Autophagy initiation triggers p150Glued-AP-2β interaction on the lysosomes to facilitate their transport

The endocytic adaptor protein 2 (AP-2) complex binds dynactin as part of its noncanonical function, which is necessary for dynein-driven autophagosome transport along microtubules in neuronal axons. The absence of this AP-2-dependent transport causes neuronal morphology simplification and neurodegeneration. The mechanisms that lead to formation of the AP-2–dynactin complex have not been studied to date. However, the inhibition of mammalian/mechanistic target of rapamycin complex 1 (mTORC1) enhances the transport of newly formed autophagosomes by influencing the biogenesis and protein interactions of Rab-interacting lysosomal protein (RILP), another dynein cargo adaptor. We tested effects of mTORC1 inhibition on interactions between the AP-2 and dynactin complexes, with a focus on their two essential subunits, AP-2β and p150Glued. We found that the mTORC1 inhibitor rapamycin enhanced p150Glued–AP-2β complex formation in both neurons and non-neuronal cells. Additional analysis revealed that the p150Glued–AP-2β interaction was indirect and required integrity of the dynactin complex. In non-neuronal cells rapamycin-driven enhancement of the p150Glued–AP-2β interaction also required the presence of cytoplasmic linker protein 170 (CLIP-170), the activation of autophagy, and an undisturbed endolysosomal system. The rapamycin-dependent p150Glued–AP-2β interaction occurred on lysosomal-associated membrane protein 1 (Lamp-1)-positive organelles but without the need for autolysosome formation. Rapamycin treatment also increased the acidification and number of acidic organelles and increased speed of the long-distance retrograde movement of Lamp-1-positive organelles. Altogether, our results indicate that autophagy regulates the p150Glued–AP-2β interaction, possibly to coordinate sufficient motor-adaptor complex availability for effective lysosome transport.


Introduction
The effective cooperation of endomembrane components and the cytoskeleton is necessary for efficient intracellular communication and cell contacts with the extracellular environment. Microtubules are cytoskeleton elements that are essential for both the integrity of membrane compartments and their long-distance movement [1,2]. Microtubules are dynamic and polarized, meaning that their ends (referred to as plus and minus) can undergo dynamic changes and are not identical [3]. This polarization determines the rules of directed cargo transport along microtubules by molecular motors, e.g., kinesins and dynein [4,5]. The latter transports cellular cargo from the plus end to the minus end of microtubules [5][6][7]. Dynein does not act alone; it requires additional protein complexes to efficiently hold cargo and move along microtubules. One of these complexes is dynactin, a large multiprotein complex that initiates dynein movement, increases its processivity, and supports cargo attachment [5,8,9]. Dynactin consists of two major parts: sidearm and actin-related protein 1 (Arp-1) rod [5,7,10,11]. The sidearm binds microtubules and dynein [5,7,10,11]. The Arp-1 rod, in cooperation with dynein activators and adaptors, is responsible for cargo binding [5,7,[10][11][12][13][14][15]. p150 Glued is part of the sidearm, the largest dynactin subunit, and a member of the microtubule plus-end tracking protein (+TIP) family [16]. Its binding to microtubule plus ends and its plus-end tracking behavior require the presence of cytoplasmic linker protein 170   [17,18]. In some model systems (e.g., neuronal axons), this is essential for the initiation of dynein-dynactinbound cargo transport along tyrosinated microtubules [8,19].
The adaptor protein 2 (AP-2) complex consists of two large subunits (α, β), one medium subunit (µ), and one small subunit (σ) [20]. All four subunits contribute to the trunk of the AP-2 complex, but α and β2 C-termini project outside the trunk as α and β2 appendages (i.e., ears), respectively [20]. Canonically, AP-2 serves as a cargo adaptor complex in clathrin-mediated endocytosis [21]. However, evidence supports AP-2 functions outside the initiation of clathrin-mediated endocytosis, particularly in macroautophagy (hereafter called autophagy), lysosome tubulation, and microtubular transport [22][23][24][25][26][27]. The latter function was first discovered in neurons, in which AP-2 was found to be central to the retrograde transport of neuronal amphisomes that are produced by autophagosome-late endosome fusion and in axons act as signaling organelles that carry activated receptors for neurotrophins, such as tropomyosin receptor kinase B (TrkB) [24,28] to the cell soma. The lack of this AP-2-dependent transport in axons resulted in disturbances in the morphology of dendrites and neurodegeneration [22,24]. For dynein cargo adaptor function, AP-2 binds microtubule-associated protein 1A/1B-light chain 3 (LC3) on the amphisome surface via its AP-2µ subunit, whereas the AP-2β ear was shown to co-immunoprecipitate with p150 Glued [24].
The effective termination of autophagy requires the fusion of autophagosomes or amphisomes with lysosomes, which contain degradative enzymes that are needed for autophagosome cargo destruction. It heavily relies on autophagosome and lysosome transport along microtubules. Dynein-dynactin transports autophagosomes retrogradely for fusion with lysosomes [45][46][47]. To meet autophagosomes, lysosomes may use both kinesins and dyneindynactin [48,49]. Lysosomes are dispersed through the cytoplasm with two distinguishable pools: perinuclear and peripheral [49]. The peripheral pool serves additional purposes (e.g., exocytosis [50]); when the demand for lysosomes greatly increases, however, such as during nutrient starvation that initiates autophagy, they move via dynein-dynactin transport toward autophagosomes that are already positioned in the cell center. To date, only two adaptors (ALG2 and JIP4) have been shown to recruit dynein-dynactin to lysosomes on demand upon nutrient starvation [51,52].
Although AP-2-dynactin was shown to transport TrkB-positive amphisomes in neurons, unclear is whether the AP-2-dynactin complex also forms naturally in non-neuronal cells, in which amphisomes are considered very transient. Further details of the AP-2-dynactin interaction and its potential regulation are lacking. Our unpublished preliminary mass spectrometry data suggested a potential role for mTOR in the regulation of the p150 Glued -AP-2β interaction. This is particularly intriguing when considering the important role of kinases in the regulation of microtubular transport [39,53] and a recent finding that dynein can be recruited to autophagosomes by LC3 and Rab-interacting lysosomal protein (RILP) when mTORC1 activity is low [39]. Therefore, the present study investigated whether mTORC1 controls the p150 Glued -AP-2 interaction and, if so, how and for what purpose. We found that mTORC1 inhibition enhanced the p150 Glued -AP-2β interaction in both neurons and nonneuronal cells. We also found that p150 Glued -AP-2β complex formation, boosted by mTORC1 inhibition, in non-neuronal cells required an intact dynactin complex and the undisturbed initiation of autophagy and endolysosomal pathway. We also found that the autophagy-induced p150 Glued -AP-2β interaction occurred on lysosomes, which accelerated their retrograde motility. Thus, we revealed a novel mechanism whereby functions of essential components of cellular transport machinery are regulated at the level of autophagy initiation.

Cell line and primary neuronal cultures and transfection
Rat2, HEK293, and primary hippocampal neurons cultures and transfections were performed using standard previously published protocols. For details, please refer to Supplementary Information.

