A role for CASM in the repair of damaged Golgi architecture

Abstract

The term CASM describes a process in which LC3 and other Atg8 proteins are covalently ligated to lipids in damaged endomembranes. While CASM is commonly described as a cytoprotective response to multiple types of membrane damage, the ways in which CASM helps cells maintain homeostasis are still unclear. Here, we show that CASM contributes to the maintenance or repair of Golgi apparatus architecture following the loss of TRIM46, a ubiquitin ligase with roles in microtubule organization. TRIM46-deficient cells were notable for enhanced TFEB-driven lysosomal biogenesis and Golgi ribbon fragmentation, with colocalization between the trans-Golgi marker TGN46 and the Atg8 proteins LC3B and GABARAP. Similar results were seen when Golgi architecture was disrupted by inhibitors of microtubule assembly or of vesicle trafficking. Further studies revealed that the Golgi atg8ylation seen in TRIM46 knockout cells was not degradative and mechanistically resembled CASM. Genetic inhibition of CASM in TRIM46-deficent cells reduced TFEB activation and exacerbated the Golgi morphology defects. Together, these studies reveal that lysosomal biogenesis and CASM are common features of a Golgi damage response, with CASM acting to preserve Golgi integrity.

INTRODUCTION

The proper organization, placement, and maintenance of membranous organelles is fundamental to the life of eukaryotic cells. This requires the repair or replacement of damaged membranes or dysfunctional membrane-associated proteins. Multiple mechanisms of membrane repair have evolved, likely due to the critical importance of maintaining membrane integrity and function, the diversity of organelle membranes within a cell, and the variety of possible causes of membrane damage. Depending on the type of membrane damage, cellular responses can include: i) ESCRT-mediated excision of damaged membranes; ii) delivery of new membrane to the site of damage by organelle fusion or by membrane-to-membrane lipid transfer; and iii) enhanced expression and biosynthesis of protein constituents of damaged membranes¹. Additionally, multiple membrane damage scenarios trigger the attachment of a set of ubiquitin-like proteins, collectively referred to as Atg8s (LC3A, LC3B, LC3C, GABARAP, GABARAP L1, and GABARAP L2), to membrane in a process termed ‘atg8ylation’^2–4.

Membrane atg8ylation contributes to membrane damage responses in several ways. These include enabling the lysosome-mediated removal of damaged membranes or membranous organelles^5–8, indirectly up-regulating the biosynthesis of new membrane components^9–11, and providing a “patch” to plug damaged membranes¹². Mechanistically, the atg8ylation process resembles ubiquitylation, with the roles of E1 activating enzyme and E2 conjugating enzyme played by ATG7 and ATG3, respectively³. There are two multi-protein E3 ligase complexes that are known to act in atg8ylation: the ATG5-ATG12-ATG16L1 complex¹³ and the ATG5-ATG12-TECPR1 complex^14,15. The E3 ligase complexes attach Atg8 proteins to either phosphatidylethanolamine (PE) or phosphatidylserine (PS). Macroautophagy (hereafter, autophagy) is the best-studied pathway that utilizes membrane atg8ylation. In autophagy, Atg8 proteins are attached to de novo-created membrane that elongates in a curved manner and ultimately sequesters cytoplasmic contents within a sealed double-membraned vesicle termed an autophagosome¹⁶. Autophagosomes are then trafficked along microtubules towards lysosomes, where the two organelles fuse and the inner autophagosomal membrane and its luminal contents are degraded¹⁷. More recently, additional processes that require the atg8ylation machinery have been identified. These processes differ from autophagy in that they involve atg8ylation of pre-existing membranes, they do not result in the formation of autophagosomes, and they are not always degradative². The term CASM (conjugation of Atg8 to single membranes) has been coined to describe these atg8ylation-dependent pathways¹⁸. While both autophagy and CASM utilize the same atg8ylation machinery, they differ in the mechanism by which the E3 ligase complex is recruited to the sites of atg8ylation. In autophagy, the ATG16L1-containing E3 complex is recruited to phosphatidylinositol-3-phosphate containing membranes generated by the BECN1/hVPS34 complex. The BECN1 (Beclin 1) complex is activated by a protein complex containing the kinase ULK1/2. On the other hand, BECN1 and the ULK1 complex are dispensable for CASM^18–20. Finally, the two pathways differ in terms of which lipid is conjugated to Atg8s, with PE being exclusively utilized by autophagy whereas both PE and PS are utilized by CASM²¹. Despite their mechanistic differences, both autophagy and CASM are typically considered protective responses to membrane damage, although the functions of CASM remain largely undefined.

Ubiquitination can be a key regulator of atg8ylation-dependent processes²². This is particularly true in the case of autophagy. For example, many members of the tripartite motif containing (TRIM) family of ubiquitin ligases are reported to act as both autophagy regulators and in the identification of membranous or proteinaceous autophagy substrates^23–25. The role(s) of TRIM proteins, or ubiquitination more generally, in other atg8ylation-dependent processes such as CASM have not yet been explored.

Here, our investigation into the actions of TRIM proteins in atg8ylation led us to find that disruption of microtubule organization is a potent activator of the non-degradative atg8ylation of trans-Golgi membranes. This subsequently activates a program of lysosomal biogenesis and autophagy activation. We found that TRIM46, a ubiquitin ligase previously implicated in the formation of an axonal structure in neurons²⁶, is important for microtubule organization. Genetic depletion of TRIM46 resulted in fragmentation of the trans-Golgi network (TGN) and activated TFEB/TFE3, the master transcriptional activators of lysosome- and autophagy-related gene expression. TRIM46 knockout also resulted in substantial atg8ylation of the TGN in a manner reminiscent of CASM. When atg8ylation is blocked, the Golgi morphology defects seen in TRIM46 knockout cells were exacerbated, while the activation of TFEB was attenuated. The activation of both CASM and TFEB seen in TRIM46 knockout is phenocopied by chemical inhibitors of microtubule function or vesicle trafficking, all of which disrupt Golgi architecture. Overall, our study indicates that CASM-dependent activation of lysosomal biogenesis is a general response perturbations of Golgi architecture, with CASM having a role in membrane repair or reorganization.

RESULTS

Screening of TRIM proteins for roles in autophagosome maturation.

Since TRIM proteins have many reported roles in regulating autophagy initiation^23,27,28, we wondered if their actions in autophagy extend to mediating autophagosome-lysosome fusion and autophagy flux. To address this question, we reanalyzed data from a previous siRNA screen that examined the effect of TRIM knockdown on the abundance of the autophagosome marker LC3B²⁹. This screen made use of cells expressing mCherry-eYFP-LC3B (tandem fluorescent LC3B, tfLC3B) as a marker of autophagosomes. Although tfLC3B can be used to monitor both the formation of autophagosomes and their delivery to lysosomes (termed autophagosome maturation) due to the differential sensitivity of the two fluorophores to low pH^30,31, in our initial analysis we only considered the abundance of YFP-positive LC3B structures²⁹. Here, we determined whether the siRNA-mediated knockdown of any one of 68 TRIMs in HeLa cells altered the delivery of the LC3B reporter to acidified compartments, which can be detected by comparing the relative abundance of “non-acidified” LC3B structures (positive eYFP signal) with that of “total” LC3B structures (positive for mCherry signal; Fig. 1A, B and S1A). We found that cells subjected to knockdown of six TRIMs (TRIMs 25, 38, 46, 47, 61, and 68) reduced the acidification of the LC3B reporter by more than three standard deviations below the mean of cells transfected with non-targeting siRNAs in two out of two experiments (Fig. 1B and S1A). This indicates a possible role for these TRIMs in autophagosome maturation. Alternatively, these TRIMs could act to inhibit non-degradative atg8ylation pathways such as CASM.

Figure 1. TRIM46 regulates non-degradative LC3 lipidation.

(A) siRNA screen for TRIMs regulating the accumulation of non-acidified LC3B. HeLa cells stably expressing mCherry-eYFP-LC3B were transfected with TRIM or control siRNA two days prior to fixation and high content imaging. Neutral pH LC3B puncta (eYFP positive; blue mask) or total LC3B puncta (yellow mask) were identified and quantitated from >500 cells per siRNA. White mask, nuclei; magenta mask, cell boundary. (B) TRIM knockdowns (red data points) with LC3B acidification reduced by more than three standard deviations (3X SD) below the mean of non-targeting siRNA controls (blue data points) were identified as hits. Magenta numbers indicate TRIMs ‘hits’ that were identified in two out of two experiments (see also Figure S1A). (C–D) High-content imaging analysis of LC3B abundance in WT and two independent TRIM46 knockout clones. Each dot represents the average LC3B puncta area from >500 cells. (E, F) Immunoblot analysis of LC3B-II levels in TRIM46 knockout cells. Quantification of LC3B-II from 3 independent experiments. (G, H) Halo-LC3 assay for autophagy flux. WT and TRIM46 knockout cells stably expressing HT-LC3B were pulsed with TMR-HL for 30 minutes prior to 6 h chase with full media. HT/HL were detected by in-gel fluorescence. LE, long exposure. Quantitation (H) shows the percent of total HT/HL signal that results from the “released” HT/HL, a product of lysosomal degradation. (I, J) High-content imaging analysis of LC3B (cyan) and LAMP2 (magenta) localization in WT and two TRIM46 knockout clones. Each data point in (J) represents the average area of LC3B and LAMP2 overlap per cell from >500 cells. (K-M) Immunoblot analysis of LC3B-II and SQSTM1 abundance in lysates from WT or TRIM46 knockout HeLa cells treated with DMSO or 100 nM bafilomycin A1 (BafA1) for 6 hours. Plots show quantification of relative protein abundance of SQSTM1 (L) and LC3B-II (M). Each dot represents an independent experiment. Data: mean ± SEM; *, p < 0.05; ***, p < 0.001; ****, p < 0.0001; ns, not significant.

