impacted in the absence of GORASP2 (Figure 3E,F). As the MON1A-CCZ1 complex is the only known RAB7A GEF, we investigated the relationship between GORASP2 and MON1A-CCZ1. Our results revealed that during glucose starvation, GORASP2 significantly interacted with CCZ1 and MON1A (Figure 4A–C), facilitating the membrane recruitment and GEF activity of MON1A-CCZ1 for RAB7A (Figure 4D–G). Depletion of GORASP2 decreased the interaction between RAB7A and CCZ1, suggesting that GORASP2 regulates the GEF activity for RAB7A (Figure 3E). The impairment of RAB7A activity is believed to contribute to the autophagosome maturation defect observed in GORASP2 knockout cells.

The HOPS complex, a conserved heterohexameric tethering complex, serves as the effector of RAB7A, tethering autophagosomes to lysosomes for fusion [22,39]. Our results suggest that GORASP2, identified as a new linker between autophagosomes and lysosomes [30,44], acts as an upstream regulator of RAB7A (Figure 4). Thus, we asked how these two distinct membrane tethers work together in the autophagy process. Our findings suggested that during glucose starvation, GORASP2 interacts more with the subunits of the HOPS complex (Figure 5A–E), suggesting collaboration in autophagy. When GORASP2 is depleted from cells, the interaction between HOPS and RAB7A is reduced (Figure 5H–J), indicating that decreased RAB7A activity due to GORASP2 depletion eliminates the interaction between RAB7A and its effector proteins. GORASP2 may also regulate the recruitment of HOPS to autophagosomes, as indicated by its direct binding with VPS41 (Figure 5F; Figure S5I).

The process of membrane fusion involves the spontaneous assembly of four complementary SNARE motifs in the SNARE complex [45]. The STX17-SNAP29-VAMP8 plays a major role in this process, while the recently identified STX7-SNAP29-YKT6 complex complements it in the fusion of autophagosomes and lysosomes [42]. During glucose starvation, GORASP2 interacts and colocalizes with subunits of both SNARE complexes (Figure 6A–F). GORASP2 interacts more strongly with STX17-SNAP29-VAMP8 (Figure 6G,H), supporting the hypothesis that it plays a more important role in autophagosome maturation. Depletion of GORASP2 impairs the assembly of both complexes, with a greater influence on the assembly of STX7-SNAP29-YKT6 (Figure 7A–D). This may be due to GORASP2 directly binding more with YKT6 than VAMP8 (Figure 6K). Additionally, depletion of GORASP2 decreases YKT6’s localization on autophagosomes (Figure 7E,F) and significantly reduces the recruitment of SNAP29 to the membrane (Figure 7G,H), indicating impaired SNARE assembly due to GORASP2 absence.

Taken together, our study provides a detailed understanding of how the Golgi tether protein, GORASP2, extends its role to other membrane fusion processes during stress. Recruited by LC3 onto autophagosomes during glucose starvation, GORASP2 collaborates with ESCRT-III and VPS4A in autophagosome sealing. It further facilitates UVRAG assembly and RAB7A activation by MON1A-CCZ1, subsequently promoting SNARE assembly for final autophagosome-lysosome fusion. GORASP2 works extensively with the classical membrane fusion machinery as a tether, indicating intricate crosstalk between membranes, especially in response to changes in the cellular microenvironment.

Materials and Methods

Plasmids and Antibodies

pEGFP-N1-GORASP2, pcDNA3.1/Myc-His(-)A-GORASP2, pGEX-4T-1-hMAP1LC3B, pET-30a(+)-GORASP2 WT were previously described [30]. GFP-RAB7A WT, GFP-RAB7A^T22N, GFP-RAB7A^Q67L were kindly provided by Dr. Haoxing Xu (Zhejiang University, China). pHAGE-N-GFP-CCZ1, pCMV6-Entry-MON1A-Myc-FLAG were kindly provided by Dr. J Wade Harper (Harvard Medical School, USA). FLAG-PIK3C3 was kindly provided by Dr. Lishenglan Xia (Huazhong Agricultural University, China). pCMV-Myc-VPS39, pCMV-Myc-VPS41, pDmyc-VPS18, pDmyc-VPS33A were kindly provided by Dr. Tuanlao Wang (Xiamen University, China). YFP-STX17 was kindly provided by Dr. Edward A. Fon (McGill University, Canada). FLAG-SNAP29 was kindly provided by Dr. Qing Zhong (Shanghai Jiao Tong University, China). Halo-LC3 was constructed into pFN21K-Halo Tag-CMV. MBP-RILP was constructed into pMAL-c2X. CHMP3-GFP, CHMP6-GFP, CHMP2A-GFP, VTA1-GFP, STAM-GFP, MVB12A-GFP, VPS41-GFP were constructed into pEGFP-N1. GFP-CHMP4B, GFP-VPS4A, GFP-VPS4B, GFP-HGS, GFP-STAM2, GFP-TSG101, GFP-SNF8, GFP-VPS36, GFP-VPS28, GFP-STX7, GFP-YKT6 were constructed into pEGFP-C1. GST-CHMP2A, GST-CHMP4B, GST-VPS4A, GST-CHMP3, GST-CHMP6, GST-RAB7A WT, GST-RAB7A^T22N, GST-RAB7A^Q67L, GST-VPS39, GST-VPS41, GST-VPS41 aa1–394, GST-VPS41 aa395–854, GST-STX17, GST-SNAP29, GST-VAMP8, GST-STX7, GST-YKT6 were constructed into PGEX-6P-1. GORASP2-FLAG, VPS4A-FLAG were constructed into pcDNA3.1-C-3×Flag. FLAG-CCZ1, FLAG-STX17, FLAG-VAMP8, FLAG-YKT6 were constructed into pBICEP-CMV2. MON1A-mCherry was constructed into pmCherry-N1. mCherry-LC3, mCherry-STX17, mCherry-SNAP29, mCherry-VAMP8, mCherry-STX7, mCherry-YKT6 were constructed into pmCherry-C1.