Drugs and drug treatment
All drugs, unless indicated otherwise, were dissolved in dimethylsulfoxide (DMSO), the final concentrations of which in the culture medium did not exceed 0.1%. For mTOR inhibition, Rat2 and HEK293T cells were treated with rapamycin (100 nM, Calbiochem, catalog no. 553210) or AZD-8055 (100 nM, Cayman Chemical, catalog no. 16978-5) for 2 h before the experiment (see figure legends for detailed descriptions). For translation inhibition, cells were treated for 2 h with 35 µM cycloheximide (Calbiochem, catalog no. 239763). For mTOR inhibition-dependent autophagy arrest at the initiation step, 25 µM SBI-0206965 (Merck, catalog no. SML1540) was used for 2.5 h (see figure legends for detailed descriptions).
When combined with rapamycin, SBI-0206965 was added 30 min before the addition of rapamycin. To inhibit autophagic flux, cells were treated for 2 h with 60 µM 1-adamantyl(5bromo-2-methoxybenzyl)amine (ABMA; Medchemexpress, catalog no. HY-124801) or 50 µM chloroquine (dissolved in water; Sigma-Aldrich, catalog no. C6628). For lysosomal vacuolar (H+)-adenosine triphosphatase (vATPase) inhibition, cells were treated for 2 h with bafilomycin A1 (Baf A1; 100 nM; Bioaustralis,. When cells were treated with an inhibitor and rapamycin, these compounds were administered at the same time and incubated for 2 h. For the alkalization of the cellular environment, 20 mM NH4Cl (dissolved in water; Sigma-Aldrich, catalog no. 213330) was added for 3 h. When cells were treated with both NH4Cl and rapamycin, NH4Cl was added to the cells 1 h prior to rapamycin, and the cells were then incubated with both drugs for an additional 2 h. To induce autophagy independently from mTORC1 inhibition, 100 μM L-690330 (Tocris, catalog no. 0681) was added to the cells for 3 h. Nocodazole (100 nM; Sigma-Aldrich, catalog no. M1404) was used to inhibit microtubule dynamics. For live-cell imaging experiments, the drug was added to Rat2 cells 1 h before imaging. For the PLA experiments, in which nocodazole was added to Rat2 cells alone or combined with rapamycin, it was added 2 h 15 min or 1 h before fixation, depending on whether it was used before or after rapamycin treatment. Before live imaging or fixation, neurons were treated with either vehicle (0.1% DMSO, 2 h) or rapamycin (100 nM, 2 h).

Animals and rapamycin treatment
Rapamycin treatment and brain protein lysate isolation were performed according to a protocol that was approved by the 1st Ethical Committee in Warsaw (Poland;decision no. 843/2008 and288/2012). Mature (3-month-old) male Wistar rats were used for the experiments.
Rapamycin was initially dissolved in 100% ethanol at a 0.1 mg/ml concentration and stored at -20C. Immediately before the injection, rapamycin was diluted in a vehicle solution that contained 5% Tween 80 and 5% PEG 400 (low-molecular-weight grade of polyethylene glycol; Sigma) and injected intraperitoneally (i.p.; 10 mg/kg) three times per week for 1 week. A control group of rats was injected with a vehicle solution that contained 5% Tween 80, 5% PEG 400, and 4% ethanol. Protein extraction from adult rat brains was performed as described previously [24] with a slight modification that involved the addition of 100 nM rapamycin to the lysis buffer for brains of rapamycin-treated animals.

Proximity ligation assay and PLA-EM
Standard PLA procedures were performed as described previously [58] and detailed procedure is described in Supplementary Information. For PLA-EM, Rat2 cells were grown for 24 h and fixed for 15 min in 4% PFA with the addition of 0.1% glutaraldehyde in PBS. The cells were then washed three times with PBS. Afterward, cells in PBS were permeabilized by three cycles of freezing in liquid nitrogen, thawed, incubated for 20 min with 1% sodium borohydride in PBS, washed three times with PBS, incubated for 20 min at room temperature with 3% hydrogen peroxide in PBS/ethanol (1:1), and washed again three times in PBS. Next, fixed and permeabilized cells were incubated for 1 h in 5% bovine serum albumin (BSA) in PBS at room temperature. Afterward, the cells were incubated for 48 h at 4°C with primary mouse anti-p150 Glued and rabbit anti-AP-2β antibodies that were diluted in 0.1% donkey serum/PBS and washed three times in PBS at room temperature. The cells were then incubated for 60 min at 37°C with PLA probes (Sigma-Aldrich, catalog no. DUO92002 and DUO92004) and washed twice for 5 min with buffer A (Sigma-Aldrich, catalog no. DUO82046). Ligation and amplification were performed according to the manufacturer's protocol using DuoLink In Situ Detection Reagents Brightfield (Sigma-Aldrich, catalog no. DUO92012). After the PLA reaction, the cells were additionally fixed with 2.5% glutaraldehyde in PBS for 2 h at 4°C and washed twice with PBS and once with deionized water. Next, the cells were incubated with 3% hexamethylenetetramine, 5% silver nitrate, and 2.5% disodium tetraborate for 10 min at 60°C, washed three times with water, once with 0.05% tetrachloroauric acid, once with 2.5% sodium thiosulfate, and finally three times with water (all at room temperature). As the last step, the cells were postfixed with 1% osmium tetroxide for 1 h at room temperature, washed with water, incubated in 1% aqueous uranyl acetate for 1 h, dehydrated with increasing dilutions of ethanol, and infiltrated with epoxy resin (Sigma-Aldrich, catalog no. 45-359-1EA-F). After resin polymerization at 60°C, fragments of coverslips with embedded cells were cut out with scissors and glued to the resin blocks. The blocks were then trimmed and cut with a Leica ultramicrotome (Ultracut R) to obtain ultrathin sections (70 nm thick) and collected on 100 mesh copper grids (Agar Scientific, catalog no. AGS138-1). Specimen grids were examined with a Tecnai T12 BioTwin transmission electron microscope (FEI) that was equipped with a 16 megapixel TemCam-F416 (R) camera (Tietz Video and Imaging Processing Systems).

Immunofluorescence and fixed cell image acquisition and analysis
Procedures used for immunofluorescence and fixed cell image acquisition and analysis are described in detail in Supplementary Information.

Live imaging of microtubule dynamics
For microtubule dynamics analysis, Rat2 cells were electroporated with pEGFP-EB3 or pEGFP-CLIP-170 plasmids. Forty-eight hours later, time-lapse movies of EGFP-EB3 or EGFP-CLIP-170 comets were taken using an Andor Revolutions XD spinning disc microscope with a 63 objective and 1.6 x tube lens (optovar) at 1004  1002 pixel resolution. Images were taken with an exposure of 200 ms and interval of 0.3 s, collecting a total of 600 frames over 3 min.
During imaging, cells were kept in a Chamlide magnetic chamber (Quorum Technologies) at 37°C with 5% CO2 in the incubator that was part of the microscope system. Only EGFP-EB3 or EGFP-CLIP-170 comets whose movements lasted at least four consecutive frames and had a displacement length of at least 10 pixels (0.68 µm) were analyzed using the ImageJ "TrackMate" plugin [66]. The reported values are the number of tracks (i.e., the quantification of objects that were detected as comets), the total run length of comets before catastrophe (Track Displacement), comet lifetime (Track Duration), and Track Mean Speed. For the analysis, values were calculated as means for each cell.

Live imaging of Lamp-1-GFP objects
For the lysosomal-associated membrane protein 1 (Lamp-1) object motility analysis, Rat2 cells were electroporated with Lamp-1-GFP plasmid and imaged approximately 22 h later with an Andor Revolutions XD spinning disc microscope, with the same setup and settings as described above. Time-lapse movies were collected over 3 min at 0.3 s intervals, resulting in 600 frames. Movies were analyzed using the ImageJ "TrackMate" plugin [66]. Only objects that were visible for more than 4 consecutive frames were considered. Objects with movement lengths shorter than 6.8 µm were considered not motile. The number of motile and non-motile objects divided by cell area and their ratio were measured. Other calculated values that were used for the analysis included the following: length of the Lamp-1 object run (Track Displacement), time during which the objects were visible (Track Duration), and speed with which the object moved (Track Mean Speed). For the analysis of Lamp-1-GFP objects in the cell center and in the periphery, Lamp-1-GFP objects were assigned manually to these compartments based on maximum projections.