Knockdown of TRIM46 increases non-degradative atg8ylation without impairing autophagy flux.

We chose to focus on TRIM46 as a possible regulator of autophagosome maturation because of its reported actions on microtubule organization in neurons, which we reasoned could impact proper autophagosome and/or lysosome trafficking. We generated several clonal populations of HeLa and HEK293T cells in which TRIM46 was knocked out by CRISPR/Cas9 (Fig. S1B, C). High content imaging of two of the HeLa cell lines showed that the abundance of endogenous LC3B puncta was elevated by ~3-fold in the TRIM46 knockout cells relative to wild-type HeLa cells stably expressing Cas9 and non-targeting gRNA (Fig. 1C, D). Immunoblot experiments also showed that TRIM46 knockout HeLa cells (Fig. 1E, F) or HEK293T cells (Fig. S1D) showed strongly elevated levels of LC3B-II, the form of LC3B that is generated by atg8ylation. Exogenous expression of TRIM46 reversed the effect of TRIM46 knockout on LC3-II abundance (Fig. S1E, F). Together, these data show that TRIM46 attenuates the accumulation of lipidated LC3B.

Although the purpose of our initial screen was to identify TRIMs that might be required for autophagosome maturation, further study showed that TRIM46 deficiency did not impair autophagy flux. This was first demonstrated using the Halo-LC3 assay, which relies on the fact that Halo tag/Halo ligand adducts (HT/HL) are protected from lysosomal degradation. When autophagy flux is active, the lysosome will incompletely degrade HT/HL-fused LC3 (48 kDa), leaving a 33 kDa HT/HL fragment that can readily be detected in SDS-PAGE gels when the Halo ligand is fused to a fluorescent molecule³². We detected the HT/HL fragment in lysates from both WT and TRIM46 knockout cells expressing HT-LC3B and treated with tetramethylrhodamine (TMR)-labeled HL under basal autophagy conditions (Fig. 1G). Although more released HT/HL is present in TRIM46 knockout cells, quantitation of the released HT/HL band relative to the total abundance of TMR signal (Fig. 1H) indicated that WT and TRIM46 knockout cells had comparable levels of autophagy flux. Image analysis showed that colocalization between LC3B and the lysosome marker LAMP2 in both lines of TRIM46 knockout HeLa cells was elevated relative to WT (Fig. 1I, J and S1G, H), indicating no impairment in autophagosome/lysosome fusion. Treatment with the lysosomal inhibitor Bafilomycin A1 (BafA1) increased the abundance of the autophagy substrate SQSTM1 in both WT and TRIM46 knockout cells (Fig. 1K, L), demonstrating functional autophagy flux.

Interestingly, however, the effect of BafA1 on LC3-II levels in TRIM46 knockout cells was muted relative to what was measured in WT cells in which it strongly increased the abundance of LC3-II (Fig. 1K, M). Similar results were seen when using the Halo-LC3 assay (Fig. S1I). BafA1 treatment increased the abundance of HT/HL-LC3B-II in WT cells pulsed with TMR-HL by ~3-fold. In contrast, BafA1 had no effect on the abundance of HT/HL-LC3B-II in the TRIM46 knockout cells. As discussed later, our interpretation for this result as well as those from our tfLC3B experiments (Fig. 1B and S1A) is that not all of the membrane-associated LC3B in TRIM46 knockout cells is delivered to acidified compartments for degradation and that atg8ylation pathways in addition to autophagy are activated in TRIM46-deficient cells.

TRIM46 restrains lysosomal biogenesis

In the experiments shown in Figure 1I and S1E, we noted that the number of LAMP2-positive structures appeared elevated in the TRIM46 knockout cells, a result that we confirmed by high content imaging (Fig. 2A, B). The expression of membrane and luminal proteins of lysosomal proteins also tended to be higher in TRIM46 knockout HeLa cells (Fig. 2C, D) and HEK293T cells (Fig. S2A, B). Thus, lysosomal abundance is increased in TRIM46 knockout cells. To further assess lysosomal function in TRIM46 knockout cells, we used DQ-BSA, a molecule that becomes fluorescent upon hydrolysis by lysosomal proteases in an acidic environment³³. TRIM46-knockout cells exhibited increased DQ-BSA signal relative to WT cells (Fig. S2C, D). However, when DQ-BSA signal was normalized to the abundance of lysosomes based on LAMP2 staining, TRIM46 knockout cells were indistinguishable from WT (Fig. S2E). This indicates that the degradative functions of lysosomes are intact in TRIM46 knockout cells. These data suggest that the increased lysosome abundance is not due to the accumulation of dysfunctional lysosomes, but instead due to lysosomal biogenesis.

Figure 2. TRIM46 deficiency increases lysosome biogenesis.

(A, B) High-content imaging analysis of LAMP2 abundance in WT HeLa and two TRIM46 knockout clones. The average area of punctate LAMP2 per cell from >500 cells was quantitated and graphed in (B). (C, D) The levels of the indicated proteins in WT and TRIM46 knockout cells were measured by immunoblotting (C) with quantitation from 6 independent experiments shown in (D). (E, F) High content image analysis of TFEB localization in WT and TRIM46 knockout HeLa cells. The percentage of cells showing nuclear-localized TFEB was plotted in (F). Each data point represents the average calculated from >500 cells. (G-I) Immunoblot analysis of the effects of TRIM46 deficiency on mTORC1 and AMPK kinase activity. Quantification of relative protein abundance of phospho-AMPKα (T172) and phospho-P70-S6K (T389) are shown in (H) and (I), respectively. Each data point represents an independent experiment. Data: mean ± SEM; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

Lysosomal biogenesis is controlled at the transcriptional level by the MiTF/TFE family of transcription factors, which includes the protein TFEB. TFEB shuttles between the cytoplasm and nucleus depending on its phosphorylation status to activate the expression of a large number of autophagy- or lysosome-related genes^34–36. TFEB staining revealed that TRIM46-knockout cells exhibited a nearly two-fold increase in TFEB nuclear localization compared to WT cells (Figure 2E, F), consistent with enhanced lysosomal biogenesis in TRIM46 knockout cells. TFEB is under the control of the master metabolic regulating kinases mTORC1 and AMPK^36–40. TRIM46 knockout reproducibly increased the level of active phosphorylated AMPK while the phosphorylation of p70 S6 kinase, a substrate of mTORC1, was decreased (Fig. 2G–I). Active mTORC1 is primarily localized on lysosomes⁴¹. In confocal microscopy experiments, we found that the lysosomal localization of mTOR was reduced in TRIM46 knockout cells (Fig. S2F, G), further indicating reduced mTORC1 activation. Overall, these data show that TRIM46 restrains TFEB-mediated lysosomal biogenesis associated with mTORC1 inactivation.

trans-Golgi apparatus is disrupted and subject to non-degradative atg8ylation in TRIM46 knockout cells.