Antibodies used in this study include monoclonal antibodies against LAMP2 (Developmental Studies Hybridoma Bank, H4B4), MYC (Santa Cruz Biotechnology, sc-40), FLAG (Proteintech Group 66,008–4-Ig), GFP (Proteintech Group 66,002–1-Ig), His (Proteintech Group,66005–1-Ig), GST (Santa Cruz Biotechnology, sc-138), MBP (Proteintech Group 66,003–1-Ig), VPS18 (Santa Cruz Biotechnology, sc -100,890), VPS39 (Santa Cruz Biotechnology, sc -514,762), VPS41 (Santa Cruz Biotechnology, sc -377,118), BLZF1/Golgin45 (Santa Cruz Biotechnology, sc -515,193), ACTB (Proteintech Group 66,009–1-Ig), SNAP29 (Santa Cruz Biotechnology, sc -390,602), VAMP8 (Santa Cruz Biotechnology, sc -166,820), YKT6 (Santa Cruz Biotechnology, sc -365,732), TUBA1B/Alpha Tubulin (Proteintech Group 66,031–1-Ig); polyclonal antibodies against LC3 (MBL International, PM036), GORASP2 (Proteintech Group 10,598–1-AP), RAB7A (Proteintech Group 55,469–1-AP), CCZ1 (Proteintech Group 22,159–1-AP), MON1A (Proteintech Group 23,772–1-AP), SQSTM1/p62 (Proteintech Group 18,420–1-AP), GOSR1/GS28 (Proteintech Group 16,106–1-AP).

Cell culture and transfection

HeLa and U-2 OS cells were cultured in Dulbecco’s modified Eagle’s medium (Gibco, C11995500BT) supplemented with 10% fetal bovine serum (NEWZERUM, FBS-S500) and 100 units/ml penicillin-streptomycin (Gibco 15,140,122) at 37°C with 5% CO₂. Polyethylenimine (PEI; BIOHUB, 78PEI25000) was used for transient transfection of plasmids. To perform co-immunoprecipitation assay, cells were seeded and grown to 70–80% confluency before transfection. For immunofluorescence assay, cells were seeded and grown to 40–50% confluency before transfection. To carry out transfection, the PEI reagent was diluted into DMEM and mixed with plasmids. The mixture was incubated for 15 min at room temperature before adding it into the cells. For GORASP2 knockdown experiments, HeLa cells were transfected with Lipofectamine RNAiMAX (Invitrogen 13,778,075) and GORASP2 siRNA oligo as described previously [28]. For MBP-RILP affinity-isolation assay, cells were analyzed 72 h after siRNA transfection. For the Co-immunoprecipitation experiment, HeLa cells were transfected with siRNA for 48 h and then transfected with indicated proteins for another 24 h. Control nonspecific RNAi oligo and GORASP2 siRNA were purchased from GenScript. GORASP2 targeting sequence (AACTGTCGAGAAGTGATTATT) was previously described [24].

Modulation of Autophagy

For glucose starvation, cells were extensively washed with phosphate-buffered saline (PBS; Invitrogen 21,600,069) and then changed with DMEM medium that contains 25 mM glucose as control group (Con) or DMEM medium without glucose (Gibco,11966025) as glucose starvation group (GS) for 4 h. 400 nM BafA1 were added to the control group (CB) or the glucose starvation group (GSB) to inhibit autophagosome-lysosome fusion.