Live imaging of neurons
An Andor Revolutions XD confocal spinning disc microscope and a Chamlide magnetic chamber were used for the in vivo imaging of cells. Cell recording was performed at 37°C with 5% CO2 in a thermostat-controlled incubator. Series of images were acquired at 502 x 501 pixel resolution, with 1 s interval between frames. The total imaging time was 3 min. The images were collected using the 63x objective and 1x optovar. Neuronal processes were manually tracked with ImageJ. Based on these tracks, kymographs were created in ImageJ with the "Kymograph Clear" plugin [67]. By tracing lines on the kymographs, the number of particles that moved at a given time and distance was determined, and their velocity was calculated by determining the difference between the height of the starting point and end point of a given particle.

Biochemical procedures
All standard biochemical procedures including protein production, pull-down assays, immunoprecipitation, Western blot, kinase assays and RNA isolation and Quantitative Real-Time PCR are described in details in Supplementary Information.

Statistical analysis
The exact numbers of cells (n) that were examined for the respective experiments and number of repetitions of each experiment (N) are provided in the figure legends. The statistical analyses were performed using GraphPad Prism 9 software or Rstudio. The Shapiro-Wilk test was used to assess whether the data distribution met the assumptions of a normal distribution.
For comparisons between two groups, the t-test (in the case of a normal distribution) or Mann-Whitney test (in the case of a non-normal distribution) was used to verify statistical significance. For comparisons between more than two groups, the data were analyzed using one-way analysis of variance (ANOVA), followed by the Bonferroni multiple-comparison post hoc test (in the case of a normal distribution) or Kruskal-Wallis test and Dunn's multiplecomparison post hoc test (in the case of a non-normal distribution). For comparisons between two factors, the data were analyzed using two-way ANOVA, followed by Tukey's multiplecomparison post hoc test. The statistical significance of qRT-PCR data was assessed with the one-sample Student's t-test.

mTORC1 inhibition increases p150 Glued -AP-2β interaction in neurons and non-neuronal cells
Recent work shows that mTOR inhibition increases the biosynthesis of RILP and enhances its recruitment to autophagosomes, potentiating their transport in different cell types, including neurons [39]. This finding prompted us to investigate whether rapamycin, an inhibitor of mTORC1, also affects the AP-2-dynactin interaction that is responsible for the axonal transport of amphisomes [24]. We performed the IP of endogenous AP-2β from brain lysates from control rats and rats that were treated with rapamycin for 8 days. Rapamycin treatment effectively decreased the phosphorylation of ribosomal protein S6 (P-S6) at Ser235/236, confirming efficient mTORC1 inhibition, and increased the co-IP of p150 Glued with AP-2β. We observed no noticeable difference in overall levels of p150 Glued or AP-2β in corresponding input fractions (Fig. 1A). mTORC1 inhibition potentiates the AP-2-dynactin interaction in the brain. Therefore, we next tested effects of rapamycin on the AP-2β and p150 Glued interaction in axons of hippocampal neurons, in which its functional significance was demonstrated [24]. We transfected neurons that were cultured in vitro (DIV5) with plasmids that encoded tdTomato-tagged p150 Glued and green fluorescent protein (GFP)-tagged AP-2β. Two days later, we imaged the behavior of fluorescently tagged proteins in axons in DMSO-treated (control) Supplementary Information]). Similar to brain tissues, rapamycin decreased P-S6 levels ( Fig. S1B, C). We also observed a significant increase in the number of p150 Glued -AP-2β-positive objects (Fig. 1D), suggesting that mTORC1 inhibition boosted the p150 Glued -AP-2β interaction in neurons, but these objects were largely immobile. The difference between fractions of mobile AP-2β/p150 Glued -positive objects was also statistically nonsignificant under the tested conditions (Fig. 1E). Thus, we concluded that mTORC1 inhibition in neurons potentiated the p150 Glued -AP-2β interaction in axons.
We next investigated whether mTORC1 inhibition has a similar effect on the co-occurrence of p150 Glued and AP-2β in non-neuronal cells using HEK293T and Rat2 cell lines. Under basal culture conditions, there was some evidence of an AP-2-dynactin interaction, demonstrated by IP (HEK293T), immunofluorescence co-localization (Rat2), and the PLA (Rat2; Fig. 1F-J; Fig.   S2). The treatment of HEK293T cells with 100 nM rapamycin for 2 h decreased P-S6 levels and increased AP-2β-p150 Glued (Fig. 1F). In Rat2 cells, rapamycin decreased S6 phosphorylation (Fig. S2A) and enhanced the p150 Glued -AP-2β interaction, measured by immunofluorescence signal co-localization analysis and PLA using antibodies against endogenous proteins (Fig. 1G, H, Fig. S2B-G). Co-localization analysis showed relatively low co-localization under basal conditions and an increase in rapamycin-treated cells ( Fig. S2B-F).
The best-known function of mTORC1 is the positive regulation of protein synthesis. Therefore, we treated Rat2 cells for 2 h with cycloheximide (CHX) (35 µM), a widely used protein synthesis inhibitor, to test whether the increase in the PLA signal was attributable to a decrease in translation. However, we did not detect a significant difference in the PLA signal in CHX-treated cells compared with control cells (Fig. S3). Overall, our results suggest that mTORC1 inhibition enhances the interaction between p150 Glued and AP-2β in different cell types but not because of the canonical function of mTORC1 as a translational enhancer.
Excluding the possibility that protein synthesis inhibition was a main driver of an increase in the p150 Glued -AP-2β interaction prompted us to test whether p150 Glued or AP-2β are substrates of mTOR. In vitro kinase assays, using GFP-AP-2β or GFP-p150 Glued and the mTOR active fragment, excluded such a possibility (Fig. S4A). Furthermore, inspection of the available datasets of mTOR-dependent phosphoproteomes ([e.g., 68, 69]) did not support the mTOR-dependent phosphorylation of other dynactin or AP-2 subunits. mTOR inhibition in HEK293T cells also did not affect the connection between the dynactin sidearm and its Arp1 rod, indicated by the lack of differences in the co-IP of p150 Glued with p62 and Arp1 between analyzed conditions (Fig. S4B). Thus, we concluded that the observed effects of mTORC1 inhibition on the p150 Glued -AP-2β interaction were not driven by direct actions of mTOR on the dynactin complex or AP-2β.

