The endolysosomal adaptor PLEKHM1 is a direct target for both mTOR and MAPK pathways

The lysosome is a cellular signalling hub at the point of convergence of endocytic and autophagic pathways, where the contents are degraded and recycled. Pleckstrin homology domain‐containing family member 1 (PLEKHM1) acts as an adaptor to facilitate the fusion of endocytic and autophagic vesicles with the lysosome. However, it is unclear how PLEKHM1 function at the lysosome is controlled. Herein, we show that PLEKHM1 coprecipitates with, and is directly phosphorylated by, mTOR. Using a phosphospecific antibody against Ser432/S435 of PLEKHM1, we show that the same motif is a direct target for ERK2‐mediated phosphorylation in a growth factor‐dependent manner. This dual regulation of PLEKHM1 at a highly conserved region points to a convergence of both growth factor‐ and amino acid‐sensing pathways, placing PLEKHM1 at a critical juncture of cellular metabolism.

tyrosine kinases (RTKs) by their cognate growth factor (e.g. epidermal growth factor receptor (EGFR) by EGF) results in dimerization of the RTK, transautophosphorylation, activation, recruitment of small GTPases and signal transduction through a cascade of kinases [2,3]. For example, EGF stimulation of EGFR leads to the activation of Ras GTPase, recruitment and activation of RAF kinase, then MEK (MAP kinase kinase; MKK) and then ERK 1/2 (MAPK). Activated ERK then phosphorylates a number of intracellular targets and simulates cellular proliferation [2,3].
In addition to stimulating MAPK, growth factormediated activation of RTKs activates the mTOR pathway. mTOR is a serine/threonine kinase and a catalytic subunit of two mTOR complexes -mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2). The two protein complexes possess different functional properties since they interact with different accessory proteins. Cell growth is primarily controlled by mTORC1, whereas mTORC2 is involved in the regulation of cell proliferation and survival and is activated by growth factor stimulation of RTKs [1]. Both complexes share mTOR, mLST8 (mammalian lethal with Sec13 protein 8) and Deptor (DEP-domain-containing mTOR-interacting protein) as subunits [4,5], with mTORC1 also containing Raptor (regulatory-associated protein of mTOR) and PRAS40 (proline-rich AKT substrate of 40 kDa), as additional subunits [6][7][8][9]. mTORC2 contains Rictor (rapamycininsensitive companion of mTOR), mSIN1 and Protor [10][11][12]. mTOR signalling is essential for monitoring nutrient availability in cells, and it has a central function in the process of autophagy [1]. mTORC1 is localized to the lysosomal surface during nutrient-rich conditions, where it interacts with Rag GTPases through Raptor [13]. During periods of amino acid starvation, mTORC1 is inhibited, leading to its dissociation from Rags, and therefore the lysosome; in contrast, mTORC2 couples growth factor and cytokine signalling to mTORC1 functions through direct phosphorylation of AKT [14] and is responsive to amino acids through the PI3K/AKT pathway [15]. mTORC1 suppresses autophagy flux through phosphorylation of the ULK1 (Unc-51 like autophagy activating kinase) complex that is restored upon mTOR inhibition [16,17]. Inhibition of mTORC1 leads to ULK1 activation and subsequent activation of the autophagy pathway through recruitment and phosphorylation of several downstream proteins, resulting in the generation of a double-membraned vesicle, the autophagosome [18][19][20][21][22]. The autophagosome sequesters the cargo marked for degradation and subsequently fuses with the lysosome where the cargo is degraded to recycle protein and lipid components back into the cytosol, re-activating mTOR [23].
The lysosomes act as a recycling hub for proteins and organelles, as well as a point of convergence for endosomal and autophagy pathways. Previously, we and others have identified that PLEKHM1 (Pleckstrin homology domain-containing family member 1) is an adaptor protein linking both endocytic and autophagy pathways through its interaction with Rab7, HOPS (homotypic fusion and vacuole protein sorting) and Arl8b [24][25][26][27]. HOPS is a tethering complex involved in the fusion of endocytic vesicles and is comprised of six subunits -VPS39, VPS41, VPS11, VPS16, VPS18 and VPS33. Rab7 and Arl8b are GTPases, involved in the transport and fusion of late endocytic/autophagy vesicles. PLEKHM1 directly binds Rab7, facilitating its own recruitment to the membranes [25], while its binding to Arl8b promotes its lysosomal localization and facilitates its interaction with HOPS complex [27]. In order to fulfil its function as an adaptor in endosome/lysosome pathway, PLEKHM1 requires simultaneous binding to both Rab7 and Arl8b [27]. However, the regulation of PLEKHM1 function and the dynamics of its intricate association with lysosomes is yet to be understood. Various mass spectrometry screens have reported phosphorylation sites on PLEKHM1; however, none of them explore in depth the consequences of such post-translational modification (phosphosite.org). Using a quantitative mass spectrometry approach, we discovered that PLEKHM1 interacts with mTOR and is directly phosphorylated at two proline-directed serine motifs within a highly conserved region of PLEKHM1 located between the RUN and first PH domain. Furthermore, we demonstrate that these sites match a MAPK consensus sequence (PxSP) and are also direct targets for MAPK-mediated phosphorylation. Using a phosphospecific antibody, we show that the inhibition of mTOR and MAPK decreases phosphorylation of PLEKHM1 at Ser432/ Ser435. Our data suggest direct implication of mTOR and MAPK in regulating PLEKHM1 phosphorylation, and potentially its function, at the late endosome/lysosome, hinting that PLEKHM1 sits at the regulatory crossroads of these two essential signalling pathways.