We next considered how TRIM46 deficiency triggered TFEB activation. As the master regulator of lysosome biogenesis, TFEB activation is strongly associated with lysosomal damage^42,43. However, our data shown in Fig. S2 indicates that the lysosomes in TRIM46 knockout cells are functional. We thus considered whether damage to other organelles could explain the phenotypes observed in TRIM46 knockout cells. Structural disruption of the Golgi has been reported to inactivate mTOR signaling while inducing atg8ylation^9,44,45. The architecture of the Golgi apparatus is maintained by microtubules^46,47. TRIM46 localizes to microtubules and has been implicated in bundling axonal microtubules in neurons²⁶. We thus hypothesized that Golgi structure might be disrupted in TRIM46 knockout cells via microtubule disorganization, and that this may explain the mTOR and TFEB phenotypes that we have observed. In agreement with this hypothesis, we found that mCherry-tagged TRIM46 exclusively localized to microtubules when expressed in HeLa cells (Fig. S3A). The microtubules in TRIM46 knockout cells were shorter and more disorganized than in WT cells (Fig. 3A–C). Along with the microtubule phenotypes, we observed that the Golgi apparatus in TRIM46 knockout cells appeared to be larger and less coalesced. We also noticed that the number of small structures positive for the trans-Golgi marker TGOLN2/TGN46 (trans-Golgi network protein 2; TGN46) was substantially increased in TRIM46 knockout cells (Fig. 3D). As a measure of Golgi fragmentation, we used high content imaging and analysis to selectively identify “small” TGN46-positive structures (smaller than 3 μm²) and quantitate them in WT and TRIM46 knockout cells. Using this analysis, we saw that the number of Golgi fragments was increased by ~2-fold in both TRIM46 knockout cell lines relative to WT (Fig. 3E and S3B). Additionally, the total area of TGN46, including both small and large TGN46-positive structures, is increased in TRIM46 knockout cells (Fig. 3F). Although imaging analysis revealed an increased Golgi area in TRIM46 knockout cells, the abundance of Golgi-resident proteins including GCP60 and the trans-Golgi protein TGN46 was unchanged, and only a slight increase was observed in the endosomal/TGN localized protein WIPI1 (Fig. S3C). These findings suggest that the increased Golgi area reflects structural dispersal rather than an upregulation of Golgi protein synthesis in TRIM46 knockout cells. Together, these data demonstrate that TRIM46 is essential for maintaining Golgi architecture, likely through TRIM46’s ability to organize microtubules.

Figure 3. TRIM46 deficiency triggers microtubule disorganization and Golgi fragmentation.

(A-C) High-content imaging analysis of microtubule organization in WT and TRIM46 knockout (shown, KO #4) HeLa cells. WT and TRIM46 knockout HeLa cells were fixed and stained with anti-α-tubulin antibodies. (A) Short (α-tubulin structures < 0.794 μm²; cyan mask) and long (α-tubulin structures > 0.794 μm²; yellow mask) microtubule structures were automatically segmented. (B) The relative abundance of short microtubules per cell was quantified as a percentage of total microtubules and compared between WT and of TRIM46 knockout cells. (C) The standard deviation of angles of long microtubule fibers, a measure of microtubule disorganization, was calculated per cell, and the mean values were compared between WT and TRIM46 knockout cells. (D) Maximum image projection confocal images of WT and TRIM46 knockout HeLa cells stained with trans-Golgi marker TGN46 and cis-Golgi marker GM130. (E, F) High-content imaging analysis of trans-Golgi network (TGN) fragmentation in WT and TRIM46 knockout cells. Representative images are in Supplementary Figure S3B. TGN fragmentation was quantified by determining the number of “small” TGN46-positive structures (<3 μm²) per cell (E). The total area of TGN46-positivity per cells was quantitated and plotted relative to what is seen in WT cells (F). Each data point represents the average of >500 cells. Data: mean ± SEM; **, p < 0.01, ****; p < 0.0001.

In our siRNA screen (Fig. 1), we observed that TRIM46 knockdown increased the abundance of non-acidified LC3B structures, suggesting that TRIM46 may inhibit a non-degradative atg8ylation pathway. Several other groups have observed non-degradative conjugation of LC3 to Golgi membranes^9,45,48, and so we wondered if some of the LC3-positive structures seen in TRIM46 knockout cells colocalized with the trans-Golgi marker TGN46. Indeed, we found substantial overlap between the small TGN46-positive structures and the Atg8 proteins LC3B and GABARAP in TRIM46 knockout cells (Fig. 4A–F; S4A–B). Analysis of three-dimensional reconstructions of deconvolved confocal images of TRIM46 knockout cells that had been probed with antibodies recognizing TGN46 and LC3B revealed that the LC3B-positive structures that colocalized with TGN46 had irregular morphologies that corresponded with the shape of the TGN46 structure (Fig. S4C). The shape of these LC3B-positive structures was different from what might be expected of autophagosomes, which tend to be round. In agreement with this observation, quantitative analysis showed that the LC3B-positive structures in TRIM46 knockout cells showed reduced sphericity when compared to LC3B-positive structures in WT cells (Fig. S4D). Subcellular fractionation experiments also indicated association between lipidated LC3B and TGN46-positive membranes (Fig. S4E). To enrich lysosomal and trans-Golgi network (TGN) vesicles, we performed sequential, differential centrifugation including ultracentrifugation at 100, 000 g. The resulting pellets were lysed, and equal amounts of protein were analyzed.

Figure 4. TRIM46 knockout induces Golgi atg8ylation.

(A) Confocal analysis of colocalization between LC3B and TGN46 in WT and TRIM46 knockout (shown, KO #4) HeLa cells. Zoomed in images of the boxed regions are shown below. Arrows indicate colocalized signal. (B) Colocalization between TGN46 and LC3B was quantified using Pearson’s correlation coefficient analysis. Each point represents data from a different confocal image. (C-F) High content image analysis of WT and TRIM46 knockout HeLa cells stained with antibodies against GABARAP and TGN46. The average abundance of GABARAP puncta, TGN46 area, and GABARAP/TGN46 overlap area were quantified per cell. Data points represent biological replicates, each based on an average of >500 cells. (G) Confocal analysis of mCherry-ATG16L1 and TGN46 localization in transiently transfected TRIM46 knockout cells. Small arrows indicate colocalizing puncta. An enlarged image of the boxed region is shown. The large arrow indicates the path of the measured intensity profile. Data: mean ± SEM; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

Fractionation was confirmed by immunoblotting with LAMP2 and TGN46 as lysosomal and TGN markers, respectively. In WT cells, lipidated LC3B was predominantly found in 21k fraction, with substantially less LC3B-II cofractionating with TGN46 in the 100k fraction. In contrast, roughly equal amounts of LC3B-II were found in the 21k and 100k fractions in TRIM46 knockout cells (Fig. S4E). The findings presented above are consistent with direct conjugation of LC3B to the TGN46-positive membranes. In agreement with this concept, ATG16L1, a key component of the atg8ylation machinery, shows strongly enhanced colocalization with TGN46 in TRIM46 knockout cells (Fig. 4G and S4F–H). We next used mKeima-fused YIPF3 as a reporter to determine if the atg8ylated Golgi fragments were delivered to the lysosome for degradation in a selective autophagy-based process called ‘Golgiphagy’^7,49,50. YIPF3 is a Golgi-resident transmembrane protein that serves as a Golgiphagy receptor and is degraded by autophagy following amino acid starvation^49,50. We saw no difference in the delivery of the mKeima-YIPF3 reporter to acidified compartments when we compared WT with TRIM46 knockout cells, despite positive controls (amino acid starvation) and negative controls (BafA1) behaving as expected in this assay (Fig. S4I). These results, along with data showing that TRIM46 knockout does not reduce the abundance of several Golgi-resident proteins (Fig. S3C), show that TRIM46 knockout does not increase Golgiphagy, and instead imply that the observed Golgi atg8ylation is non-degradative.

TRIM46 knockout activates CASM of trans-Golgi membranes

Our data indicate that TRIM46 knockout induces CASM-mediated atg8ylation of TGN46-positive membranes. To further validate this, we inhibited or knocked down proteins that are essential for autophagy but are dispensable for CASM^51–53. When WT cells are treated with VPS34-IN1, a compound that inhibits phagophore formation, LC3-II levels were reduced by ~60%. In contrast, this treatment only modestly impacted LC3B-II levels in TRIM46 knockout cells (~20% reduction; Fig. 5A, B). Next, we employed siRNA to knock down the expression of the core autophagy proteins ATG7, ATG13, BECN1, and ULK1 to determine how this impacted the elevated levels of LC3-II and the increased abundance of LC3-positive structures in TRIM46 knockout cells (Fig. 5C–F; S5A–F). As expected, knockdown of ATG7 reduced lipidated LC3B and LC3B puncta. ATG7 knockdown also reduced the colocalization between LC3B and TGN46 (Fig. 5G; S5C). However, knocking down the expression of ATG13 and BECN1 did not reduce the abundance of LC3-II or LC3 puncta in TRIM46 knockout cells. ULK1 knockdown seemed to reduce atg8ylation in TRIM46 knockout cells, but this trend was not statistically significant (Fig. 5C–G; S5D–G). These data confirm that much of the excessive atg8ylation seen in TRIM46 knockout cells is attributable to CASM.

Figure 5. Enhanced atg8ylation in TRIM46 knockout is independent of canonical autophagy factors.