p150 Glued interaction with AP-2β is indirect and requires dynactin integrity
Our previous study [24] and the data described above show that AP-2 and p150 Glued can form a complex, but further characterization is needed. Therefore, in the following experiments, we first focused on the biochemical characterization of this interaction. Using an Avi-tag pulldown assay, we previously demonstrated that full-length p150 Glued that is produced in HEK293T cells can effectively bind the E. coli-produced β2 ear of AP-2β [24]. Therefore, we used this system to characterize the p150 Glued -β2-ear interaction further and clarified which p150 Glued domains are required. We first compared the ability of the N-terminal (1-490 aa; N) and C-terminal (490-end; C) parts of p150 Glued and the full-length protein ( Fig. 2A) that is produced in HEK293T cells to bind the His-tagged β2 ear. The C-terminal part of p150 Glued was as effective as the full-length protein, whereas its N-terminus did not bind the AP-2 fragment, exactly like β-galactosidase that served as the negative control (Fig. 2B). Moreover, a shorter fragment of the C-terminus (1049-end; C2; Fig. 2A), which is known for its contribution to the dynactin interaction with cargo adaptors [13,70,71], also bound the β2 ear (Fig. 2C). Notably, however, the newest structural and biochemical data raise the issue of whether the C-terminus of p150 Glued binds cargo adaptors directly [10,11]. Indeed, when both C-terminal fragments of p150 Glued were produced in E. coli, no interaction with the AP-2β fragment was observed (Fig.  2D, E). But, the β2 ear interacted with Eps15 protein (541-790 aa fragment fused to GST; [63,72]), which was used as a positive control. The most C-terminal part of p150 Glued is known for its role in connecting the dynactin sidearm with the Arp-1 rod that binds cargo [73]. Indeed, Western blot indicated that during the p150 Glued Avi-tag pull-down, regardless of the harsh washing conditions, the dynactin Arp1-rod proteins (Arp1 and p62) also co-purified from HEK293T cells (Fig. 2F). This suggests that intact dynactin might be involved in formation of the multi-protein complex that contains p150 Glued and AP-2β. Indeed, overexpression of the p50 subunit of dynactin, which is routinely used to disrupt dynactin complex integrity by dissociation of the sidearm and Arp1 rod [74,75]; Fig. 2G), completely blocked the rapamycindriven increase in the p150 Glued -AP-2β PLA signal in Rat2 cells (Fig. 2H, I). Thus, although p150 Glued and AP-2β are unlikely to bind each other directly, their interaction (evident upon mTORC1 inhibition) requires the C-terminus of p150 Glued and an undisturbed interaction between the dynactin sidearm and Arp-1 rod.

CLIP-170 is needed for p150 Glued -AP-2β interaction
p150 Glued is a +TIP. Therefore, we tested whether the p150 Glued -AP-2β interaction requires its ability to target dynamic microtubules. p150 Glued microtubule plus-end targeting requires the presence of CLIP-170, which is also an mTOR substrate [17,18,58]. Thus, we tested whether CLIP-170 knockdown impacts the p150 Glued -AP-2β interaction. We knocked down CLIP-170 in Rat2 and HEK293T cells using rat-and human-specific siRNAs, respectively, and performed the PLA and IP 72 h later. Both siRNAs against CLIP-170 effectively reduced CLIP-170 levels compared with cells that were transfected with control siRNAs (Fig. 3A-D). The loss of CLIP-170 also potently prevented the rapamycin-induced p150 Glued -AP-2β interaction (Fig. 3A, E, F). Notably, CLIP-170 knockdown did not affect the overall distribution of AP-2β in either control or rapamycin-treated cells (Fig. S5A).
CLIP-170 knockdown may result, in addition to p150 Glued displacement, in substantial changes in microtubule dynamics, especially in cells that do not express .
Although Rat2 cells express both CLIPs and our siRNAs did not target CLIP-115 (Fig. 3B), we directly tested effects of CLIP-170 knockdown on microtubule dynamics and the role of microtubule dynamics in the AP-2-dynactin interaction. Indeed, CLIP-170 knockdown had no effect on EB3-GFP mobility that highlights dynamic plus ends of microtubules ( Fig. S5B-F, Movies 5-6). Next, we performed a p150 Glued -AP-2β PLA in Rat2 cells that were treated with 100 nM nocodazole that was added either 15 min before or 1 h after rapamycin treatment. At such low concentrations, nocodazole blocks plus-end microtubule dynamics instead of depolymerizing microtubules [59]. Indeed, 1 h of the nocodazole treatment of Rat2 cells resulted in the loss of EB3-GFP and CLIP-170 comets, confirming the inhibition of microtubule dynamics (Fig. S6, Movies 7-10). Such treatment did not affect the p150 Glued -AP-2β PLA signal under basal conditions or in response to rapamycin treatment ( Fig. 3G-J). In contrast to nocodazole, rapamycin treatment did not affect EB3-GFP microtubule plus-end tracking behavior ( Fig. S7A-D, Movie 11, 12). Notably, rapamycin did not change the plus-end tracking behavior of CLIP-170-GFP that was overexpressed in Rat2 cells ( Fig. S7E-H, Movie 13, 14).
This observation is consistent with our previous data on the lack of effect of rapamycin on endogenous CLIP-170 microtubule binding in neurons and HeLa cells [58].
Overall, the data suggest that microtubule dynamics, at least in the short term, is not needed for the p150 Glued -AP-2β interaction. Low-dose nocodazole treatment should also result in p150 Glued displacement from dynamic microtubule plus ends, like CLIP-170, but we did not observe any impact of nocodazole on p150 Glued -AP-2β complex formation. Thus, p150 Glued displacement from microtubules unlikely explains the effects of CLIP-170 knockdown on the p150 Glued -AP-2β PLA interaction. We confirmed this hypothesis using a dominant-negative CLIP-170 mutant that lacked the N-terminal part of the protein and was previously shown to severely affect microtubule dynamics and displace p150 Glued from microtubule plus ends [60].
The 48 h overexpression of this protein in Rat2 cells did not prevent p150 Glued -AP-2β interaction measured with PLA in the rapamycin-treated cells (Fig. S8). In summary, CLIP-170 is needed for the p150 Glued -AP-2β interaction, but two of its canonical functions (i.e., the regulation of microtubule dynamics and targeting p150 Glued to microtubule plus ends) do not appear to be directly or immediately involved.

mTORC1-dependent autophagy triggers p150 Glued -AP-2β interaction
Like mTOR, AP-2 was postulated to be essential for autophagy initiation [25]. Thus, we investigated whether mTORC1-controlled autophagy under the conditions that were used in the present study is induced and required for the rapamycin-driven p150 Glued -AP-2β interaction.
Treatment with rapamycin decreased P-S6 level, demonstrating mTORC1 inhibition and resulting in an increase in beclin-1 phosphorylation at Ser30, which is an Ulk-1 target and considered an early marker of autophagy (Fig. S9A). Furthermore, after 2 h, rapamycin also increased the ratio of the lipidated form of LC3 (LC3B II) to non-lipidated LC3B I, which is routinely used to assess autophagy (Fig. S9A, B). Furthermore, rapamycin also increased the formation of large LC3B foci, further confirming autophagy induction (Fig. S9C). To verify whether autophagy induction is required for the mTORC1 inhibition-driven AP-2-dynactin interaction, we investigated whether pretreatment with the autophagy initiation inhibitor SBI-0206965 (25 µM; 30 min before rapamycin administration) counteracts the effects of rapamycin. As expected, pretreatment with SBI-0206965 was sufficient to decrease rapamycininduced beclin-1 phosphorylation at Ser30, LC3 lipidation, and the formation of endogenous large LC3 foci ( Fig. S9A-C). Notably, blocking autophagy initiation completely abolished the rapamycin-induced increase in the AP-2β-p150 Glued interaction, measured by IP and the PLA in HEK293T and Rat2 cells, respectively ( Fig. 4A-C). To further confirm that early steps of autophagy are required for the rapamycin-induced interaction of AP-2β and p150 Glued , we tested whether the knockdown of Atg5, a key protein for this process, exerts an identical effect as SBI-0206965. Rat2 cells were transfected with siAtg5 and siCtrl for 72 h. Western blot showed that siAtg5 effectively reduced Atg5 levels in transfected cells compared with the control and simultaneously inhibited the autophagy process, indicated by a decrease in the LC3B II/LC3B I ratio (Fig. S9D-F). The silencing of Atg5 also effectively counteracted the rapamycin-induced increase in the p150 Glued -AP-2β PLA signal (Fig. 4D, E). Thus, the induction of autophagy is needed for p150 Glued -AP-2β protein complex formation upon mTORC1 inhibition.