Cell culture
HeLa Kyoto [52] and U2OS cells were obtained from ATCC and grown in full medium (DMEM supplemented with 10% foetal bovine serum, 1% non-essential amino acids, 1% sodium pyruvate and penicillin/streptomycin). HeLa FlpIn-TREx PLEKHM1-GFP cells were previously described [26]. Protein expression was induced by 1 µgÁmL À1 doxycycline. U2OS cells with endogenous GFP-PLEKHM1 were generated using CRISPR/Cas9-mediated modification of human PLEKHM1 (NM_014798.3) using the Cas9 D10A 'nickase' mutant. Optimal sgRNA pairs were identified and chosen on the basis of being as close as possible to the point of GFP insertion while having a low combined off-targeting score (sgRNA 1 5 0 -GATGACACC TGAAGAGACAGA-3 0 and sgRNA2 5 0 -TACTTGCTG CACCACTACTTAGG-3 0 ). The antisense guide (sgRNA2) was cloned onto the spCas9 D10A expressing pX335 vector (Addgene plasmid no. 42335) and the sense guides (sgRNA1) into the puromycin-selectable pBABED P U6 plasmid (Dundee-modified version of pBABE-puro plasmid). A donor construct consisting of GFP flanked by approximately 500-bp homology arms was synthesized by GeneArt (Invitrogen, Life Technologies, Paisley UK); each donor was engineered to contain sufficient silent mutations to prevent recognition and cleavage by Cas9 nuclease. The U2OS cells were transfected with both sgRNA and donor constructs and selected in 1 lgÁmL À1 puromycin for 48 h, retransfected and allowed to recover in puromycin-free full medium. When confluent, cells were single-cell sorted for GFP-positive populations and homozygous clones selected for further analysis. MEFs expressing HA-PLEKHM1 were generated through retro-viral transduction with MSCV-N-Flag-HA-IRES-PURO-PLEKHM1-WT and selected using 2 µgÁmL À1 puromycin. Where indicated, cells were incubated in Earle's balanced salt solution (EBSS) medium to induce amino acid starvation; or treated with Bafilomycin A1 (200 nm) or chloroquine (20 µm) to block fusion with lysosomes; torin 1 (250 µm) or Ku-0063794 (10 µm) (both from Tocris) to inhibit mTOR activity; or U0126 (10 µm) (Promega, Southampton, UK) to inhibit MAP kinases, for the time indicated in the figure legends. Transient transfection of HeLa cells was performed using TurboFect, according to the manufacturer's instructions.

Immunoprecipitation
Cells were lysed in CHAPS lysis buffer (50 mm Tris/HCl [pH 7.5], 0.5% CHAPS, 115 mm NaCl and 5 mm MgCl 2 ) supplemented with EDTA-free protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail (Sigma). Lysates were cleared by centrifugation, and the clear supernatant was incubated with primary antibody, rotating overnight at 4°C, followed by incubation with protein A sepharose beads (1 h with rotation at 4°C), for endogenous immunoprecipitation, or with either GFP-TRAP beads (ChromoTek GmbH, Martinsried, Germany) or HAaffinity matrix (Sigma) for 30 min, rotating at 4°C. Pellets were washed three times with lysis buffer and four times with wash buffer (50 mM Tris/HCl [pH 7.4], 150 mM NaCl, 2 mM MgCl 2 and 0.1% Triton X-100). Proteins were eluted in 2x Laemmli sample buffer and subjected to SDS/PAGE and western blotting.