(A-B) WT or TRIM46 knockout HeLa cells were treated with 10 μM of VPS34-IN1 for 4 hours prior to lysis and immunoblotting (A). Graph (B) shows the relative reduction, expressed as a percent, in LC3-II abundance following VPS34-IN1 treatment. Data points represent independent experiments. (C-D) WT or TRIM46 knockout HeLa cells were transfected with the indicated siRNA prior to lysis, immunoblotting (C), and quantitation of LC3B-II levels relative to that seen in WT cells (D). Data points represent independent experiments. (E-G) Representative high content images of TRIM46 knockout HeLa cells transfected with the indicated siRNA and stained with antibodies recognizing LC3 and TGN46. Plots show the abundance of punctate LC3B area per cells (F) and the overlapping area between TGN46 and LC3B (G). (H) Left, immunoblot analysis of the effect of VPS34-IN1 and BafA1 on LC3B lipidation in WT and TRIM46 knockout cells. Cells were treated with inhibitors (VPS34-IN1: 10 μM, BafA1: 100 nM) or DMSO vehicle control for 4h. Plot (right) shows the abundance of LC3B-II relative to that seen in DMSO-treated WT cells. (I) Representative images showing TFEB staining in TRIM46 knockout HeLa cells after transfection with control or ATG7 siRNA. Graph shows the percentage of cells with nuclear TFEB localization as determine by high content imaging. (J) Impact of ATG7 knockdown on TGN46 area in TRIM46 knockout cells. Data: mean ± SEM; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001; ns, not significant. Data points represent biological replicates, each based on an average of more than 500 cells for microscopy experiments.

Deacidification of cellular compartments such as lysosomes or ER is a known driver of CASM⁵⁴. This process, also called VAIL (V-ATPase-ATG16L1-induced LC3 lipidation)⁵⁵, requires the complete assembly and activity of V-ATPase structures on the membranes of the deacidified organelles⁵⁶. V-ATPase then recruits the ATG16L1 E3 ligase complex through direct protein-protein interaction⁵⁷. We asked whether V-ATPase assembly is required for the elevated atg8ylation seen in TRIM46-knockout cells. V-ATPase-dependent LC3 lipidation can be blocked by BafA1⁵⁵, which inhibits V-ATPase assembly. However, the inhibitory effects of BafA1 on CASM-related atg8ylation can be hard to detect since BafA1 also causes the accumulation of autophagy-related lipidated LC3B. To overcome this challenge, we used VPS34-IN1, which minimally impacts atg8ylation in TRIM46 knockout cells (Fig. 5A, B) to block autophagy while simultaneously treating cells with BafA1. When this experiment was performed on WT cells, our results indicated that autophagy was the primary driver of atg8ylation, with BafA1 treatment increasing and VPS34IN-1 decreasing the abundance of LC3B-II. VPS34-IN1 completely prevented the BafA1-dependent accumulation of lipidated LC3B (Fig. 5H). In contrast, TRIM46 knockout cells showed evidence for both ongoing autophagy and CASM, since neither BafA1 nor VPS34IN-1 on their own impacted the already elevated levels of lipidated LC3B seen in these cells. However, the levels of LC3B-II were significantly reduced when TRIM46 knockout cells were treated with both inhibitors (Fig. 5H). Thus, when autophagy is inhibited with VPS34-IN1, we can see a clear inhibitory effect of BafA1 on atg8ylation in TRIM46-deficient cells. This result establishes a role for the V-ATPase in Golgi membrane atg8ylation in TRIM46-deficient cells.

Impacts of CASM on TFEB activation and Golgi morphology in TRIM46-deficient cells

In most cases, the physiological consequences of CASM are not entirely clear. We next used ATG7 knockdown to determine whether atg8ylation contributes to the phenotypes seen in TRIM46 knockout cells. We found that siRNA-mediated knockdown of ATG7 expression substantially reduced the percentage of TRIM46 knockout cells that were positive for TFEB nuclear localization (Fig. 5I), suggesting that CASM may contribute to the increased lysosomal biogenesis seen in TRIM46 knockout cells. Our data also indicate a role for CASM in governing Golgi architecture in TRIM46 knockout cells. As discussed above, the trans-Golgi network in TRIM46 knockout cells is disrupted and shows an increased area per cell (Fig. 3D–F). We found that ATG7 knockdown in TRIM46 knockout cells exacerbated this phenotype, further increasing the cross-sectional area of TGN46 staining per cell relative to TRIM46 knockout cells transfected with non-targeting siRNA (Fig. 5J). Since we did not see evidence of Golgiphagy (Fig. S3C and S4I), we assume that Golgi atg8ylation is likely serving to either maintain or repair Golgi architecture.

Chemical disruption of the Golgi architecture phenocopies TRIM46 knockout

We have demonstrated that TRIM46 deficiency leads to disorganization of the microtubule network, Golgi fragmentation, and CASM activation. To determine whether microtubule disorganization alone is sufficient to mimic the phenotypes seen in TRIM46 knockout cells, we treated WT HeLa cells with the microtubule-depolymerizing agents vinblastine and nocodazole. As expected, both treatments significantly increase TGN46-positive fragments (Fig. 6A, B and S6A, B). Similar to TRIM46 knockout cells, both vinblastine- and nocodazole-treated cells showed a significant increase in LC3B puncta and LC3B/TGN46 colocalization, suggesting CASM activation (Fig. 6C–F; S6C–E). Microtubule disruption also increased lysosome abundance (Fig. 6G, H; S6F–G) and increased TFEB nuclear localization (Fig 6I, J; S6H–I), mirroring the effects observed in TRIM46 knockout cells. Thus, disorganization of the microtubule network, whether by drug treatment or by TRIM46 deficiency, results in CASM and lysosomal biogenesis.

Figure 6. Microtubule disruption phenocopies TRIM46 deficiency, triggering Golgi atg8ylation and increased lysosomal biogenesis.

WT HeLa cells were treated with 100 nM of Vinblastine for 16 hours. (A-B) High-content imaging analysis of TGN fragmentation measured by counting TGN46-positive fragments smaller than 3 μm² (cyan mask). Magenta mask indicates TGN46-positive structures with surface areas >3 μm². (C) Confocal microscopic analysis of TGN46 and LC3B colocalization in HeLa cells treated with Nocodazole, Vinblastine, or DMSO control. Arrows indicate colocalizes structures. Insets show magnified images of the area within the dashed-line boxes. (D-F) Representative high-content microscopy images of TGN46 and LC3B colocalization following vinblastine treatment. Plots show coalesced LC3-positive area per cell (E), and overlapping area between TGN46 and LC3B (F). (G-H) High-content imaging analysis of WT HeLa cells of LAMP2 abundance following vinblastine treatment. (I-J) High content imaging analysis of TFEB nuclear localization following vinblastine treatment. (K) Model of TRIM46 role in maintaining microtubule organization. Left panel, TRIM46 localizes to microtubules and maintains the organization of microtubule fibers. Right panel, in the absence of TRIM46, microtubules are less organized, leading to Golgi fragmentation and mTORC1 inactivation. In response, v-ATPase subunits are assembled on Golgi-membranes, triggering atg8ylation. Atg8ylation of the Golgi helps restore Golgi architecture and promotes TFEB activation and lysosomal biogenesis. Data: mean ± SEM; ****, p < 0.0001. Each data point represents the average of more than 500 cells.

We next asked whether we could recapitulate the phenotypes seen in TRIM46 knockout cells by disrupting Golgi apparatus architecture without targeting microtubules. To test this, we treated HeLa cells with brefeldin A, a compound that induces Golgi disassembly by interfering with the formation of COP-I vesicles. As expected, brefeldin A increased the number of Golgi fragments per cell in WT HeLa cells (Fig. S6J, K). Additionally, we found that it also significantly increased the abundance of LC3B puncta per cell and the colocalization between LC3B and TGN46 (Fig. S6L–N). Brefeldin A treatment also resulted in a >2-fold increase in TFEB nuclear localization (Fig. S6O, P). These data show that Golgi atg8ylation and TFEB activation are generalized cellular responses to disruption of Golgi architecture.

DISCUSSION

Golgi architecture disruption is associated with multiple pathological conditions including viral infection, cancer, and neurodegenerative disease^58–62. How cells respond to or repair these architectural changes remains unclear. Our study has shown that disruption of the Golgi apparatus activates a generalized response including Golgi atg8ylation and TFEB-driven lysosomal biogenesis (Fig. 6K). We started our studies focusing on TRIM46, which likely alters Golgi structure indirectly through its actions in establishing microtubule organization. Supporting this concept, we found that treatment of wild type cells with the microtubule depolarizing compounds vinblastine or nocodazole phenocopied what we observed with the TRIM46 knockout cells. Interestingly, brefeldin A, which disrupts Golgi structure and function without impacting microtubule organization, also induced Golgi atg8ylation and TFEB activation. This result suggests that loss of Golgi architecture, rather than disruption of microtubule organization, is the primary trigger for these phenotypes. In agreement with this concept, blockage of post-Golgi trafficking by over-expression of the secreted protein DLK1⁹, down-regulation of a Golgi-localized membrane tether⁶³, or various Golgi-damaging treatments^45,64 all induce LC3 conversion.