Autophagy initiation is sufficient for p150 Glued -AP-2β interaction
Our results above show that the initiation of autophagy is essential for effects of mTORC1 inhibition on p150 Glued -AP-2β complex formation. However, a key issue is whether autophagy initiation, even when mTORC1 is active, is sufficient to induce a similar effect. To investigate this possibility, we treated Rat2 cells with L-690330, an inhibitor of inositol monophosphatase and mTOR-independent activator of autophagy [76], and performed an p150 Glued -AP-2β PLA. After 3 h of L-690330 (100 µM) treatment, autophagy levels increased, indicated by the LC3B II/LC3B I ratio and formation of large LC3B foci, whereas the level of phosphorylated S6 at Ser235/236 did not decrease as expected ( Fig. 5A-C). This treatment also increased the p150 Glued -AP-2β PLA signal similarly to rapamycin treatment (Fig. 5D, E).
Altogether, our results show that autophagy induction alone is sufficient to induce AP-2dynactin complex formation. Additionally, as in the case of rapamycin treatment, CLIP-170 knockdown blocked the L-690330-induced increase in the p150 Glued -AP-2β interaction ( Fig.   5F-H). This observation suggests potential novel autophagy-related activities of CLIP-170.
Thus, we tested whether CLIP-170 knockdown affects autophagy that is induced by rapamycin or L-690330. The loss of CLIP-170 prevented the increase in the LC3B II/LC3B I ratio and LC3 foci formation that were caused by both autophagy inducers ( Fig. 5I-K). These findings suggest that the dynactin interaction with the AP-2 adaptor complex requires autophagy initiation, which depends on the presence of CLIP-170.

ABMA and chloroquine prevent rapamycin-induced p150 Glued -AP-2β interaction
Based on the data that were obtained, the initiation of autophagy appears to play a key role in the p150 Glued -AP-2β-protein interaction, but unknown is whether undisturbed autophagic flux is also required. Therefore, we treated cells with rapamycin in the presence of chloroquine (CQ) and ABMA, two compounds that affect this process via different mechanisms [77,78].
ABMA stimulates the formation of amphisomes, which however are unable to fuse with lysosomes to finish the autophagy [79]. Chloroquine appears to have pleiotropic effects that include direct blockade of the fusion of autophagosomes with lysosomes by preventing the recruitment of SNAP29 to the fusion site, slowing the acidification of lysosomes or disturbing endosomal flow [78,80]. Rat2 cells were treated with rapamycin in the presence of ABMA (60 µM) or CQ (50 µM) for 2 h. Cells that were treated with DMSO, ABMA, or CQ alone served as controls. Although fluorescence analysis did not show a profound enhancement of LC3B protein cluster formation after treatment with ABMA ( Fig. S10A), Western blot confirmed that it increased the LC3B II/LC3B I ratio as expected ( Fig. S10B, C) [77]. Treatment with CQ also resulted in autophagic flux inhibition as described previously (Fig. S10D-F) [78]. Both ABMA and CQ blocked the increase in the p150 Glued -AP-2β PLA signal that was caused by rapamycin ( Fig. 6A-D). ABMA and CQ prevent the fusion of autolysosome-preceding compartments with lysosomes but may also have additional effects on the endomembrane system. Therefore, we tested whether the knockdown of SNAP29, which plays a key role in autophagosome-lysosome fusion [81], affects the rapamycin-induced p150 Glued -AP-2β interaction. Rat2 cells were transfected with control siRNA or siRNA against Snap29 (siSnap29) for 72 h. The qRT-PCR analysis of RNA that were isolated from siRNA-transfected cells showed a significant decrease in Snap29 mRNA levels compared with controls ( Fig. S10G). Additionally, immunofluorescent staining and Western blot analysis showed an increase in p62/SQSTM1 protein ( Fig. S10H- which is expected to accumulate in the cell in the absence of Snap29 and with autophagosomelysosome fusion inhibition. However, Snap29 knockdown did not prevent the enhancement of the p150 Glued -AP-2β PLA signal by rapamycin (Fig. 6E, F). Thus, we concluded that both ABMA and CQ effectively prevented the rapamycin-induced p150 Glued -AP-2β interaction, but SNAP29-dependent fusion unexpectedly did not appear to be required.
Because CQ can alkalize the lysosome environment [80] and because ABMA was not tested for it in Rat2 cells, we used lysotracker staining to investigate effects of rapamycin, CQ, ABMA, and their combination on lysosomal acidification in our experimental model. As a control, we treated cells for 2 h with the vATPase inhibitor Baf A1. Two hours of treatment with rapamycin significantly increased the intensity of lysotracker staining and the number of positive lysotracker structures (Fig. S11). In contrast, both ABMA and CQ alone and in the presence of rapamycin significantly decreased both parameters (Fig. S11). However, effects of CQ and ABMA were different. Chloroquine had a stronger effect that was comparable to treatment with Baf A1, while influence of ABMA was much milder (Fig. S11). Thus, we concluded that although autophagosome fusion with the endolysosomal pathway did not affect the p150 Glued -AP-2β interaction, ABMA and CQ might affect it, leading to improper lysosome acidification or disruption of the endolysosomal pathway as previously shown [77,78]. To test the hypothesis that the correct pH in the cell is necessary for the rapamycin-induced interaction of p150 Glued and AP-2β, we repeated the experiment, this time treating the cells with 20 mM NH4Cl to alkalize the cells before administering rapamycin and running PLA (Fig. 6G, H). As in the previous experiments, rapamycin caused an increase in p150 Glued -AP-2β PLA. At the same time, administration of NH4Cl under basal conditions had no significant effect on PLA signal. However, incubation with NH4Cl counteracted the increase in p150 Glued -AP-2β PLA in response to rapamycin administration, supporting the hypothesis that appropriate acidification of the endolysosomal pathway is critical for the interaction of p150 Glued and AP-2β.

Rapamycin enhances the lysosomal p150 Glued -AP-2β interaction and affects Lamp-1 mobility
Based on the above observations, we investigated where in Rat2 cells p150 Glued and AP-2β interact. We used the PLA that was adjusted for EM, revealing that the p150 Glued -AP-2β PLA signal in rapamycin-treated Rat2 cells localized primarily to organelles that contained electron dense material that is characteristic for lysosomes (43% of cells with PLA-EM signal) and in double-membrane organelles, which we classified as autophagosomes (57% of cells with PLA-EM signal). In control, in DMSO-treated cells, the PLA-EM signal was also spotted but less frequently and rarely on lysosome-like structures ( Fig. 7A; 14% vs. 43% of cells with PLA-EM signal). Because the PLA procedure did not allow the preservation of a high-quality ultrastructure, we additionally analyzed in rapamycin-treated Rat2 cells the co-occurrence of the p150 Glued -AP-2β PLA signal with the LC3B or lysosomal marker Lamp-1, either endogenous or overexpressed as a GFP fusion, using AiryScan high-resolution light microscopy. For LC3B, some co-occurrence was detected but not frequently (Fig. S12). For Lamp-1, we noticed an apparent, although still partial, co-localization of the PLA signal with lysosomes ( Fig. 7B, C). Thus, the combined observations from PLA-EM and Airyscan images indicated that lysosomes are the primary localization of the p150 Glued -AP-2β interaction under rapamycin treatment, raising the issue of whether rapamycin affects the mobility of Lamp-1positive compartments.
We found that rapamycin significantly increased lysosome acidification (Fig. S11), and lysosome acidification was previously related to their position in the cell. Therefore, we investigated whether the rapamycin-induced p150 Glued -AP-2β interaction contributes to the mobility of lysosomes. We performed the live imaging of Rat2 cells that were transfected with Lamp-1-GFP and treated with 100 nM rapamycin for 2 h. Two distinct populations of Lamp-1-GFP objects were spotted. One aggregated near the cell center, and the other was more motile around the peripheries. When we focused on the latter one, we observed that rapamycin increased the speed of long-distance movements (> 6 µm) toward the cell center. Movement from the central Lamp-1-positive organelle pool toward the cell periphery was unaffected ( Fig.   7D, E).