Western blotting
CHAPS (50 mm Tris/HCl [pH 7.5], 0.5% CHAPS, 115 mm NaCl, and 5 mm MgCl 2 ) or total cell lysis buffer (50 mm Tris/HCl [pH 7.4], 5 mm MgCl 2 , 150 mm NaCl, 1% SDS), supplemented with EDTA-free protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail (Sigma), was used for cell lysis. Phosphatase inhibitors were omitted from samples treated with lambda-phosphatase. Lysates were cleared by centrifugation. For dephosphorylation of PLEKHM1, either whole cell lysates or PLEKHM1 immunoprecipitated on GFP-TRAP beads were incubated with 1 µL of lambda-phosphatase (NEB) with the accompanying manufacturer's 10x phosphatase buffer and 1 mM MnCl 2 , for 30 min at 30°C. 2x Laemmli sample buffer was added to samples to stop the reaction. Proteins were resolved in 7% polyacrylamide gels or 5% polyacrylamide containing 20 µM Phos-Tag (Nard) and 10 mM MnCl 2 , for protein mobility shifts. The proteins were then transferred onto a 0.2-µm PVDF membrane (Merck Millipore). Following incubation with primary and secondary antibodies, the blots were developed by ImmunoCRUZ western blotting Luminol reagent (Santa Cruz). Quantification was performed with FIJI/IMAGEJ software.

Mass spectrometry
Sample preparation for mass spectrometry For testing of interaction partners, PLEKHM1-GFP was immunoprecipitated in HeLa FlpIn-TREx PLEKHM1-GFP cells on GFP-TRAP beads. Additionally, for detection of phosphorylation sites, SILAC labelling was used. HeLa FlpIn-TREx PLEKHM1-GFP cells were cultured in DMEM supplemented with either L-arginine 12 C 6 14 N 4 (Arg0) and L-lysine 12 C 6 14 N 2 (Lys0) or L-arginine 13 C 6 15 N 4 (Arg10) and L-lysine-U 13 C 6 15 N 2 (Lys8) as previously described (Ong et al., 2002). SILAC-labelled cells were lysed in CHAPS lysis buffer. PLEKHM1-GFP expression was induced for 16h with doxycycline, and HeLa FlpIn-TREx cells, cultured in light SILAC-labelled medium, were treated in EBSS for 4 h to starve cells of amino acids. PLEKHM1-GFP was immunoprecipitated using GFP-TRAP, and eluates were mixed 1:1 (v/v) and run on an SDS/PAGE gel. After excision of individual gel bands and destaining according to manufacturer´s instructions, proteins were reduced with 10mM DTT for 45 min at 56°C and alkylated with 55mM Iodoacetamide for 30 min at RT under light protection. After digestion with Trypsin at 37°C overnight, tryptic peptides were extracted from the gel pieces consecutively with 30% acetonitrile (ACN) containing 3% trifluoroacetic acid (TFA), then 80% ACN containing 0.1% TFA and finally 100% ACN for 30 min each. Extraction solutions were combined and dried by vacuum centrifugation before reconstitution in 0.1% formic acid for LC-MS/MS analysis.

LC-MS analyses
Peptides were analysed on an Orbitrap Elite mass spectrometer coupled to an easy nLC II (Thermo Fisher Scientific, Paisley, UK) using a 20 cm long, 75 µm ID fused-silica column, which has been packed in house with 3 µm C18 particles (ReproSil-Pur, Dr. Maisch), and kept at 45°C using an integrated column oven (Sonation). Peptides were eluted by a non-linear gradient from 8-40% acetonitrile over 240 min and directly sprayed into the mass spectrometer via a nano-Flex ion source (Thermo Fisher Scientific) at a spray voltage of 2.3 kV. Full scan MS spectra (300-2000 m/z) were acquired at a resolution of 120,000 at m/z 200, a maximum injection time of 100 ms and an AGC target value of 1 9 10 6 charges. Up to 20 most intense peptides per full scan were isolated in the ion-trap using a 2 Th window and fragmented using collision-induced dissociation (CID, normalized collision energy of 35). MS/MS spectra were acquired in rapid mode using a maximum injection time of 25 ms and an AGC target value of 5 x 10 3 . Ions with charge states of 1 and > 6 as well as ions with unassigned charge states were not considered for fragmentation. Dynamic exclusion settings were 1 repeat count and 30-s repeat duration as well as an exclusion duration of 90s in order to minimize repeated sequencing of already acquired precursors.

Mass spectrometry data processing
Mass spectrometry raw data processing was performed with MaxQuant (v 1.3.0.5) applying default parameters. Acquired spectra were searched against the human reference proteome (Taxonomy ID 9606) downloaded from UniProt (09-2012) and a collection of common contaminants (244 entries) using the Andromeda search engine integrated in MaxQuant [53,54]. Identifications were filtered to obtain false discovery rates (FDR) below 1% for both peptide spectrum matches (PSM; minimum length of 7 amino acids) and proteins using a target-decoy strategy [55]. See Data S1 for analysis. Raw data are available via Pro-teomeXchange with identifier PXD021199.

