The small GTPase Rab32 resides on lysosomes to regulate mTORC1 signaling

The small Rab GTPases comprise a family of over 60 proteins in humans that are recognized as key members of the vesicular trafficking machinery. Specific Rab proteins associate with respective intracellular compartments where they function as ‘molecular switches’ by cycling between GTP-bound (ON) or GDP-bound (OFF) states to support cellular endocytosis, exocytosis, intracellular vesicle and protein trafficking, and cytoskeletal dynamics (Stenmark, 2009; Zhen and Stenmark, 2015). This is particularly important for metabolism in epithelial cells, such as the liver hepatocytes that utilize this machinery to regulate lipid catabolism, nutrient uptake and secretion as well as detoxification of the blood (Schulze et al., 2019). Thus, it is not surprising that disruption of normal function of several Rab proteins is implicated in a range of metabolic diseases and cancer (Li, 2011; Li et al., 2016b; Schroeder et al., 2015; Schulze et al., 2017a,b).

In addition to Rab proteins, hepatocytes and other epithelial cells utilize an ancient signaling cascade that is highly conserved in all eukaryotes, and is considered to be a master regulator of cellular metabolism (Saxton and Sabatini, 2017). This mechanistic target of rapamycin complex 1 (mTORC1) pathway is uniquely positioned to receive inputs from nutrients, growth factors and stress, and responds accordingly to support cell growth and metabolism, whereas a related – but functionally distinct – mTORC2 integrates growth factor signaling to support cytoskeletal dynamics and cell survival (Kim and Guan, 2019). Highly metabolic organs, such as the liver, require mTORC1 signaling to support growth and regeneration (Fausto et al., 2006; Fouraschen et al., 2013; Sengupta et al., 2010), while a deregulated mTORC1 pathway has been implicated in a variety of human pathologies, such as cancers, diabetes, obesity and neurodegeneration (Laplante and Sabatini, 2012; Matter et al., 2014).

Interestingly, several Rab proteins have been found to regulate the mTORC1 signaling pathway. A genetic screen performed in Drosophila S2 cells revealed that knockdown of Rab5, Rab11 and Rab1 decreases phosphorylation of the well-defined mTORC1 substrate p70 ribosomal protein S6 kinase (RPS6KB1, hereafter referred to as S6K) (Li et al., 2010). Additionally, overexpression of constitutively active GTP-bound forms of Rab5 and Rab7 inactivates mTORC1 signaling, perhaps indirectly due to a disruption of endocytic trafficking (Flinn et al., 2010; Li et al., 2010). Rab1A, which supports the vesicular trafficking pathway from the endoplasmic reticulum (ER) to Golgi, is frequently overexpressed in colorectal cancer and has also been shown to activate the mTORC1 pathway (Thomas et al., 2014). Specifically, amino acids stimulate Rab1A activity and its binding with mTORC1, which then promotes mTORC1 interaction with its activator Rheb present on the Golgi apparatus, ultimately leading to increased mTORC1 activity. The contribution of other Rab GTPases to mTORC1 activity across different cells and tissue types remains poorly understood.

Rab32 was originally found to localize to mitochondria in the fibroblast-like WI-58 and COS-7 cells and, subsequently, became known for promoting the biogenesis of lysosome-related organelles, particularly pigment granule and/or melanosome biogenesis in melanocytes (Alto et al., 2002; Bultema et al., 2014; Wasmeier et al., 2006). In addition, Rab32 can associate with multiple metabolic organelles in epithelial cells, such as the ER and mitochondria-associated membranes (MAMs), where it is proposed to function as a cAMP-dependent protein kinase A-anchoring protein (AKAP), and to regulate mitochondrial dynamics and apoptosis (Bui et al., 2010). A different study suggested that its localization at the ER also supports autophagic membrane formation and autophagy under basal, nutrient-rich conditions (Hirota and Tanaka, 2009). Interestingly, the role of Rab32 in autophagy appears to be conserved in the Drosophila fat body, where it localizes to lysosomes and autophagosomes, and regulates lipid storage (Wang et al., 2012). Similarly, in hepatocytes, depletion of Rab32 results in decreased lipid droplet content, suggesting that Rab32 also regulates hepatic lipid metabolism (Li et al., 2016a). Overall, this diverse subcellular localization of Rab32 suggests that it provides a physiological link between metabolic organelles; however, its function in cellular growth and energy metabolism remains poorly understood.