Fluorescence protease protection (FPP)

HeLa wild-type (WT) and GORASP2-depleted (GORASP2 KO) cells were plated on glass bottom dishes (Biosharp, BS-20-GJM), respectively, and transfected with mCherry-LC3 for 20 h, treated with glucose starvation medium with 400 nM BafA1 for another 4 h. Remove cell culture medium and wash cells three times in KHM buffer (100 mM potassium acetate, 20 mM HEPES, 2 mM MgCl₂, pH 7.4) at room temperature. Cells were then incubated with 20 μM digitonin (Sigma-Aldrich, D141) for 5 min. Then cells were treated with 50 μg/ml proteinase K (MedChemExpress, HY-108717) and real-time images were immediately recorded by Andor confocal spinning disk on a Nikon inverted microscope Eclipse Ti-E.

HaloTag-LC3 assay

Halo-LC3 assays were performed as previously described [8]. Briefly, HeLa WT and GORASP2 depleted cells (GORASP2 KD) were transfected with Halo-LC3 for 24 h and treated with GSB medium for another 4 h. Cells were washed with KHM buffer and treated with 20 μM digitonin in KHM buffer at 37°C for 2 min. Cells were incubated with membrane-impermeable ligand (MIL) HaloTag® Alexa Fluor® 488 (Promega, G1001) at 37°C for 15 min. Cells were then fixed with 4% paraformaldehyde (PFA; Sigma-Aldrich 158,127) for 15 min, washed three times with PBS, and incubated with membrane-permeable ligand (MPL) HaloTag® TMR (Promega, G8251) for 25 min at room temperature. Coverslips were washed with PBS three times and mounted on the microscope slides with antifade mounting medium with DAPI (Beyotime, P0131) and examined by confocal microscope (Nikon A1 HD25).

Immunoblotting

To perform immunoblotting, the cells were first washed twice with cold PBS, then lysed at 4°C for 20 min in lysis buffer (20 mM HEPES, pH 7.4, 140 mM KCl, 5 mM MgCl₂, 1% Triton X-100 [BioFroxx, 1139ML500] and EDTA-free protease inhibitor cocktail [MedChemExpress, HY-K0010]). The lysate was centrifuged for 20 min at 4°C and 16,000 g. The supernatants were collected and mixed with 6 × SDS sample loading buffer. The mixture was boiled at 98°C for 5 min, then subjected to SDS-PAGE. The nitrocellulose membrane (Cytiva 10,600,002) was used to transfer the protein and was blocked with 5% skim milk (BioFroxx, 1172GR500). Corresponding primary and secondary antibodies were used to incubate the membrane. The signals were detected using SuperSignalTM West Pico Plus Chemiluminescent Substrate (Thermo Fisher Scientific 34,578) and Bio-Rad ChemiDoc XRS+ System with Image Lab™ Software.

Co-immunoprecipitation

Indicated plasmids were used to transfect HeLa cells. GFP was used as a negative control to detect interaction proteins of GFP-tagged proteins. The transfected cells were then treated with different types of medium (control medium, control medium plus BafA1, glucose starvation, or glucose starvation plus BafA1) for 4 h. After the treatment, the cells were washed with cold PBS and lysed in immunoprecipitation (IP) buffer (20 mM HEPES, pH 7.4, 140 mM KCl, 5 mM MgCl₂, 1% Triton X-100 and EDTA-free protease inhibitor cocktail) for 20 min at 4°C. 600 μL IP buffer was used for a 10-cm petri dish cell lysis. The lysate was cleared by centrifugation, 80 μL supernatants were used as input sample and the remaining supernatants were incubated with the corresponding beads overnight at 4°C. GFP antibodies were pre-incubated with Protein A 4FF Sefinose (TM) Resin (Sangon Biotech, C600957) to generate GFP-trap beads for use in GFP trap. The bound proteins were washed three times with IP buffer and analyzed by immunoblotting after being eluted with 6 × SDS sample buffer. 20% of the immunoprecipitated samples were used to detect GFP proteins, while the remaining 80% were used to detect co-immunoprecipitated proteins.

Immunofluorescence microscopy

Immunofluorescence microscopy was performed as described previously [29]. Briefly, cells were grown on glass coverslips and transfected with the indicated plasmids, then performed corresponding treatment after transfection. After washed twice with PBS, cells were fixed in 4% PFA, followed by quenching with 50 mM NH₄Cl and permeabilized with 0.3% Triton X-100. The fixed cells were incubated with indicated primary antibodies at 4°C overnight, then washed with PBS three times, and further incubated with secondary antibodies at room temperature for 1 h. The coverslips were finally mounted with antifade mounting medium with DAPI. The images were collected with the corresponding microscopic imaging system. Nikon Structured Illumination Microscope (N-SIM) was used for the super-resolution microscopy imaging in Figure 1A. Images were captured with an EMCCD camera (Andor iXon DU-897) and a 100 × 1.49 NA TIRF objective (Nikon CFI Apo TIRF). Image acquisition and reconstruction were performed with Nikon NIS-Elements software. Confocal and STED (Stimulated Emission Depletion Microscopy) images in Figure 1B were acquired using Abberior STEDYCON (Abberior Instruments GmbH, Göttingen, Germany) fluorescence microscope built on a motorized inverted microscope I×83(Olympus UPlanXAPO 100×, NA1.45, Tokyo, Japan). Huygens professional software was used for post-processing deconvolution of the raw STED images. Line profile was analyzed with STEDYCON Gallery software. Confocal Microscope (Nikon A1 HD25) was used for cell imaging in all the other immunofluorescence images. Images were captured with a100 × 1.49 NA TIRF objective (Nikon CFI Apo TIRF), and the following image acquisition and deconvolution were performed with Nikon NIS-Elements software.