Confocal microscopy
U2OS cells endogenously expressing GFP-PLEKHM1 were seeded on glass coverslips and, on the following day, starved in EBSS or treated with Bafilomycin A1 for the time periods indicated in the figure legends. Cells were then fixed in 4% paraformaldehyde (PFA; Santa Cruz, 30525-89-4) for 12 min minimum. Cells were permeabilized with 0.1% saponin in PBS at room temperature, followed by incubation with the primary antibodies in 0.1% saponin/ 5% BSA (bovine serum albumin)/PBS (phosphate-buffered saline). The coverslips were washed three times in 0.1% saponin/PBS, before being incubated with the secondary antibodies and DAPI (Molecular Probes, Lifetechnologies, Paisley, UK) for 1 h in 0.1% saponin/5% BSA/PBS. For detection of endogenous GFP-PLEKHM1, nanobody boosters towards GFP (anti-GFP, Atto-488 coupled, Chromotek; gba488-100) were used to enhance the signal. After incubation in secondary antibodies, the coverslips were washed twice in PBS/0.1% saponin, once in PBS and once in ddH 2 O to remove residual saponin prior to mounting in ProLong Diamond Antifade containing Mowiol (Invitrogen, p36965). Cells were imaged using a Zeiss 710 confocal microscope with a 63 9 objective lens. Subsequent image analysis was performed using FIJI/IMAGEJ [56].

Protein expression and purification
GST-PLEKHM1 aa418-453 was cloned into pGEX-4T-1 (GE Healthcare) and expressed in E. coli BL21 (DE3) cells in LB medium. Expression was induced by addition of 0.5 mM IPTG, and, subsequently, cells were incubated at 16°C overnight. Harvested cells were lysed using sonication in a lysis buffer (20 mM Tris/HCl [pH 7.5], 10 mM EDTA, 5 mM EGTA, 150mM NaCl), and the supernatant was subsequently applied to Glutathione Sepharose 4B beads (GE Healthcare, Hatfield, UK). After several washes, fusion protein-bound beads were used directly in the kinase assay.

In vitro kinase assay
Purified GST-tagged PLEKHM1 (aa 418-453) was washed 3 times in kinase buffer (10 mM HEPES, [pH 7.5], 50 mM NaCl, 50 mM b-glycerophosphate, 1 mM dithiothreitol [DTT], 10 mM MgCl 2 , 4 mM MnCl 2 ) and subsequently incubated with 300 ng of a catalytic fragment (spanning amino acids 1362 to 2549) of human mTOR kinase (40061; BPS Bioscience) in the presence or absence of ATP (100 lM) for 1 h at 30°C. 250 nM torin 1 was used to inhibit the in vitro kinase activity. For MAPK in vitro assay, ERK2 WT or kinase dead (D168A) (10 ng/100 µL reaction) was activated by incubating with MEK (10 ng/100 µL reaction) in kinase buffer (100 mM NaCl, 50 mM Tris/HCl, pH 7.0, 10 mM MgCl2, 100 mM ATP and 0.1% bovine serum albumin) for 30 min at 30°C. Recombinant GST-PLEKHM1 was added with additional cold ATP (100 µM), and the reaction was continued for 30 min at room temperature. After incubation, the reaction was stopped by adding 2x Laemmli buffer and by boiling at 95°C for 5 min prior to SDS/PAGE and western blot analysis.

Statistical analysis
Results were plotted and analysed using GraphPad Prism software using the statistical method outlined in the figure legends.