In this study, we identified that the localization of Rab32 to lysosomes is particularly prevalent in primary rat hepatocytes and Hep3B human hepatoma cells. By using an unbiased reverse phase protein array screen, we found the metabolic mTORC1 signaling pathway to be markedly altered in the absence of Rab32. Subsequently, western blot analysis showed that knockdown of Rab32 inhibits both basal and amino acid-induced activation of mTORC1 signaling to S6K and, to a lesser extent, unc-51-like autophagy activating kinase 1 (ULK1). Consistent with attenuation of mTORC1 activity, we observed that Rab32 knockdown reduced cell proliferation and cell size, together with increased nuclear localization of transcription factor EB (TFEB) and lysosome biogenesis. Mechanistically, Rab32 appeared to interact with mTOR kinase in a unique GTP-independent manner. Finally, by using biochemically isolated lysosome fractions, we demonstrated a reduction in the lysosome association of mTOR, regulatory-associated protein of mTOR (RPTOR, hereafter referred to as Raptor) and mTORC1 pathway proteins, such as RagC and Lamtor1, following knockdown of Rab32, suggesting that Rab32 functions to regulate mTOR trafficking to lysosomes. Together, these findings suggest that Rab32 acts as a novel regulator of cell metabolism and growth through the promotion of mTORC1 activity.

Rab32 has been previously reported to localize to a number of metabolic organelles, such as the ER, mitochondria and lysosomes and/or autophagosomes in COS1 and COS7 and HeLa cells, as well as pigment granules in melanocytes (Alto et al., 2002; Bui et al., 2010; Hirota and Tanaka, 2009). To further define the role of Rab32 in more-differentiated epithelial cells, such as hepatocytes, we first sought to analyze its subcellular localization in both primary rat hepatocytes and Hep3B human hepatoma cells. As shown in Fig. 1, GFP-tagged wild-type (WT) Rab32 displayed a prominent localization around lysotracker-positive acidic vesicles, suggestive of late endosomes/lysosomes. Similar localization was observed with FLAG-tagged Rab32, and with Lamp2A- or Lamp1-labeled lysosomes in primary rat hepatocytes or Hep3B cells, respectively. Both WT and GTP-bound (Q85L) Rab32 forms exhibited similar late endosome and/or lysosome associations, whereas GDP-bound (T39N) Rab32 showed reduced localization to these organelles. In support of previous studies (Alto et al., 2002; Bui et al., 2010), Rab32 was also observed to localize to mitochondria and the ER in a subset of Hep3B cells, but this was rarely observed in primary rat hepatocytes (data not shown). Altogether, Rab32 distributes to various metabolic organelles but appears most prominent on acidic vesicles, such as late endosomes and lysosomes in hepatocytes.

To gain a better understanding of the role that Rab32 has in epithelial cell function, we conducted an unbiased reverse phase protein array (RPPA) to screen for changes in protein signaling pathways involving cell growth, proliferation, metastasis, metabolism and apoptosis. Hep3B cells were subjected to control or Rab32 siRNA treatment for 72 h, followed by cell lysis, and analysis of changes in the levels of total or phosphorylated protein by using a ∼220 antibody-based array (Chang et al., 2015). This assay revealed that several prominent signaling proteins related to the mTOR pathway are attenuated in response to Rab32 knockdown (Table S1). Included on this list are upstream regulators of mTORC1, such as phosphorylated Akt and proline-rich Akt substrate of 40 kDa (PRAS40, officially known as AKT1S1), mTOR itself as well as downstream mTORC1 targets, such as S6K (Fig. 2A) (Dibble and Cantley, 2015). To further assess the impact Rab32 depletion has on mTORC1 signaling, we analyzed the phosphorylation of the individual downstream mTORC1 substrates by western blot analysis. As shown in Fig. 2B, Rab32-depleted Hep3B cells exhibited a ∼50% decrease in the phosphorylation of the mTORC1 substrate S6K and its downstream substrate ribosomal protein S6. Interestingly, a significant impact on the phosphorylation of the mTORC1 substrate eukaryotic translation initiation factor 4E-binding protein 1 (EIF4EBP1, hereafter referred to as 4EBP1) was not observed relative to total 4EBP1 amounts in the Hep3B cell model (Fig. 2B). These effects were confirmed across four individual Rab32 siRNAs (Fig. S1). mTORC1 is also known to phosphorylate the autophagy regulator ULK1 at Ser757 (Kim et al., 2011). Phosphorylation of ULK1 at this residue appeared modestly reduced relative to total ULK1 following knockdown of Rab32 (Fig. 2B and Fig. S1). Similar inhibition of mTORC1 was also observed in HeLa cells upon Rab32 depletion (Fig. S2A). These results show that loss of Rab32 affects phosphorylation of mTORC1 substrates – with the most dramatic effects on S6K and S6.