In vitro GST affinity-isolation assay

GST, GST-tagged CHMP2A, CHMP4B, VPS4A, CHMP3, CHMP6, RAB7A WT, RAB7A^T22N, RAB7A ^Q67L, VPS39, VPS41, VPS41 aa1–394, VPS41 aa395–854, STX17, SNAP29, VAMP8, YKT6, STX7 or His-tagged GORASP2 were expressed in BL21(DE3) bacteria and purified with glutathione Sepharose 4B beads (GE Healthcare 17,075,605) and Ni-NTA agarose beads (Thermo Fisher Scientific 88,222), respectively. To identify the protein that interacts directly with GORASP2, 3 μg of the GST or GST-tagged protein was mixed with 3 μg of His-tagged GORASP2 in GST affinity-isolation buffer (20 mM Tris-HCl, pH7.5, 150 mM NaCl, 1 mM MgCl₂, 0.5% NP-40 [Solarbio, N8030]). The mixture was incubated at 4°C for 2 h and then glutathione Sepharose 4B beads were added. After another hour, the bound proteins were thoroughly washed and eluted with 6 × SDS sample. Finally, the eluted proteins were analyzed by western blot with an anti-His antibody.

MBP-RILP affinity-isolation assay

RILP affinity-isolation assay was performed as described previously [37]. In brief, MBP or MBP-RILP protein was expressed in BL21 bacteria and purified using Amylose Resin (New England Biolabs, E8021S). HeLa WT and GORASP2 knocked-down cells were treated with glucose starvation medium and lysed with lysis buffer (20 mM Tris HCl, pH 7.8, 137 mM NaCl, 1% glycerol, 0.5% Triton X-100, 1 mM MgCl₂, and EDTA-free protease inhibitor cocktail). After preparing the lysates, 10% of them were kept aside for the input analysis. Following this, MBP-control beads and MBP-RILP beads were incubated with the lysates separately overnight at 4°C. The mixture was then washed three times with lysis buffer and analyzed through western blot to obtain the results.

Subcellular fractionation

The process of subcellular fractionation was carried out following the protocol described in a previous study [29]. First, the cells were washed with PBS and then homogenization buffer (0.25 M sucrose, 1 mM EDTA, 1 mM magnesium acetate, 10 mM HEPES-KOH, pH 7.2, and protease inhibitors). Cells were pipetted and resuspended in 800 μl of homogenization buffer. To crack the cells, a ball bearing homogenizer (Isobiotec Cell Homogenizer) was used, and the process was monitored under a microscope by trypan blue (Thermo Fisher Scientific 15,250,016), until 75–80% of the cells were broken. The homogenate was then centrifuged at 1000 × g and 4°C for 10 min, and the pellet was isolated as the post-nuclear supernatant (PNS). The PNS was then subjected to ultracentrifugation in a TLA55 rotor at 120,000 × g for 1 h, resulting in the collection of cytosols in the supernatant and membranes in the pellet. Equal volume fractions of the cytosol and membrane were analyzed by western blotting.

Generation of CRISPR-Cas9 KO cell lines

The targeting sequence for GORASP2 (TCGCAAAGCGTCGAGATCCC) was inserted into lentiCRISPR v2 plasmid (Addgene 52,961; deposited by Dr. Feng Zhang). Constructed plasmids were transfected into HeLa cells for 24 h. Cells were treated with Puromycin (MedChemExpress, HY-B1743A) 1.5 μg/ml for 4–5 days and single cell clones were isolated on a 96-well plate. The KO cells were identified through DNA sequencing and immunoblotting.

Quantification and statistical analysis

Quantification data represent the mean ± SEM (standard error of the mean) of at least three independent experiments. At least 20 cells were counted for colocalization analysis. A statistical analysis was conducted with a two-tailed Student’s t-test. Differences in means were considered statistically significant at p < 0.05. Significance levels are: *, p < 0.05; **, p < 0.01; ***, p < 0.001. Analyses were performed using ImageJ.

Funding Statement

This work was supported by the National Natural Science Foundation of China (X.Z., grant number: 32070693).

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2024.2375785