Discussion
Recent work demonstrated that AP-2 is an adaptor protein for the dynein-dynactin transport of amphisomes along neuronal axons [22,24]. To date, however, important questions about mechanistic details of the regulation of the dynactin-AP-2 interaction have not been answered. Here, we demonstrate that AP-2 and dynactin cooperate in neurons and non-neuronal cells under autophagy-permissive conditions, including, but not limited to, mTOR inactivation.
Furthermore, we show that the co-occurrence of AP-2β with p150 Glued does not require binding the latter to microtubule plus ends or microtubule dynamics. However, this interaction requires the presence of CLIP-170, which contributes to autophagy initiation. Finally, we show that the autophagy-induced p150 Glued -AP-2β interaction likely occurs on lysosomes, possibly increasing their mobility toward autophagosomes in the perinuclear area.

Autophagy induction and a properly functioning endolysosomal pathway are essential for the p150 Glued -AP-2β interaction
The initial finding that stimulated our research was the observation that rapamycin enhanced AP-2-dynactin complex formation in neurons. This discovery raised the issue of how mTORC1 inhibition stimulates this interaction. The present findings show that AP-2β or p150 Glued is not an mTORC1 substrate, instead supporting the scenario that mTORC1 inhibition promotes p150 Glued -AP-2β complex formation via autophagy initiation (Fig. 4). Moreover, the pharmacological induction of autophagy that does not involve mTORC1 inhibition was sufficient to induce this interaction (Fig. 5), indicating that the direct trigger for the p150 Glued -AP-2 interaction is autophagy. Furthermore, it explains why the coexistence of AP-2β and p150 Glued is readily visible in axons, whereas it is barely detectable in non-neuronal cells under basal conditions. In cultured neurons, axonal autophagy is relatively high and stable at a steadystate level. Autophagosomes are continuously formed at the axonal growth cone or at presynaptic sites and likely fuse with late endosomes, forming amphisomes, and are transported toward the cell soma [24,28,45,[82][83][84]. Thus, there is a continual need for AP-2-dynactin complexes in axons to transport amphisomes [24]. However, if axonal autophagy is constitutive and if autophagy is sufficient to drive the p150 Glued -AP-2β interaction, then how does mTORC1 inhibition potentiate it? Although some studies show that mTORC1 inhibition does not increase autophagy in neurons [41,42], other studies reported that it is nevertheless possible [43,44].
Such observations show that this process can still be upregulated, despite its relatively high basal level. In contrast to neuronal axons, autophagy in many cultured non-neuronal cells is at a low level under basal conditions, and its induction requires additional stimuli (e.g., mTORC1 inactivation). Consequently, the degree of the p150 Glued -AP-2β interaction in such cells is likely to be adjusted to the current level of autophagy in the cell. In the context of requiring the initiation of autophagy for the p150 Glued -AP-2β interaction, it is also an interesting observation that this interaction requires the presence of CLIP-170. However, its important canonical functions (e.g., the regulation of microtubule dynamics) seem unnecessary. In contrast, we have shown that CLIP-170 is required for proper autophagy, but unclear is how CLIP-170 regulates this process. Thus, determining precisely how CLIP-170 regulates autophagy will require further studies.
Our previous experiments showed that AP-2β interacts with p150 Glued in neurons to transport amphisomes. Thus, we investigated whether unperturbed autophagosomal flux and the fusion of autophagosomes with organelles of the endolysosomal pathway are also required for this interaction in non-neuronal cells. Both inhibitors of these processes that we used (CQ and ABMA) significantly prevented the increase in the p150 Glued -AP-2β PLA signal. This result may suggest that unperturbed autophagic flux is indeed crucial. Because both ABMA and CQ counteract the fusion of autophagosomes and amphisomes with lysosomes, this could suggest that this step is critical. Indeed, previous studies showed that AP-2 attaches to autolysosomes when autophagic lysosome reformation (ALR) is initiated. Autophagic lysosome reformation is a process that links autophagy, lysosomes, and AP-2 [85]. During ALR, protolysosomes emerge from autolysosomes and mature into lysosomes. Protolysosome formation requires clathrin and AP-2 [26]. Because ALR requires the merging of autophagosomes with the endolysosomal pathway, this would support the importance of this event for the p150 Glued -AP-2β interaction. However, the lack of an effect of SNAP29 knockdown on the p150 Glued -AP-2β interaction likely excludes this possibility. Thus, a question arises about how ABMA and CQ can block the p150 Glued -AP-2β interaction. A common feature of these two compounds is their negative effect on the normal ultrastructure of the endolysosomal pathway [77,78]. Our results clearly demonstrate that in Rat2 cells both ABMA and CQ lead to a significant decrease in lysotracker-positive organelles and a decrease in its intensity in the remaining ones. A similar effect of CQ on the pH of acidic organelles has been reported [80], although its absence has also been reported [78], suggesting that this effect may depend on the cell type. ABMA is a relatively recently described compound [77,86], and its effect may also depend on the cell line and should always be tested experimentally. Thus, our results suggest that formation of the p150 Glued -AP-2β complex requires proper function of the endocytic pathway and/or lysosome acidification. The latter possibility is further supported by the results of experiments with NH4Cl, which directly indicate that alkalinization of the cell prevents the increase in the p150 Glued -AP-2β interaction upon rapamycin administration that initiates autophagy.