Key questions remain about why cells engage the atg8ylation machinery in response to Golgi or other membrane stress. We found that TRIM46 deficiency constitutively activated Golgi atg8ylation. Inhibition of Golgi atg8ylation by knocking down ATG7, the E1 activating enzyme of the atg8ylation cascade, had two effects in TRIM46 knockout cells. First, it augmented the Golgi architecture defects caused by TRIM46-deficiency. This suggests that atg8ylation plays a role in the maintenance or repair of the Golgi ribbon structure, since our data excluded the possibility that Golgiphagy was responsible for removing damaged Golgi fragments in TRIM46 knockout cells. Our data also disfavors models in which atg8ylation results in increased Golgi biosynthesis or the delivery of new membrane to damaged Golgi structures, since these mechanisms would be expected to add size to the Golgi in an ATG7-dependent manner, which is the opposite of what we found. Further investigation is needed to identify how atg8ylation acts to control Golgi architecture.

Second, we found that the increased activation of the transcription factor TFEB seen in TRIM46 knockout cells was reversed following ATG7 knockdown. This result is consistent with several other recent findings that revealed a positive feedback loop between atg8ylation and TFEB activation^10,11. Golgi damage, by either TRIM46 deficiency or by chemical disruption of the Golgi network, strongly promotes TFEB activation to enhance lysosomal biogenesis and upregulate the expression of proteins involved in CASM and autophagy. We speculate that the TFEB-dependent responses to Golgi damage may be a cellular response to counter the source of Golgi damage or possibly in preparation to initiate Golgiphagy should the extent of Golgi damage become more severe. The second concept aligns with what is known about lysosomal membrane permeabilization, in which CASM is activated in response to mild lysosomal damage while lysophagy, the autophagic targeting of damaged lysosomes, seems to be reserved for lysosomes that are extensively damaged⁶⁵. It is possible that a similar relationship exists with Golgi membranes: while TRIM46 knockout does not cause sufficient damage to induce Golgiphagy, complete disruption of Golgi architecture with Brefeldin A is reported to induce autophagic degradation of Golgi-localized proteins⁶⁴.

Our studies with TRIM46, as well as other observations of Golgi atg8ylation in response to a variety of Golgi damaging conditions, raise the question of how the cell senses Golgi damage to activate atg8ylation. However, it is currently unknown if all of the various causes of Golgi damage activate the same damage-sensing pathways or if different forms of Golgi damage trigger different danger signals. Since the Golgi is a slightly acidic organelle, it is possible that some of the aforementioned Golgi damaging conditions could lead to proton leak and consequent deacidification of the Golgi lumen. In membranes of the endolysosomal system, this will lead to CASM in a manner requiring assembly of active V-ATPase complexes which recruits ATG16L1 complexes⁵⁴. Our data demonstrating that TRIM46 knockout increases ATG16L1 recruitment to Golgi membranes and that V-ATPase inhibition with BafA1 reduces LC3 lipidation in TRIM46 knockout cells are consistent with this model. Lysosomal deacidification also results in mTORC1 inactivation⁴¹. Interestingly, several studies have indicated that a pool of mTORC1 resides on the Golgi in addition to its well-established localization on lysosomes^63,66,67. We found that mTORC1 activity was inhibited in TRIM46-deficient cells, suggesting similar damage-responsive mechanisms between Golgi and lysosomes. Inhibition of mTORC1 activity can then result in or enhance autophagy and lysosomal biogenesis while also altering Golgi architecture^67,68.

Given the broad functions of TRIMs in autophagy, our initial goal was to screen TRIMs for roles in the late stages of autophagy. Interestingly, follow up experiments demonstrated that TRIM46 deficiency led to non-degradative CASM activation. In addition to TRIM46, the knockdown of five other TRIMs also showed reduced LC3B acidification in our screen (Fig. 1). Further investigations are warranted to determine whether these TRIMs regulate atg8ylation pathways indirectly, as in the case with TRIM46, or if they directly regulate autophagosome maturation.

In conclusion, our study shows that TRIM46 contributes to the organization of the microtubule network. TRIM46 deficiency destabilized microtubules and resulted in alteration to the Golgi ribbon structure, triggering CASM, lysosomal biogenesis, and changes in the activity of the key metabolic regulators mTORC1 and AMPK. Our work extends the growing body of literature documenting Golgi atg8ylation by demonstrating that CASM functions to maintain or repair Golgi architecture in response to Golgi damage.

METHODS

Antibodies

Primary antibodies were obtained as follows: mCherry (ab183628), GABARAP (ab109364; 1:500 for IF), and NPC2 (ab2181921:1,000 for WB) from Abcam; AMPK (2532; 1:1000 for WB), phospho-AMPK T172 (2535; 1:1,000 for WB), ATG13 (13468; 1:1,000 for WB), BECN1 (3459; 1:1,000 for WB), GM130 (12480; 1:500 for IF, 1:1,000 for WB), mTOR (2983; 1:500 for IF), phospho-P70 S6K (9205; 1:1,000 for WB), TFEB (4240; 1:500 for IF), and ULK1 (8054; 1:1,000 for WB) from Cell Signaling Technology; actin (sc-58673; 1:1,000 for WB), Cas9 (sc-517386, 1:1000 for WB), LAMP2 (sc-18822; 1:1,000 for WB, 1:500 for IF), P70 S6K (sc-8418; 1:1,000 for WB), and alpha-tubulin (sc-23948; 1:500 for IF) from Santa Cruz Biotechnology; Cathepsin D (21327–1-AP, 1:1,000 for WB), myc (16286–1-AP: 1:1,000 for WB), LIMP2 (27102–1-AP, 1:1,000 for WB), TRIM46(21026–1-AP, 1:1,000 for WB) from Proteintech; LC3B (PM036; 1:500 for IF) from MBL and LC3B (L7543; 1:2,000 for WB) from Sigma; p62/SQSTM1(610833; 1:2,000 for WB) from BD; ATG7 (MA5–32221; 1:1,000 for WB), GCP60 (MA5–25999; 1:1,000 for WB), TGN46 (MA5–37930; 1:1,000 for WB), and WIPI1 (PA5–34973; 1:1000 for WB) from Thermo Fisher and TGN46 (AHP500G; 1:500 for IF) from Bio-Rad. Secondary antibodies were purchased as follows: anti-rabbit Alexa Fluor 488 (A11008; 1:1,000 for IF), anti-sheep Alexa Fluor 568 (A21099; 1:1,000 for IF), anti-rabbit Alexa Fluor 647 (A21244; 1:1,000 for IF), and anti-mouse Alexa Fluor 647 (A21235; 1:1,000 for IF) from Thermo Fisher; anti-mouse IRDye 680LT (925–68020; 1:10,000 for WB) and anti-rabbit IRDye 800CW (926–32211; 1:5,000 for WB) from LI-COR.

Cell culture and treatment

All cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM; 11965084, Thermo Fisher) supplemented with 10% fetal bovine serum (FBS; 26140095, Thermo Fisher) and 100 U/mL penicillin-streptomycin (15140122, Thermo Fisher) in a humidified incubator at 37 °C with 5% CO₂. Bafilomycin A1 (tlrl-baf1, Invivogen; 100 μM), brefeldin A (B6542, Sigma; 5 mM), nocodazole (M1404, Sigma; 5 mM), Vps34-IN1 (Cayman chemical; 10 mM), vinblastine (11762, Cayman chemical; 100 μM) were prepared as stock solution in DMSO at the indicated stock concentration. To measure autophagy flux and CASM in TRIM46 knockout cells, cells were treated with 100 nM bafilomycin A1 with or without 10 μM vps34-IN1 for 4 or 6 hours. For experiments examining TFEB nuclear localization or LC3B and LAMP2 induction, cells were treated with 100 nM brefeldin A, nocodazole, or vinblastine for 16 hours.

Development of knockout or overexpressing cell lines

Knockout and stably overexpressing cell lines were generated by lentiviral transduction. Lentivirus was produced by co-transfecting HEK293T cells with pMD2.G, pPAX2, and pLV[CRISPR]-hCas9:T2A:Puro-U6-hTRIM46 (VB900138–6978huh, VectorBuilder) or pLEX307-Halo-LC3 at a 2:3:5 ratio using the ProFection Mammalian Transfection System (Promega). After 48hours, virus-containing supernatants were collected, cleared of residual mammalian cells by centrifugation at 500 × g for 10 minutes, and filter through a 0.45 μm vacuum filter (SE1M003M00, SIGMA). Target cells (HeLa or HEK293T) were incubated with viral supernatants diluted in DMEM for 48 hours and subsequently selected with 1 μg/mL puromycin (J61278.MB, Thermo Fisher). For TRIM46 knockout cells, single-cell clones were isolated, and knockout was validated using the Guide-it Complete sgRNA Screening System (632636, Takara). Briefly, sgRNA targeting TRIM46 was generated using a customized forward PCR primer containing T7 promoter sequence, TRIM46 target sequence, and Cas9 scaffold templates with the Guide-it sgRNA In Vitro Transcription Kit (Takara). The sgRNA template was validated on a 2% agarose gel, which showed a single band of ~130 bp. sgRNA was synthesized by In Vitro transcription using the validated template. Genomic DNA was extracted from WT or TRIM46 knockout cells using a genomic DNA extraction kit (95213–050, QuantaBio), and the target region containing the CRISPR/Cas9 cleavage site was amplified by PCR. After purification of PCR fragments, the Cas9 cleavage assay was performed. The reaction products were loaded on a 1.5% agarose gel and imaged using a ChemiDoc Imaging System (Bio-Rad). For Halo-LC3B expressing HeLa wild type or TRIM46 knockout cells, Halo-LC3B expression was confirmed by immunoblotting with anti-HaloTag antibody.