PLEKHM1 is found in complex with mTOR
Previously, we identified PLEKHM1 as an interaction partner of Rab7 [25] and subsequently showed that PLEKHM1 is present on LAMP1/LAMP2/ Rab7-positive vesicles, which facilitates the fusion of autophagosomes and lysosomes [26] and influences the Salmonella-containing vacuole (SCV) formation [25]. As PLEKHM1 localizes at the lysosomal compartment through its interaction with Rab7 [27], we performed mass spectrometry analysis of the PLEKHM1 interactome using a Zwitter-ionic buffer (CHAPS) that enriches membrane-bound proteins, to gain a better understanding of the role of PLEKHM1 in this compartment. We performed SILAC-based LC-MS-MS on trypsinized immunoprecipitated complexes derived from HeLa-TREX cells expressing WT PLEKHM1-GFP (HEAVY) under doxycycline control for inducible expression and compared the interactome to un-induced cells (LIGHT). Using this method, we detected a number of late endocytic and lysosome-associated proteins (Fig. 1A), including Rab7, LAMP1 and LAMP2, in addition to HOPS complex and multiple SNAREs (Fig. 1A) [25,26]. These were found readily in our interactome and served as validation of the assay (Fig. 1A). Given our interest in the regulation of PLEKHM1 and the quantitative nature of LC-MS-MS, we observed a distinct subset of interaction partners of which mammalian target of rapamycin (mTOR) was one of the most abundant. Using the number of unique peptides detected (arrow thickness Fig. 1A), we pinpointed a strong interaction between mTOR and PLEKHM1 (Fig. 1A). As mTOR is known to localize to lysosomes and regulate autophagy [1], we focused our studies on components of the mTOR signalling pathway. We were able to identify components of the mTORC1 complex (Raptor), mTORC1 regulatory GTPases (Rheb, RagA/B), mTORC2 complex (Rictor, Protor), Ragulator (Lam-tor1, Lamtor2), Gator1 (NPLR3) and Gator2 complex (MIOS, SEH1L, WDR59, SEC13), as well as the constitutive mTOR interaction partner mLST8 (Fig. 1A). PLEKHM1-GFP could coprecipitate with mTOR, Raptor, Rictor, mLST8 and VPS41 in doxycycline-induced HeLa-TREX cells, but not in un-induced immunoprecipitates (Fig. 1B). Using CRISPR/ Cas9, we generated endogenous GFP-tagged PLEKHM1 (Fig. S1A) and could observe PLEKHM1 colocalizing with mTOR-on LAMP2-positive lysosomes (Fig. 1C, upper panels) that were enlarged upon Bafilomycin A1 treatment (V-ATPase inhibitor; Fig. 1C lower panels). Under nutrient-rich (full medium) conditions, we observed PLEKHM1-GFP (green) localized on LAMP2-(magenta) and mTOR-(red) positive vesicles (Fig. 1D upper panels and quantified in Fig. 1E). Upon amino acid starvation, PLEKHM1 was retained on LAMP2-positive vesicles, but mTOR became dispersed and reduced colocalization with PLEKHM1-GFP (Fig. 1D, lower panels, insets and quantified in Fig. 1E). Thus, PLEKHM1-GFP localizes and interacts with mTOR on the lysosome. Given the previously reported role of PLEKHM1 in facilitating the fusion of late endosomes and autophagosomes with the lysosome [25,26], we hypothesized that mTOR phosphorylates and potentially regulates PLEKHM1 at the lysosome to control its function.
mTOR directly phosphorylates PLEKHM1 at a conserved serine motif mTOR is a key regulator of cellular metabolism, plays a key role in amino acid sensing at the lysosome and acts as one of the main 'on/off' switches for the autophagy pathway [16,17]. Therefore, we next asked whether PLEKHM1 is phosphorylated in an mTORdependent manner. To inhibit mTOR, we used either torin 1, or Ku-0063794, both of which are highly specific towards mTOR, and do not affect PI3K activity at the concentrations used [28,29]. Upon either mTOR inhibition with Ku-0063794 (KU) or starvation (EBSS) conditions, we could observe increased mobility shift of endogenous PLEKHM1, compared to full medium controls ( Fig. 2A). After a period (4 h) of starvation and refeeding for 40 min in full medium, this increased mobility shift was reversed ( Fig. 2A), implying that under these conditions PLEKHM1 may be post-translationally modified. Immunoprecipitation of endogenous PLEKHM1 and subsequent treatment with lambda-phosphatase (k-PPase) resulted in increased mobility of the immunoprecipitated PLEKHM1 (Fig. 2B). Next, HeLa FlpIn-TREx cells were induced for 16 h overnight with doxycycline to express PLEKHM1-GFP and were treated for increasing time in starvation medium. The cells were subsequently lysed and run on a phos-tag gel. The inclusion of the phos-tag chelating agent in conventional polyacrylamide gels retards the migration of phosphorylated protein species and can help distinguish different states of phosphorylation of the protein of interest [30]. Phosphatase-treated PLEKHM1-GFP sample showed a large increase in mobility, and increasing time of mTOR inhibition showed a range of mobility shifts between the untreated (full medium) and phosphatase-treated (Fig. 2C). This indicates that PLEKHM1-GFP phosphorylation is decreased upon mTOR inhibition (Fig. 2C). Next, we induced PLEKHM1-GFP expression in HeLa-TREX cells and left the cells untreated or starved them of amino acids and growth factors for up to 2 h (starve). Using GFPtrap beads, we immunoprecipitated PLEKHM1-GFP and probed with a generic phosphoserine antibody and monitored the effect on PLEKHM1 phosphorylation. As a negative control, we treated immunoprecipitated PLEKHM1 for 30 min using k-PPase to dephosphorylate PLEKHM1. Increased time in starvation medium resulted in decreased PLEKHM1 phosphorylation up to approximately 70% of total phosphorylated PLEKHM1 after 2 h of starvation ( Fig. 2D and quantified in 2E). PLEKHM1 phosphorylation (using phosphoserine antibody) was decreased by almost 70% after treatment with 10 µM Ku-0063794 for 4 h (Fig. 2F). This was associated with decreased mTOR association with PLEKHM1 under these conditions (~60% loss of mTOR; Fig. 2F) that was not prevented by lysosome inhibition (chloroquine, CQ; Fig. 2F). The majority of VPS41 (HOPS complex) was still bound to PLEKHM1 under these conditions (~70% bound; Fig 2F). Next, we wanted to map potential PLEKHM1 phosphorylation sites that changed upon amino acid starvation and mTOR inhibition. Using SILAC (stable isotope labelling of amino acids in cell culture), to differentiate changes in phosphorylation status of PLEKHM1, we labelled doxycycline-induced PLEKHM1-WT cells with either LIGHT or HEAVY amino acids, and either starved (LIGHT) the cells or left them in full medium (HEAVY) for 2 h. We also compared the phosphorylation status of PLEKHM1-GFP in untreated (full medium + DMSO) versus KU-treated cells. We immunoprecipitated PLEKHM1-GFP using GFP-Trap beads and analysed by LC/MS to map-specific amino acid residues on PLEKHM1 that change upon amino acid depletion/KU treatment (Fig. 2G). In total, we mapped nine potential phosphorylation sites that decreased upon starvation or KU treatment. Of these sites, only Ser280, Ser305, Ser432, Thr434, Ser435 and Ser501 (Fig. S1B-G) overlapped between the two conditions (Starved and KU). Of particular interest were the clusters of phosphoserine/threonine around 431-435 on PLEKHM1 (Fig. 2H,I). These sites have also previously been found to be enriched in multiple independent mass spectrometry analyses (phosphosite.org). These include analysis of the ultradeep human phosphoproteome [31] and were found to be increased in a number of cancer cell types including breast cancer [32], non-small-cell lung cancer [33] and leukaemia [34]. Interestingly, Ser432/435 was shown to decrease upon rapamycin treatment [35], and the same sites have also been shown to be increased upon insulin treatment [36]. Therefore, due to the number of hits at these sites and due to the fact that this cluster of phosphoserine residues is present within a highly conserved region of PLEKHM1 ( Fig. 2H; alignment), we wanted to verify that these were indeed mTOR-directed phosphorylation sites on PLEKHM1.