Autophagy-induced p150 Glued -AP-2β interaction occurs on lysosomes
Our results indicate that autophagy is a critical cellular process for the interaction of AP-2β with p150 Glued . Nevertheless, our results did not identify autophagosomes as the primary p150 Glued -AP-2β interaction site in non-neuronal cells. An important question is where the p150 Glued -AP-2β interaction occurs upon autophagy induction. As mentioned above, the AP-2dynactin complex in neurons is involved in the transport of signaling amphisomes in axons [24]. However, in non-neuronal cells, amphisomes are considered temporary structures [32,87]. Moreover, ABMA, which potentiates their formation [77], did not enhance the p150 Glued -AP-2β interaction. Finally, our PLA-EM and PLA-AiryScan confocal microcopy findings in Rat2 cells revealed a p150 Glued -AP-2β PLA signal that often co-occurred with lysosomes.
Notably, the presence of lysosomal AP-2 was previously reported and not only in the context of ALR, which corroborates our observation [88]. Thus, another issue emerges about the purpose of p150 Glued -AP-2β complex formation on lysosomes.
Korolchuk et al. [89] showed that the dynein-dynactin-dependent transport of lysosomes toward the cell nucleus under cell starvation conditions is crucial for their pH regulation and fusion with autophagosomes. Hence, recruitment of the p150 Glued -AP-2β complex to lysosomes at the onset of autophagy may be designed to ensure the proper fusion of these two organelles at the end of the process. A similar mechanism was previously reported for another microtubular transport lysosomal adaptor, ALG2 [51]. Under amino acid starvation or mTOR inhibition conditions, ALG2 and dynein are recruited to lysosomes in a Ca 2+ -dependent manner for their retrograde transport. Previous studies showed that RILP protein is responsible for lysosome transport by dynein-dynactin in response to changes in cholesterol levels [90]. This suggests that non-neuronal cells use several transport systems for one organelle, depending on the cellular conditions. This raises the question about what could trigger p150 Glued -AP-2β complex binding to lysosomes upon mTOR inhibition. Based on our findings that CQ and ABMA decreased lysotracker fluorescence intensity, a tempting speculation is that changes in pH could serve as one such mechanism, but further research is needed to confirm this possibility.
In summary, our study provides new insights into the mechanisms that regulate formation of the p150 Glued -AP-2β complex, which is essential for cargo transport along microtubules. Importantly, we showed that autophagy initiation is necessary and sufficient to trigger the formation of this complex. This finding exemplifies a basic mechanism that allows the coordination of various elements that are involved in a vital cellular process.

Funding Statement
Research was supported by Polish National Science Centre Opus grant no.

Conflict of Interest
None of the authors have any financial or non-financial competing interests.

Data availability
All data generated or analyzed during this study are included in this published article and its supplementary materials. Raw data from all quantitatively analyzed experiments are available from the corresponding author upon reasonable request.        area inside the golden circle is considered the "center" compartment, and all movements outside this area are "peripheries." The lower row shows trajectories (tracks) that were identified by the ImageJ "TrackMate" plugin that were longer than 6.8 mm (100 pixels). Trajectories were colorcoded based on their directions, which were established using Pearson correlation coefficient (PCC) calculated by change in the distance from the cell center in time. If the distance was increasing with consecutive frames (PCC: 0.5 to 1) tracks were considered to move to the cell membrane. If distance was decreasing (PCC: -1 to -0.5), direction was described as moving to the center. Values in between were marked as oscillating. (E) Difference in speed of Lamp1-GFP vesicles' movements between rapamycin-treated (RAPA) and control (DMSO) cells. The values are mean trajectories that were identified by the ImageJ "TrackMate" plugin that were at least 6.8 µm (100 pixels) long, divided according to their initial location (center or periphery) and direction (center or cell membrane) as indicated above the graphs. The single dot represents the mean value from one measured cell. N = 4 independent experiments. n = 25 cells for both RAPA and DMSO. *p < 0.05, ns, nonsignificant (Mann-Whitney test).

Fig. 8. Postulated mechanism of autophagy initiation-induced recruitment of p150Glued
and AP -2β to lysosomes. Administration of rapamycin or L-69030 initiates autophagosome formation (AP), which depends on the presence of CLIP -170. Simultaneously, autophagy inducers lead to a decrease in lysosomal pH (LY pH) and recruitment of the dynatin-AP2 complex to the lysosome (LY), presumably leading to intensification of LY retrograde transport to the perinuclear region and subsequent LY fusion with the mature AP. kndknockdown, SBI-0206965 -Ulk inhibitor.

Cell line cultures and transfection
Rat2 and HEK293T cells were purchased from the American Type Culture Collection and grown in Dulbecco's modified Eagle's medium (DMEM) with 4500 mg/ml of D-glucose that contained 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (all from Sigma-Aldrich) at 37°C in a 5% CO2 atmosphere. For the proximity ligation assay (PLA) and PLAelectron microscopy (EM) experiments, Rat2 cells were grown on glass coverslips or Thermanox plastic coverslips (Thermo Fisher; catalog no. 174950), respectively, coated with 0.2% gelatin for 1 h at 37°C. For live experiments, Rat2 cells were seeded on glass coverslips that were coated with poly-L-lysine (50 μg/ml in H2O for 1 h). For plasmid DNA transfection, HEK293T cells at 70% confluency were transfected using polyethylenimine PEI 25K (Polysciences, catalog no. 23966) according to the manufacturer's protocols. After transfection, HEK293T cells were grown in DMEM that was supplemented with 5% FBS for 48 h. Rat2 cells were transfected with plasmid DNA using electroporation. For each transfection, a total of 10 6 cells were suspended in Opti-MEM medium (Thermo Fisher, catalog no. 31985-047), mixed with 10 μg of DNA, and added to 2 mm gap cuvettes (Nepagene, catalog no. EC-002S). Cells were electroporated using a NEPA21 electroporator (Nepagene) with poring pulses (6, 150 V, 2.5 ms length, 50 ms interval, 10% decay rate), followed by transfer pulses (5, 20 V, 50 ms length, 50 ms interval, five pulses, 40% decay rate) with "±" polarity set for both. After electroporation, the cells were grown in DMEM with 10% FBS without antibiotics. The medium was changed the next day for a medium that contained 1% penicillin-streptomycin.
The siRNA transfection of HEK293T and Rat2 cells was performed on trypsinized cells using Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher, catalog no. 13778150) according to the manufacturer's protocol. For the combined siRNA transfection/DNA electroporation experiments, cells after siRNA transfection were seeded on 10 cm plates and allowed to attach. After 24 h, cells were harvested by trypsinization and electroporated as described earlier. After electroporation, each variant was seeded on glass coverslips that were coated with poly-L-lysine (50 μg/ml in H2O for 1 h).

Primary neuron preparation and transfection
Primary hippocampal cultures were prepared from embryonic day 18 rat brains as described previously [1]. The rats that were used to obtain neurons for further experiments were with Lipofectamine2000 (Thermo Fisher, catalog no. 11668019) as described previously [1], except that the incubation time with the transfection mixture was reduced to 2 h. DNA (2 µg) and 1.5 µl of Lipofectamine2000 were used per well of a 12-well dish.

Proximity ligation assay
For the PLA, cells were fixed for 5 min with ice-cold 100% methanol and 2 mM ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA) at -20°C, followed by 10 min with 4% paraformaldehyde (PFA)/4% sucrose in phosphate-buffer (pH 7.4). Fixed cells were washed three times with phosphate-buffered saline (PBS) and incubated with mouse anti-p150 Glued and rabbit anti-AP-2β antibodies that were diluted in PBS that contained 1% donkey serum and 0.2% Triton-X100 at 4°C overnight. The next day, the cells were washed twice for 5 min with PBS with 1% donkey serum and 0.2% Triton-X100, washed once for 1 min in antibody diluent that was provided in the manufacturer's PLA kit (Sigma-Aldrich, catalog no. DUO92101), and then incubated for 60 min at 37°C with relevant secondary antibodies that were conjugated to oligonucleotide PLUS or MINUS (Sigma-Aldrich; catalog no. DUO92002 and DUO92004, respectively) and diluted in antibody diluent. The coverslips were then washed twice for 5 min with buffer A (Sigma-Aldrich, catalog no. DUO82046).
Ligation and amplification were performed according to the manufacturer's protocol using  processed with a median filter (7  7  3 size) and dilated (11  11  3 neighborhood). The cells were segmented using Otsu thresholding, and single cell masks were constructed with iterative dilation of the nuclear masks. The dilation steps were ordered by the intensity of processed cell images. Volumes that corresponded to nuclei were then excluded from the cell masks.