RNA interference and transfection

Plasmid transfections were performed using Lipofectamine 2000 (ThermoFisher). Small interfering RNAs (siRNAs) were purchased from GE Dharmacon and were supplied as pools of four different siRNA oligos against the same mRNA. siRNA transfections were performed using Lipofectamine RNAiMAX (ThermoFisher) according to the manufacturer’s instructions.

Cloning

pDONR221-hTRIM46 (HsCD00862242, DNASU) was used as the donor for cloning human TRIM46 into Gateway-compatible destination vectors, generating myc- or mCherry-tagged TRIM46 using LR Clonase (11791020, Thermo Fisher). All Plasmid constructs were verified by whole-plasmid sequencing. For addback experiments, the PAM site in myc- or mCherry-tagged TRIM46 was disrupted to prevent degradation by constitutively expressing Cas9 in TRIM46 knockout cells by introducing a silent mutation (1407G→T) using a site-directed mutagenesis kit (210518, Agilent).

TRIM family siRNA screen

The methods employed for this experiment, and the results from analyzing eYFP signal on its own, were previously published²⁹. Briefly, HeLa cells expressing mCherry-eYFP-LC3B were plated into 96-well plates containing pre-plated siRNA smart pools and transfection reagent (Dharmacon). Cells were fixed 48 hours after plating, stained with Hoechst 33342, and then imaged. Cell boundaries were determined based on nuclear staining with eYFP-positive and mCherry-positive LC3B puncta detected based on pre-set parameters in the iDev software. LC3B acidification was determined by subtracting the ratio of punctate eYFP-LC3B signal (neutral pH) to that of mCherry-LC3B signal (total LC3B) from 2; thus, values closer to 2 have increased delivery of the LC3B reporter to the lysosome while cells with inhibited delivery of LC3B to lysosomes will have reduced values. Those TRIMs whose LC3B acidification had values >3 standard deviations below the mean of cells transfected with non-targeting siRNA were considered hits. >500 cells per siRNA were analyzed in two independent experiments.

Immunofluorescent staining

For high content imaging, cells were seeded in either 24-well or 96-well plates (2 × 10 cells per well for a 24-well plate or 1 × 10 cells per well for a 96-well plate). For confocal experiments, 1 × 10⁵ cells were plated onto coverslips in 12-well plates. Cells were fixed with 4% paraformaldehyde for 3 minutes and then washed twice with 1 × PBS. In general, permeabilization was performed using 0.2% saponin and 1% BSA in PBS for 30 minutes, but for TFEB staining cells were permeabilized with 0.1% Triton X-100 and 0.1% Tween 20 for 10 minutes, followed by two washes with PBS and blocking with 1% BSA in PBS for 30 minutes. Primary antibodies were applied at the dilutions specified above and incubated with the cells for 1 hour at room temperature. Following two washes with PBS to remove residual primary antibody solution, cells were incubated with appropriate secondary antibodies for 1 hour at room temperature. Samples were then washed twice with 1 × PBS. Coverslips were mounted with ProLong Diamond Antifade mounting media (P36970, Invitrogen), while cells in multi-well plates were left in PBS for high content imaging.

High content microscopy and analysis

Following immunofluorescence staining as described above, cells were counterstained with Hoechst 33342 to visualize nuclei. High-content imaging was performed using a Cellomics HCS or CellInsight CX7 instruments and analyzed with iDEV software (Thermo Fisher). Hoechst 33342 nuclear staining was utilized for autofocusing, and regions of interest (ROIs) were defined relative to nuclear positions. Targets were identified based on fluorescence intensity within the ROI, and parameters such as object count, total area, total intensity, average intensity, co-localization area, and correlation coefficient between targets were analyzed on a per-cell basis using automated algorithms.

Puncta identification and quantitation, colocalization analysis, and analysis of nuclear localization: Puncta detection, colocalization, and nuclear localization were analyzed using the Cellomics colocalization BioApplication. Cell boundaries were defined based on either CellMask staining (H32722, Thermofisher) or the nuclei stained with Hoechst 33342 (H3570, Thermofisher). When nuclei were used to define cell boundaries, the nuclear parameter was extended to the surrounding cytoplasmic region. Single-cell segmentation was performed using nuclear intensity-based methods, and nuclei with irregular morphology or insufficient intensity were excluded to minimize segmentation errors. For puncta identification and quantification as well as colocalization analysis, the region of interest (ROI) was set to the cytoplasm, whereas for nuclear localization analysis, only the nuclear ROI was used. Target signals were measured within the defined ROI. Colocalization was quantified by automated calculation of the overlapping area between two target signals. Nuclear localization of TFEB was assessed by measuring the total nuclear intensity of TFEB, and the percentage of cells exceeding predefined nuclear intensity threshold was calculated. Analysis of trans-Golgi network fragmentation: Golgi morphology was analyzed using the Morphology Explorer BioApplication. TGN46-postitive structures larger than 3 μm² were classified as coalescent Golgi, whereas structures smaller than 3 μm² were classified as fragmented Golgi. The number of fragmented Golgi per cell was quantified and compared between WT and TRIM46 knockout cells. Analysis of microtubule organization: Microtubule organization was quantified using the Cellomics Morphology Explorer Bioapplication. Microtubules were visualized by immunofluorescence staining with α-tubulin (sc-23948, Santa Cruz) as described above. The software classified microtubule structures as either spotted or fibrous based on morphology, and fibers smaller than 0.794 μm² were defined as ‘short’ microtubules. To assess microtubule organization in TRIM46 knockout cells, the ratio of the total area of short microtubules versus the total microtubule area was calculated and compared to WT. Microtubule bundling was analyzed using the built-in fiber alignment feature of the software. The orientation of individual fibers was determined relative to the image axis, and the standard deviation of fiber angles within a single cell was used as a measure of microtubule alignment. A minimum of 500 cells per well were analyzed, with data from multiple wells per experiment and ≥3 independent experiments pooled for quantitative analysis.

Confocal microscopy and deconvolution

Sub-airy unit (0.6AU) pinhole confocal microscopy with a Zeiss LSM900 or a Leica TCS-SP8 microscope was performed followed by computational image restoration with Huygens Essential (Scientific Volume Imaging, Hilversum, Netherlands) utilizing a constrained maximum likelihood estimation algorithm. Images were acquired using 63X/1.4NA plan apochromat oil immersion objective lenses and sampled at ideal Nyquist sampling rates in x, y, and z planes, allowing for sub-diffraction limited resolution following image restoration. All images were rendered on a high performance CUDA-GPA enabled workstation and 3D renders were generated for morphological analysis with Huygens Object Analyzer software. Sphericity values for LC3B-positive structures were obtained using the “RoughSphericity” feature. Colocalization between LC3B and TGN46 was quantified using the colocalization coefficients feature, reporting Pearson’s correlation coefficient. In addition, profile intensity plots were generated to visualize the overlap between LC3B and LAMP2 signals.

Western blot analysis

Cells were washed twice with PBS and lysed in RIPA buffer (89901, Thermo Fisher) supplemented with protease inhibitors (11836170001, Sigma) and phosphatase inhibitors (PHOSS-RO, Sigma). Protein concentrations were determined using the BCA reagent (23228, Thermo Fisher). Equal amounts of protein were mixed with Laemmli Sample Buffer (1610747, Bio-Rad) and boiled at 100°C for 10 minutes. Denatured proteins were separated on either 4–20% gradient or 10% SDS-PAGE gels (Bio-Rad) at 95 V and transferred to methanol-activated PVDF membranes or nitrocellulose membranes at 100 V for 1 hour at 4°C. Membranes were blocked with 5% skim milk in PBS and incubated with primary antibodies overnight at 4°C. After three 10-minute washes with PBST (0.1% Tween-20), membranes were incubated with secondary antibodies for 1 hour at room temperature. Following three additional washes with PBST, membranes were developed using ECL substrate (1705061, Bio-Rad) for HRP-conjugated antibodies or directly imaged using a ChemiDoc imager (Bio-Rad) for fluorescent secondary antibodies. Protein quantification was performed using Image Lab software (Bio-Rad).