PLEKHM1 is directly phosphorylated by mTOR
Next, we developed phosphospecific antibodies directed against a dual phosphorylated peptide of PLEKHM1 containing phosphorylated Ser432/Ser435. Upon transient transfection of the PLEKHM1 WT and Ser432A/Ser435A (2A) plasmids, we could show that the phosphospecific antibody only recognized phosphorylated PLEKHM1-WT and not the PLEKHM1 protein in which Ser432/Ser435 sites are mutated to alanine (Fig. 3A). Moreover, phosphorylation of PLEKHM1 at these sites decreased upon amino acid starvation in HeLa FlpIn-TREx cells overexpressing PLEKHM1-GFP (Fig. 3A), and phosphorylation at these sites was sensitive to torin 1, Ku-0063794 and amino acid starvation (Fig. 3B). We were also able to detect phosphorylated PLEKHM1 that was sensitive to amino acid starvation using mouse embryonic fibroblasts expressing HA-tagged PLEKHM1 (Fig. 3C). Importantly, we could detect endogenous levels of phosphorylated Ser432/435 PLEKHM1 in HeLa cells that was sensitive to amino acid starvation (Fig. 3D). Using bacterially purified GST-PLEKHM1 (amino acids 418-453) that contains the potential mTOR phosphorylation sites, we showed that purified mTOR can phosphorylate Ser432/435 in vitro as phosphorylation (as detected by pS432/ pS435 antibody) was sensitive to the presence of ATP and torin 1 (Fig. 3E). Overexpression of WT PLEKHM1-GFP showed that PLEKHM1 is phosphorylated at S432/S435 on LAMP2-positive vesicles (lysosomes; (Fig. 3F, upper panels) and pS432/pS435 is lost when these sites are mutated (Fig. 3F, lower panels). Next, we wanted to test whether mutation of these phosphosites affected the previously identified interactions of PLEKHM1 [25,26,37]. We tested the interaction of overexpressed HA-tagged PLEKHM1 WT, 2A or 2D (S432D/S435D) with either GFP alone, GFP-Rab7 (Fig. S2 A-B) or GFP-GABARAP (Fig S2  C-D). Interestingly, neither mutant significantly affected the interaction of PLEKHM1 with either Rab7 or GABARAP. Moreover, overexpression of PLEKHM1-WT, 2A or 2D mutants did not have a dominant effect on LC3B lipidation or p62/SQSTM1 turnover, when overexpressed in HEK293T cells (Fig. S2E-G). Taken together, we show that the Ser432/435 sites are phosphorylated by mTOR are sensitive to conditions that alter mTOR activity and have an, as yet, undefined effect on PLEKHM1 function.