S6K is known to regulate protein synthesis and cell size through phosphorylation of S6 as well as activate the transcription factor sterol regulatory element binding protein 1 (SREBF1, hereafter referred to as SREBP1) to express genes involved in fatty acid, triglyceride and cholesterol synthesis (Düvel et al., 2010; Owen et al., 2012). Given the impact of Rab32 on S6K activity described above, we assessed the activation of SREBP1 in control or Rab32-depleted cells as an additional downstream read-out of S6K activity. Indeed, as seen in the luciferase assay shown in Fig. 2C, transcription of the SREBP1 target promoter sequence of fatty acid synthase (FASN) was reduced by a significant 70% in Rab32-depleted cells.

As Rab32 appears to localize on lysosomes and play a role in the regulation of mTORC1 signaling, we further predicted that Rab32 depletion would impact cell proliferation and size – properties known to be physiologically driven by mTORC1. First, Hep3B cell proliferation and viability was diminished nearly 50% following Rab32 knockdown, as measured by a standard crystal violet assay compared to control cells (Fig. 2D). This finding was further confirmed by an alternative colorimetric MTS proliferation assay, which revealed a ∼40% loss in cell proliferation and/or viability after Rab32 depletion (Fig. 2E). Additionally, Rab32-depleted Hep3B and HeLa cells exhibited a dramatic decrease in the average 2D cell area on glass coverslips, and also in the average cell diameter of trypsinized cells in suspension compared to control cells (Fig. 2F,G and Fig. S2B,C). Collectively, these results indicate that loss of Rab32 decreases cell proliferation, viability and cell size, consistent with an attenuation of mTORC1 signaling.

It is well-documented that amino acids are crucial for mTORC1 activation and signaling (Saxton and Sabatini, 2017). During periods of starvation when the supply of amino acids is low, mTORC1 is inactive and thought to reside primarily within the cytosol. However, upon amino acid replenishment, mTOR is rapidly recruited to the lysosomal surface by heterodimeric Rag GTPases – within a matter of minutes – where it becomes activated by Rheb GTPase (Sancak et al., 2010, 2008). To determine whether Rab32 is necessary for amino acid-induced activation of mTORC1, Rab32-depleted cells were starved in amino acid-free medium for 50 min followed by amino acid re-feeding for an additional 15 min. In cells treated with control siRNA, re-addition of amino acids resulted in a marked (10-fold) increase of mTORC1 activity, as assessed by phosphorylation of its downstream targets S6K and S6. However, in Rab32-depleted cells, phosphorylation of S6K was inhibited dramatically whereas phosphorylation of S6 was nearly abolished after amino acid re-addition (Fig. 3A,B). Phosphorylation of ULK1 was not significantly altered in Rab32-depleted Hep3B cells upon amino acid re-addition (Fig. 3A,B). These experiments produced similar findings in HeLa cells (Fig. S2D).