Halo LC3 assay for autophagy flux

HeLa WT or TRIM46 knockout cells stably expressing Halo-LC3 (HT-LC3B) were seeded at a density of 3 × 10⁵ cells per well in 6-well plates. The following day, cells were pulsed with 2.5μM TMR-labeled HaloTag ligand (HL) for 30 minutes, followed by three washes with DMEM. Cells were then treated with either DMSO or 100 nM Bafilomycin A1 for 4 or 6 hours. After treatment cells were washed twice with PBS and lysed in RIPA buffer supplemented with protease inhibitors. Protein concentrations were determined using a BCA assay, and equal amounts of protein were subjected to SDS-PAGE. Released HT/HL was detected by TMR fluorescence in-gel using a ChemiDoc^™ MP Imager (Bio-Rad) prior to immunoblotting for actin.

DQ-BSA assay

To assess lysosomal function, cells were seeded in a 96-well plate and incubated overnight with 10 μg/ml DQ-BSA (D12050, Thermo Fisher) diluted in culture medium. Following incubation, cells were fixed with paraformaldehyde prior to immunolabeling with anti-LAMP2 and high content imaging.

Membrane fractionation

Membrane fractionation was performed as described previously⁶⁹. Cells from the two 15-cm dishes grown to confluence were harvested and homogenized in 5X cell pellet volume of buffer B1 buffer (20 mM HEPES-KOH, 400 mM sucrose, and 1 mM EDTA) supplemented with protease and phosphatase inhibitors and 0.3 mM DTT by passing through a 22-G needle. Homogenates were subjected to sequential differential centrifugation at 1000 × g for 10 min, 3000 × g for 10 min, 25, 000 × g for 20 min, and 100, 000 × g for 30 minutes to collect the pelleted membranes. Pellets were lysed in RIPA buffer supplemented with protease inhibitors and protein concentrations were determined using a BCA assay. Samples were mixed with Laemmli Sample Buffer (1610747, Bio-Rad), boiled for 10 minutes, and analyzed by SDS-PAGE and immunoblotting with antibodies against LC3B, LAMP2, and TGN46.

Flow cytometric measurement of Golgiphagy

HeLa wild-type or TRIM46 knockout cells (2 × 10⁶) were seeded in 10 cm dishes one day before transfection. Cells were transfected with 14 μg of the pKH116-Keima-YIPF3 plasmid (214970, Addgene) using Lipofectamine 2000 (Thermo Fisher) according to the manufacturer’s protocol. 16 hours post-transfection, the culture medium was replaced with fresh medium. On day 2, cells were starved in Earle’s Balanced Salt Solution (EBSS; 24010043, Thermo Fisher) with or without 100 nM bafilomycin A1 for 6 hours. After treatment, cells were harvested by trypsinization and resuspended in PBS containing 5% FBS for flow cytometry analysis. Acidified Golgi structures were assessed using dual-excitation ratiometric pH measurements at 405 nm (neutral pH ~7) and 561 nm (acidic pH ~4) laser excitations, with emission detected at 603/48 nm and 620/15 nm respectively using the flow cytometry (Attune NxT, Thermofisher). Flow cytometry data were acquired and analyzed using FlowJo software (version 10, Tree Star).

Statistical analysis, graphing and figure assembly

All data are presented as mean ± s.e.m. Statistical significance between two groups was assessed using a two-sided unpaired Student’s t-test for normally distributed data or a two-sided Mann-Whitney U-test for non-normally distributed data. For multiple group comparison, one-way or two-way Anova followed by Tukey’s post hoc test was applied. Statistical analyses and graph generation were conducted using Graphpad Prism v10.6.0. Schematic diagrams were prepared with BioRender, and final figure assembly was performed using Adobe Illustrator.