PLEKHM1 Ser432/Ser435 is MAPK target sites
The sites we identified as being phosphorylated by mTOR conform to the proline-directed mTOR consensus motif [38]. When we further inspected the phosphomotif within PLEKMH1, we realized that the same sites also conform to the PxSP MAPK-directed phosphorylation motif (Fig. 4A) [39,40]. This prompted us to explore whether these sites were also possible substrates of the MAPK pathway. Therefore, we tested whether PLEKHM1 phosphorylation could be induced in a growth factor/MAPK-dependent manner. MEFs expressing either vector only or HA-tagged PLEKHM1 were grown in 1% serum-containing medium for 16 h and subsequently stimulated with either full medium (FM) or with 50 ng/mL epidermal growth factor (EGF) for 0, 30, 60, 120 and 180 min. In the latter case, PLEKHM1 phosphorylation peak coincided with the peak in the activation of the MAPK pathway, as reflected in the levels of phosphorylated ERK (phospho-ERK1/2) after a 30-minute stimulation (Fig. 4B). As the sites conformed to the MAPK consensus sequence and phosphorylation of PLEKHM1 followed that of ERK1/2 activation (Fig. 4B), we next tested whether PLEKHM1 could be directly phosphorylated by MAPK1 (ERK2). Using GST-PLEKHM1 (418-456, Fig. 3F), we incubated recombinant MKK1 (MEK) alone, MAPK1 WT alone, MKK plus MAPK1 and MKK1 plus kinase dead MAPK1 (D168A), in the presence or absence of ATP. Under these conditions, PLEKHM1 is directly phosphorylated by MAPK1 at Ser432/Ser435 only in the presence of ATP (Fig. 4C, lanes 3 and 5), but not by MKK1 alone (Fig. 4C, lane 1) or by kinase dead MAPK (Fig. 4C, lanes 6 and 7). Next, we wanted to test the specificity of such response by applying a number of different growth factor conditions in combination with the inhibition of MAPK and mTOR pathways and subsequently probing the levels of phosphorylated PLEKHM1 at the phosphospecific sites. MEFs expressing HA-tagged PLEKHM1 were grown overnight in 1% serum and stimulated in the presence or absence of 10 µM U0126 or 250 nM torin 1 with either EGF (epidermal growth factor), FGF2 (fibroblast growth factor 2), IGF-1 (Insulin-like growth factor 1), PDGF-AA (Platelet-derived growth factor-AA), PDGF-BB or PDGF-AB for 30 min. Using phospho-ERK1/2 (Thr202/Tyr204) and phospho-p70S6K (T389), as a read-out MAPK and mTOR activity, respectively, we observed a robust phosphorylation (>2-fold increase in phospho-PLEKHM1 compared with unstimulated) of PLEKHM1 at Ser432/435 treated with EGF, FGF2 and PDGF-AB (Fig. 4D). FGF2-mediated PLEKHM1 phosphorylation was partially (approx. 40%) sensitive to MAPK inhibition, but not to mTOR inhibition (Figs 4D and S3). Next, we wanted to assess the relative contributions of the mTOR and MAPK pathways to Ser432/Ser435 phosphorylation. Stimulation of MEFs containing HA-PLEKHM1 with full medium resulted in a robust phosphorylation at Ser432/Ser435, while treatment with the mTOR inhibitor torin 1 resulted in an approximate 50% decrease (Fig. 4E, F). Treatment with U0126 resulted in a 60% decrease, and inhibition of both MAPK and mTOR resulted in an almost complete loss of PLEKHM1 phosphorylation, similar to that observed after a 4-h starvation (EBSS medium; Fig. 4E, F). This indicates that both the mTOR and the MAPK pathways can phosphorylate directly PLEKHM1 at Ser432/Ser435, pointing to a convergence of pathways for a function that has yet to be determined.