Coefficients of intensity correlation (Pearson and Spearman) between Alexa 488 and Alexa 568
images were calculated on a cell-by-cell basis using the respective masks. The operation was repeated using the Alexa 568 images that were shifted by ±8 voxels in the x and y directions and ±2 voxels in the z direction. The respective coefficients were calculated using the union of two respective cellular masks and corresponded to the random association of Alexa 488 and Alexa 568 fluorescence in a cell. The raw correlation coefficients were then divided by their random counterparts (on a cell-by-cell basis) to create standardized Pearson and Spearman values.

Opera Phenix High content imaging of fixed cells
For the high-throughput analysis of effects of different drugs on lysosomal acidity, Rat2 cells were seeded on CellCarrier-96 well Black glass bottom plates (Perkin Elmer) that were coated with 0.2% gelatin at a density of 7  10 3 cells per well 1 day before treatment (see above). After 80 min of drug exposure, LysoTracker Red DND-99 (500 nM, Invitrogen, catalog no. L7528) was added to the cells for the last 40 min. Subsequently, the cells were fixed with 4% PFA for 10 min and subjected to three 10-min washes with PBS. High-content screening microscopy was conducted using the Opera Phenix system (PerkinElmer) that was equipped with a 40 1.1 NA water immersion objective. Image acquisition and subsequent analysis were performed using Harmony 4.9 software (PerkinElmer). Statistical analyses were performed using the R and RStudio software packages (Cran). To ensure consistent and reliable data analysis, cells were initially filtered based on morphological criteria to obtain uniform cell populations. The identification of lysotracker-positive compartments was achieved using the "Find spots" algorithm that is integrated within the Harmony software. Additionally, the total signal intensity of Lysotracker staining was quantified within both the cell cytoplasm and the previously defined compartment regions.

Protein production in bacteria and pull-down experiment
Recombinant proteins were produced in the E. coli BL21 strain. Individual clones that were transformed with plasmids that encoded proteins of interest were picked from the plates and inoculated to 5 ml of LB with appropriate antibiotic. After overnight culture, at 37°C with shaking, bacteria were refreshed with new medium in a 1:50 ratio and further cultured until reaching an optical density at 600 nm (OD600) of 0.6-0.8. Protein production was induced using 1 mM isopropyl-β-D-1-thiogalactopyranoside (Carl Roth, catalog no. CN08.3). The His6-AP-2β appendage domain, GST-Eps15, and free GST-tag were produced at 37°C for 2 h, and GST-p150 Glued fragments were produced at 21°C overnight. For protein purification, cultures were centrifuged at 4,500  g for 10 min, and the pellet was resuspended in lysis buffer (50 mM Tris [pH, 8.0], 150 mM NaCl, 0.1% Triton X-100, and protease inhibitors). For all subsequent stages, lysates were kept on ice. After resuspension, the cells were lysed by sonication in a Sonics VCX130 PB sonicator (Vibra-Cell) in two 30-s sessions with a 70% amplitude. After sonication, lysates were centrifuged at 13,000  g for 5 min to remove the insoluble fraction.
For the pull-down experiment, the bait protein was then added to Glutathione-Sepharose 4B resin (Merck, catalog no. GE17-0756-01) at a ratio of 30 μl beads for each 10 ml of original bacterial culture and incubated for 1 h with end-to-end rotation. After incubation, the resin was washed three times with lysis buffer, and the prey protein lysate was added. After another 1 h of incubation, the resin was again washed three times with lysis buffer and prepared for Western blot by dissolving in 1 Laemmli buffer and incubation in a heat block at 94°C for 15 min.

Whole-cell lysate preparation
For the Western blot analysis of proteins in whole-cell lysates, followed by Western blot. The IP of AP-2β from rat brain extracts was performed as described previously [2]. For the IP of heterologous proteins, HEK293T cells were transfected with pEGFPC1-Ap2b1, GFP-β-actin-p150 Glued , or pEGFPC1 (as a negative control). Forty-eight hours after transfection, the cells were lysed for 15 min on ice in lysis buffer (10 mM Tris [pH 7.5], 150 mM NaCl, 0.5 mM EDTA, and 0.15 % CHAPS supplemented with protease and phosphatase inhibitors) and spun using a benchtop centrifuge at maximum speed for 15 min at 4°C. The lysates were then incubated with GFP-Trap Agarose (20 µl per variant; Chromotek, catalog no. gta-10) for 2 h at room temperature, followed by four washes with wash buffer (10 mM Tris [pH 7.5], 150 mM NaCl, and 0.5 mM EDTA). The immunoprecipitated proteins were next used for the mTOR kinase assay.

Avi-tag pull down of biotinylated proteins and AP-2β-ear binding
The His6-AP-2β appendage domain was produced as described in the Protein production in bacteria section. For protein purification, cultures were centrifuged at 4,500  g for 10 min, and the pellet was resuspended in lysis buffer (50 mM NaH2PO4/Na2HPO4 [pH 8], 300 mM NaCl, 10 mM imidazole, 10% glycerol, 10 mM β-mercaptoethanol, 0.1% Triton X-100, and protease inhibitors). For all subsequent stages, lysates were kept on ice. After resuspension, the cells were lysed by sonication in a Sonics VCX130 PB sonicator (Vibra-Cell) in two 25-s sessions with a 70% amplitude. After sonication, the lysates were centrifuged at 13,000  g for 5 min to remove the insoluble fraction. The column with 5 ml agarose -Ni -NTA used for affinity chromatography of recombinant protein was equilibrated with 20 ml of lysis buffer and a flow rate of approximately 1 ml/min. The cell lysate that was obtained from 2 L of culture was then added to the column at a flow rate of approximately 0.5 ml/min. After the lysate passed through the column, the agarose resin was washed successively with 50 ml of

Western blot
Protein samples were analyzed by SDS-PAGE and immunoblotting according to standard laboratory protocols. Primary antibodies that were used for Western blot are listed in Table 1. Proteins of interest were detected using HRP-or IRdye-conjugated secondary antibodies in a chemo-luminescence reaction or with the Odyssey LiCor Biosciences system, respectively. The densitometry analysis of the amount of LC3B was performed using Image Studio Lite (LiCor Biosciences) and a previously described method [3]. Specifically, the signal intensities for LC3B I and LC3B II were first measured. Based on these values, normalization was performed. First, for all variants in a single replicate, the sum of intensities for a given protein was calculated. Second, individual protein band intensities were divided by this sum.
Such normalization was performed for LC3B I and LC3B II. Finally, after normalizing the values, the ratio between LC3B II and LC3B I was calculated by dividing the normalized values of LC3B II by LC3B I.

Fig. S1. mTORC1 inhibition in neurons increases p150 Glued -AP-2β interaction in axons.
(A) Snapshots of DIV7 neurons that were treated as indicated and expressed p150 Glued -Tdtomato (red) and AP-2β-GFP (green) and stained for the axon initial segment with CF640R-        cells that were treated as in G, with immunofluorescently labeled endogenous p62/SQSTM1 (red) and nuclei that were stained with Hoechst 33258 (blue). Scale bar = 10 µm. (I) Western blot analysis of endogenous protein levels as indicated in Rat2 cells that were treated as in G.
(J) Densitometry analysis of normalized p62/SQSTM1 in Rat2 cells that were treated as in G.
The data are presented as mean of the normalized ratio of p62 to tubulin levels ± SEM. N = 3 independent experiments. **p < 0.01 (Student's t-test).   Movie 3. Hippocampal neuron that was transfected with plasmids that encoded p150 Glued -Tdtomato and AP-2β-GFP and treated for 2 h with 100 nM rapamycin. Speed = 12 real time.