KEY RESOURCES

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-mCherry antibody	Abcam	ab183628, RRID: AB_2650480
GABARAP	Abcam	ab109364, RRID: AB_10861928
NPC2	Abcam	ab218192, RRID: AB_2941808
Purified Mouse Anti-p62 Ick ligand	BD biosciences	P0067, RRID: AB_398151
Goat Anti-Mouse IgG (H+L)-HRP	BIO-RAD	1721011, RRID: AB_2617113
Goat Anti-Rabbit IgG (H+L)-HRP	BIO-RAD	1721019, RRID: AB_11125143
TGN46	BIO-RAD	AHP500G, RRID: AB_2203291
AMPK	Cell Signaling Technology	2532, RRID: AB_330331
phospho-AMPK T172	Cell Signaling Technology	2535, RRID: AB_331250
Atg13 (E1Y9V) Rabbit mAb	Cell Signaling Technology	13468, RRID: AB_2797419
BECN1	Cell Signaling Technology	3459, RRID: AB_560924
GM130	Cell Signaling Technology	12480, RRID: AB_2797933
phospho-P70 S6K	Cell Signaling Technology	9205, RRID: AB_330944
TFEB	Cell Signaling Technology	4240, RRID: AB_11220225
mTOR	Cell Signaling Technology	2983, RRID: AB_2105622
ULK1	Cell Signaling Technology	8054, RRID: AB_11178668
IRDye® 680LT Goat anti-Mouse IgG Secondary Antibody	LI-COR Biosciences	925-68020, RRID: AB_2687826
IRDye 800CW Goat anti-Mouse IgG Secondary Antibody	LI-COR Biosciences	925-32210, RRID: AB_2687825
LC3B	MBL	PM036, RRID: AB_2274121
Anti-Halotag monoclonal antibody	Promega	G921A, RRID:AB_2688011
Cathepsin D	Proteintech	21327-1-AP, RRID: AB_10733646
LIMP2	Proteintech	27102-1-AP, RRID: AB_2880756
TRIM46	Proteintech	21026-1-AP, RRID: AB_10732843
Cas9	Santa Cruz Biotechnology	sc-517386, RRID: AB_2800509
LAMP2	Santa Cruz Biotechnology	sc-18822, RRID: AB_626858
P70 S6K	Santa Cruz Biotechnology	sc-8418, RRID: AB_628094
Anti-Actin Antibody (2Q1055)	Santa Cruz Biotechnology	sc-58673, RRID: AB_2223345
alpha-tubulin	Santa Cruz Biotechnology	sc-23948, RRID: AB_628410
Rabbit Anti-LC3B	Sigma Aldrich	L7543, RRID: AB_796155
GCP60	Thermofisher	MA5-25999, RRID: AB_2723827
TGN46	Thermofisher	MA5-37930, RRID: AB_2897850
WIPI1	Thermofisher	PA5-34973, RRID: AB_2552322
HCS CellMask™ Near-IR Stain	Thermofisher	H32722
Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 488	ThermoFisher	A-11034, RRID: AB_2576217
Goat anti-Mouse IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 647	ThermoFisher	A32728, RRID: AB_2633277
Bacterial and virus strains
NEB 5-alpha Competent E. coli (High Efficiency)	New England Biolabs	C2987
XL10-Gold Ultracompetent cells	Agilent	210518
Chemicals, peptides, and recombinant protein
10x Tris/Glycine/SDS buffer	Bio-Rad	1610732
2x Laemmli Buffer	Bio-Rad	1610737
4x Laemmli Buffer	Bio-Rad	1610747
Clarity ECL	Bio-Rad	1705061
Glycine	Bio-Rad	1610718
Tris Base	Bio-Rad	1610719
Vinblastine	Cayman chemical	11762
Vps34-IN1	Cayman chemical	17392
Bovine Serum Albumin	Fisher Scientific	CAS 9048-46-8
DTT (Dithiothreitol)	Gold Biotechnology	DTT10
Bafilomycin A1	Invivogen	tlrl-baf1
HaloTag® TMR Ligand	Promega	G8251
Brefeldin A	Sigma Aldrich	B6542
Ethylenediaminetetraacetic acid disodium salt dihydrate	Sigma Aldrich	E5134
2-mercaptoethanol	Sigma Aldrich	M3148
Nocodazole	Sigma Aldrich	M1404
PHOSSTOP	Sigma Aldrich	4906837001
cOmplete™, Mini, EDTA-free Protease Inhibitor Cocktail	Sigma Aldrich	11836170001
Puromycin dihydrochloride	Sigma Aldrich	P9620
Saponin	Sigma Aldrich	84510
Sucrose	Sigma Aldrich	84097
Tween 20	Sigma Aldrich	P1379
Millipore® Steriflip® Vacuum Tube Top Filter	Sigma Aldrich	SE1M003M00
Dulbecco’s modified Eagle’s medium	ThermoFisher	11965084
Earle’s Balanced Salt Solution	ThermoFisher	24010043
fetal bovine serum	ThermoFisher	26140095
Hoechst 33342	ThermoFisher	H3570
Lipofectamine 2000 Reagent	ThermoFisher	11668019
Opti-MEM Reduced Serum Medium	ThermoFisher	31985070
penicillin-streptomycin	ThermoFisher	15140122
Restore Plus Western Blot Stripping Buffer	ThermoFisher	46430
RIPA lysis buffer	ThermoFisher	89901
Trypsin-EDTA (0.25%), phenol red	ThermoFisher	25200072
Ampicillin Sodium Salt	VWR	IC19014805
Dimethyl Sulfoxide (DMSO)	VWR	EMMX14586
Ethanol	VWR	89125172
Kanamycin Sulfate	VWR	97061-600
Methanol	VWR	BDH20291GLP
Paraformaldehyde	VWR	JTS8987
Potassium Chloride	VWR	EMPX14051
Potassium Phosphate Monobasic	VWR	EMDPX15651
2-Propanol (Isopropyl Alcohol)	VWR	BDH20271GLP
Sodium Chloride	VWR	BDH928625KG
Sodium Phosphate Dibasic	VWR	97061472
Triton X-100	VWR	EM9410
Critical commercial assays
Agilent QuikChange Lightning Site-Directed Mutagenesis Kit	Agilent	210518
BCA reagent	ThermoFisher	23228
Extracta Plus DNA	QuantaBio	95213-050
Gateway™ BP Clonase™ II Enzyme mix	ThermoFisher	11789100
Gateway™ LR Clonase™ II Enzyme mix	ThermoFisher	11791020
Guide-it™ Complete sgRNA Screening System	Takara Bio	632636
QIAprep Spin Miniprep Kit	Qiagen	27104
QIAquick PCR Purification Kit	Qiagen	28106
PureLink™ HiPure Plasmid Midiprep Kit	Thermo Fisher	K210005
ProFection Mammalian Transfection System	Promega	E1200
Experimental models: Cell lines
HeLa TRIM46 knockout	This study
HEK293T TRIM46 knockout	This study
HeLa stably expressing HT-LC3	This study
HeLa TRIM46 knockout cells stably expressing HT-LC3	This study
Oligonucleotides
TRIM46 Forward primer for sgRNA	This study	5’-CCTCTAATACGACTCACTATAGGTACCGTTGAGTTCCGGCGCAGTTTAAGAGCTATGC-3’
TRIM46 Forward primer for genomic DNA amplification	This study	5’-GCTGCTTTCCCTTTTCCT-3’
TRIM46 Reverse primer for genomic DNA amplification	This study	5’-CTCTGAAGTTCAGAGAGGGT-3’
TRIM46 Forward primer for site-directed mutagenesis g1407t	This study	5’-AGTTCCGGCGCACTGATGTGCCTGCTC-3’
TRIM46 Reverse primer for site-directed mutagenesis g1407t	This study	5’-GAGCAGGCACATCAGTGCGCCGGAACT-3’
siControl	GE Dharmacon	D-001206-13-05
siATG7	GE Dharmacon	M-020112-01-005
siATG13	GE Dharmacon	M-020765-01-0005
siBECN1	GE Dharmacon	M-010552-01-0005
siULK1	GE Dharmacon	M-005049-00-0005
siTRIM1/MID2	GE Dharmacon	M-007076-01
siTRIM2	GE Dharmacon	M-006955-00
siTRIM3	GE Dharmacon	M-006931-00
siTRIM4	GE Dharmacon	M-007101-00
siTRIM5	GE Dharmacon	M-007100-00
siTRIM6	GE Dharmacon	M-007121-01
siTRIM7	GE Dharmacon	M-007077-01
siTRIM10	GE Dharmacon	M-006920-01
siTRIM11	GE Dharmacon	M-007075-00
siTRIM13	GE Dharmacon	M-006923-00
siTRIM14	GE Dharmacon	M-010976-00
siTRIM15	GE Dharmacon	M-007102-01
siTRIM16	GE Dharmacon	M-012220-01
siTRIM16L	GE Dharmacon	M-023055-01
siTRIM17	GE Dharmacon	M-006981-01
siTRIM18/MID1	GE Dharmacon	M-006537-01
siTRIM19/PML	GE Dharmacon	M-006547-01
siTRIM20/MEFV	GE Dharmacon	M-011081-00
siTRIM21	GE Dharmacon	M-006563-02
siTRIM22	GE Dharmacon	M-006927-03
siTRIM23	GE Dharmacon	M-006523-00
siTRIM24	GE Dharmacon	M-005387-03
siTRIM25	GE Dharmacon	M-006585-00
siTRIM26/ZNF174	GE Dharmacon	M-019558-02
siTRIM27	GE Dharmacon	M-006552-01
siTRIM28	GE Dharmacon	M-005046-01
siTRIM29	GE Dharmacon	M-012409-01
siTRIM31	GE Dharmacon	M-006939-01
siTRIM32	GE Dharmacon	M-006950-01
siTRIM33	GE Dharmacon	M-005392-03
siTRIM34	GE Dharmacon	M-006997-01
siTRIM35	GE Dharmacon	M-006952-02
siTRIM37	GE Dharmacon	M-006538-02
siTRIM38	GE Dharmacon	M-006929-01
siTRIM39	GE Dharmacon	M-007028-01
siTRIM40	GE Dharmacon	M-007129-01
siTRIM41	GE Dharmacon	M-007105-02
siTRIM42	GE Dharmacon	M-007173-00
siTRIM43	GE Dharmacon	M-007127-01
siTRIM44	GE Dharmacon	M-017337-01
siTRIM45	GE Dharmacon	M-007073-01
siTRIM46	GE Dharmacon	M-007071-01
siTRIM47	GE Dharmacon	M-007106-02
siTRIM48	GE Dharmacon	M-007059-01
siTRIM49	GE Dharmacon	M-007030-01
siTRIM50	GE Dharmacon	M-007130-00
siTRIM51	GE Dharmacon	M-010079-02
siTRIM52	GE Dharmacon	M-007095-00
siTRIM54	GE Dharmacon	M-007032-01
siTRIM55	GE Dharmacon	M-007092-01
siTRIM56	GE Dharmacon	M-007079-00
siTRIM58	GE Dharmacon	M-013985-02
siTRIM59	GE Dharmacon	M-007172-01
siTRIM60	GE Dharmacon	M-007153-00
siTRIM61	GE Dharmacon	M-028281-01
siTRIM62	GE Dharmacon	M-007010-02
siTRIM63	GE Dharmacon	M-007093-01
siTRIM64	GE Dharmacon	M-026740-04
siTRIM65	GE Dharmacon	M-018490-01
siTRIM66	GE Dharmacon	M-026772-01
siTRIM67	GE Dharmacon	M-032288-01
siTRIM68	GE Dharmacon	M-007007-01
siTRIM71	GE Dharmacon	M-023459-01
siTRIM72	GE Dharmacon	M-032293-02
siTRIM73	GE Dharmacon	M-028896-01
siTRIM74	GE Dharmacon	M-031736-01
siTRIM76/CMYA5	GE Dharmacon	M-016373-01
Recombinant DNA
mCherry	Pankiv et al, 200770
mCherry-ATG16L1	Kumar et al, 202171
pDONR221-hTRIM46	DNASU	HsCD00862242
mCherry-TRIM46	This study	N/A
Myc-TRIM46	This study	N/A
mKeima-YIPF3	Addgene	214970
pLV[CRISPR]-hCas9:T2A:Puro-U6-hTRIM46	VectorBuilder	VB900138-6978huh
pLEX307-Halo LC3	Javed et al, 202572
Software and algorithms
Prism 8	GraphPad	N/A
Image Lab	BIO-RAD	N/A
FlowJo (v10.10.0)	BD Biosciences	N/A
iDEV software	ThermoFisher	N/A
Huygens Object Analyzer and Colocalization	Scientific Volume Imaging	N/A
LASX acquisition software	Leica	N/A
BioRender	BioRender.com	N/A
ICE software	Synthego

Abbreviations:

AMPK

AMP-activated protein kinase

ATG3

autophagy related 3

ATG5

autophagy related 5

ATG7

autophagy related 7

ATG12

autophagy related 12

ATG13

autophagy related 13

ATG16L1

autophagy related 16 like 1

BECN1

Beclin 1

CASM

conjugation of Atg8 to single membranes

GABARAP

GABA type A receptor-associate protein

GABARAP L1

GABA type A receptor-associate protein like 1

GABARAP L2

GABA type A receptor-associate protein like 2

HT

HaloTag

HL

HaloTag ligand

hVPS34/PIK3C3

phosphatidylinositol 3-kinase catalytic subunit type 3

LC3A

microtubule associated protein 1 light chain 3 alpha

LC3B

microtubule associated protein 1 light chain 3 beta

LC3C

microtubule associated protein 1 light chain 3 gamma

mTORC1

mechanistic target of rapamycin complex 1

PE

phosphatidyl ethanolamine

PS

phosphatidyl serine

TECPR1

tectonin beta-propellor repeat containing 1

TFEB

transcription factor EB

TFE3

transcription factor binding to IGHM enhancer 3

TGN46/TGOLN2

trans-golgi network protein 2

TRIM46

tripartite motif containing 46

ULK1

unc-51 like autophagy activating kinase 1

VAIL

V-ATPase-ATG16L1induced LC3 lipidation