Discussion
The late endocytic adaptor protein PLEKHM1, which is involved in osteoclast vesicle transport [41] through an interaction with Rab7 [24][25][26], plays an important role in the maturation of late endosomes [24,26,27], biogenesis of the Salmonella-containing vacuole [25] and fusion of autophagosomes and lysosomes [26,37,42]. How these distinct, yet related, functions are regulated was unclear. Herein, we discover post-translational modification of human PLEKHM1 in a highly conserved region spanning amino acids 420-450. Phosphorylation of serine residues in a proline-directed manner is carried out directly by mTOR kinase, a binding partner of PLEKHM1. Interestingly, it has recently been shown that mTORC2 complex in C. elegans influences autophagosome biogenesis [43]. Since we also found mTORC2 components coprecipitating with PLEKHM1 (Fig. 1A,B), we cannot rule out a role for mTORC2-mediated PLEKHM1 phosphorylation at S432/S435. We additionally show that the very same sites that are targeted by mTOR are also targets for the MAPK pathway; hence, the inhibition of both MAPK and mTOR results in a complete loss of phosphorylated PLEKHM1 at these sites. This indicates that PLEKHM1 functions at the crossroads of these two signalling pathways and potentially integrates them into unique cellular response reflected in specific dynamics of autophagosome-lysosome fusion (mTOR pathway) or growth factor receptor degradation (MAPK pathway), depending on cell type and function, as well as on physiological demands. Regulation of adaptor and receptor proteins in both autophagy and the endocytic pathway is an emerging topic in how cells fine-tune their responses to stress, and current data are highly supportive of phosphorylation as an important regulatory mechanism. For example, during selective autophagy, the receptor proteins optineurin and p62/SQSTM1 can both target ubiquitinated Salmonella for destruction by the autophagosome-to-lysosome pathway [44,45]. Interestingly, both are phosphorylated by TBK1 to either enhance MAP1LC3B interaction [45,46], enhance the interaction with ubiquitinated Salmonella [47,48] or increased activation of oxidative stress pathways [49]. This modifiable switch helps in the mobilization and activation of receptor activity, rendering it more effective in the clearance of intracellular pathogens. In addition to autophagy receptors, adaptor proteins that control formation, trafficking and fusion of autophagosomes can also be regulated in a similar manner. One such example is protein Rubicon, which shares a high degree of homology with PLEKHM1, is phosphorylated by HUNK at two serine residues in the region that binds Vps34 and regulates the autophagy-suppressive function of Rubicon [50]. PLEKHM1 is phosphorylated at Ser432/Ser435 under normal growth conditions, but becomes rapidly dephosphorylated under starvation conditions or upon chemically induced inhibition of mTOR. As part of our studies, we used doxycycline-inducible system for the induction of overexpressed PLEKHM1. Although we saw very little change in mTOR activity after doxycycline addition, any potential metabolic changes caused by the doxycycline may need to be explored in future studies [51]. Potentially, there could be rapidly acting phosphatases that change the dynamic pool of phosphorylated PLEKHM1 that could affect the dynamics of PLEKHM1-mediated fusion events in the cells depending on the cellular demand. This pointed towards high sensitivity of PLEKHM1 to the activity of mTOR and physiological state of the cell.
In this study, our proteomics data correlates with previous reports of multiple deep-sequence mass spectrometry analyses (www.phosphosite.org). The phosphosites are located within a highly conserved unstructured region of PLEKHM1 and our data suggests an involvement of major metabolic kinases in regulation of PLEKMH1 (Fig. 2H). In order to study this regulation in more detail, we have developed several tools to study these mechanisms, including a phospho-specific antibody that detects a double-serine phosphorylation motif on PLEKHM1 (S432/S435). In combination with CRISPR/ Cas9-mediated endogenous tagging of PLEKHM1 (Fig-ure S1 and 1C), these tools will enable us to challenge the cells with broader set of chemical compounds and directly test both post-translational modifications, as well as interacting partners of PLEKHM1 in a systematic top-down approach. Further, combining such approaches will allow us, and others, to accurately detect and quantify levels and spatial organization of phosphorylated PLEKHM1 in various cell types and stress conditions. This will help expand our understanding of the interconnectivity between autophagy and growth factor receptor signalling.
Similar to the examples of other autophagy adaptors, our data suggest that the function of phosphosites in PLEKHM1 could be regulation of interactions of PLEKHM1 with components of the lysosome. Under conditions of starvation or inhibition of mTOR, mTOR dissociates from LAMP2-positive structures, but PLEKHM1 remains associated (Fig. 1E). Interestingly, mutation of PLEKHM1 at Ser432/435 did not affect the known interaction with GABARAP (Fig. S2A,B) [24,25,41] or Rab7 ( Fig. S2C-D) [24,25,41] and so potentially does not function in these distal interactions. However, we cannot rule out any changes in the interaction with HOPS/SNARE complex [26] or the recently identified interaction with Arl8b that was mapped to the RUN domain [27], which is in close proximity to phosphoserine Ser432/Ser435 site. These phosphosites may, therefore, act as on/off switch for PLEKHM1 function during conditions such as cellular stress. In addition, we also showed that these sites are not only a direct target for mTOR, but that they also share similarities with a MAPK PxSP motif and are sensitive to MAPK inhibition alone. Indeed, phosphorylation of PLEKHM1 at Ser432/Ser435 was induced by certain growth factors via both MAPK-dependent (EGF, FGF2, IGF-1) and MAPK-independent (PDGF-BB, PDGF-AB) manner, indicating compartmentalization of PLEKHM1 function. These data firmly position PLEKHM1 regulation at the crossroads of two major growth and proliferative signal transduction pathways and, given our previous findings showing that PLEKHM1 is required for both autophagosome and EGFR degradation [26], our results shed some light on possible mechanisms that may govern its activity. These early findings will help pave a way to dissect how these signalling pathways converge and regulate the function of this adaptor protein at the molecular level.