Because Rab32 knockdown appeared to diminish amino acid-induced mTORC1 activation, we predicted that overexpression of Rab32 would potentiate mTORC1 signaling either in the presence or absence of amino acid manipulation. As seen in Fig. S3, overexpression of Rab32 WT, Q85L or T39N did not significantly increase mTORC1 signaling after stimulation with amino acids. Overall, these data suggest that Rab32 is necessary but not sufficient for mTORC1 activation following stimulation with amino acids.

In addition to stimulation with amino acids, mTORC1 can also be activated by cell culture serum supplemented with growth factors that activate the PI3K-Akt pathway (Manning et al., 2002; Menon et al., 2014). We therefore asked whether Rab32 participates in mTORC1 activation in response to serum and/or growth factor stimulation. To this end, cells were placed in serum-free HBSS for 1 h, then stimulated with 10% FBS-containing growth medium for 15 min and analyzed by western blotting. As seen in Fig. 3C,D, Rab32 depletion significantly inhibited phosphorylation of S6 but not of S6K or ULK1 relative to each protein in total. Overall, these data suggest that Rab32 regulates mTORC1 activity following stimulation with amino acids and, perhaps to a lesser extent, growth factors.

mTORC1 activity has been shown to play a critical inhibitory role in lysosome biogenesis through phosphorylation of transcription factor EB (TFEB), a key transcriptional regulator of numerous lysosomal and autophagy genes. As shown in several studies, phosphorylation of TFEB by mTORC1 promotes TFEB retention in the cytoplasm, thereby preventing its transcriptional activity (Martina et al., 2012; Napolitano and Ballabio, 2016; Sardiello et al., 2009; Settembre et al., 2011). By contrast, inhibition of mTORC1 leads to TFEB dephosphorylation and translocation to the nucleus to promote transcription of genes related to lysosome biogenesis and autophagy (Roczniak-Ferguson et al., 2012; Settembre et al., 2012). Considering that Rab32 depletion inhibits mTORC1 activity, we predicted that this would also lead to increased nuclear TFEB, lysosome biogenesis and autophagy. To address this question, GFP-tagged TFEB was expressed in control and Rab32-depleted Hep3B and HeLa cells, and the percentage of cells with nuclear TFEB localization was quantified. Interestingly, a nearly 2-fold increase in the frequency of nuclear TFEB localization was observed following Rab32 knockdown (Fig. 4A-C and Fig. S4A), a result consistent with reduced mTORC1 activity in cells lacking Rab32.

To further assess the impact of Rab32 on lysosome biogenesis, the number and size of LysoTracker-stained lysosomes was measured together with the relative acidity of these vesicles by using the pH-sensitive LysoSensor dye. Consistent with an increase in nuclear TFEB localization, Rab32-depleted Hep3B cells exhibited an increase in lysosome number per cell, with no significant effect on lysosome size or relative acidity (Fig. 4D,E). Western blot analysis showed that an increase in lysosome protein levels [Lamp1, Lamp2A and cathepsin D (CTSD)] was also observed following Rab32 knockdown (Fig. 4F,G). Rab32 knockdown in HeLa cells caused a significant increase in Lamp1 protein levels, although no significant effect was observed on lysosome number per cell area (Fig. S4B-E).

To confirm that Rab32 depletion did not affect lysosome function, we assessed the degradation of epidermal growth factor receptor (EGFR) following stimulation with its ligand, epidermal growth factor (EGF). As seen in Fig. 4H,I and Fig. S4F,G, Rab32 knockdown did not diminish EGFR degradation compared to controls in both Hep3B and HeLa cells. In fact, EGFR degradation was slightly accelerated in the absence of Rab32 at t=60 min in Hep3B cells (Fig. 4I). Overall, the above data further support the concept that loss of Rab32 inhibits mTORC1 signaling that, in turn, induces nuclear TFEB translocation and lysosome biogenesis. This effect appears to contrast the role of Rab32 in pigment-producing melanocytes, which show reduced biogenesis of the lysosome-related organelles (i.e. melanosomes) following Rab32 depletion.

Inhibition of mTORC1 signaling is well known to activate autophagy to degrade cytosolic materials through lysosomal enzymes (Rabinowitz and White, 2010). On the basis of this, we predicted that this ‘self-eating’ process is enhanced following Rab32 knockdown. To test the effect of Rab32 knockdown on autophagic flux, cells were treated with the lysosome inhibitor Bafilomycin A1 (BafA1, 100 nM, 2 h) to measure the relative increase in the accumulation of autophagosomes (increase in LC3-II levels) and the autophagy substrate p62 by western blot analysis. Under basal conditions (medium supplemented with 10% FBS) Rab32 knockdown modestly decreased LC3-II and p62 accumulation after treatment with BafA1, suggesting a perturbation in autophagic flux (Fig. 5A,B). In response to treatment with the potent mTOR inhibitor and autophagy activator Torin1 for 2 h, autophagic flux was similarly perturbed by Rab32 knockdown (Fig. 5A,C). The effect of Rab32 knockdown on autophagy was also measured using the RFP-GFP-LC3 fluorescence reporter. This probe indicates neutral-pH autophagosomes as yellow (combined RFP and GFP fluorescence). Upon autophagosome fusion with an acidic lysosome, the GFP fluorescence becomes quenched and appears as red-only puncta indicating a progression of the autophagic process. As seen in Fig. 5D,E, Rab32 knockdown caused a dramatic decrease in autophagosome number per cell (yellow puncta), with a modest increase in autolysosomes (red puncta) indicating that loss of Rab32 perturbs autophagosome formation. In HeLa cells, Rab32 knockdown caused an even more dramatic perturbation in autophagic flux, as assessed by western blot and fluorescence of the RFP-GFP-LC3 reporter (see Fig. S5). Together these data suggest that Rab32 knockdown perturbs autophagic flux, perhaps by attenuating autophagosome biogenesis. This result is surprising given that mTORC1 is inhibited following Rab32 knockdown but is nonetheless consistent with a previous study that suggests a role for Rab32 in autophagosome biogenesis (Hirota and Tanaka, 2009).

To gain a more detailed mechanistic understanding of mTORC1 regulation by Rab32, we assessed the subcellular localization of mTORC1 in relation to the lysosomal surface. Recent studies suggest that mTORC1 activation occurs at the lysosome in response to nutrient signaling (Manifava et al., 2016; Sancak et al., 2010, 2008). Because Rab32 is present on lysosomes, we hypothesized that it colocalizes and associates with mTORC1 on these organelles. To test this, FLAG-tagged Rab32 was expressed in Hep3B cells together with GFP-tagged Lamp1 to label lysosomes; cells were then co-stained with an mTOR antibody. Notably, FLAG-Rab32 colocalized with endogenous mTOR on Lamp1-positive lysosomes in Hep3B cells, as shown in Fig. 6A. Additionally, colocalization between GFP-tagged Rab32, Lamp1-mCherry and mTOR was readily observed in primary rat hepatocytes (Fig. 6A). This result prompted us to test whether Rab32 associates with either mTOR kinase or any other mTOR complex components, using biochemical methods, such as GFP-Trap pulldown approach. Hep3B cells expressing GFP-Rab32 were lysed and incubated with GFP-binding agarose beads to pull down GFP vector control or GFP-tagged Rab32 and analyzed for binding to mTOR complex components. As depicted in Fig. 6B, pulldown of Rab32 revealed an association with mTOR as well as the mTORC1/2 subunit mLST8, but not with the mTORC1-specific subunit Raptor or the mTORC2-specific subunit rapamycin-insensitive companion of mammalian target of rapamycin (Rictor) under the conditions tested. Reciprocal coimmunoprecipitation of endogenous mTOR confirmed an interaction with FLAG-Rab32 together with the mTORC1/2 subunit mLST8 (Fig. 6C). To test whether this interaction is dependent upon the GTP-bound state of Rab32, we performed GST-Rab32 pulldown experiments using WT, Q85L and T39N Rab32 variants from Hep3B and HeLa cell lysates. The capacity of Rab32 to bind to a known effector (VARP) was used as a positive control (Tamura et al., 2009). Surprisingly, similar levels of endogenous mTOR were observed in pulldown assays using all the GST-Rab32 variants, whereas, as expected, only WT and QL Rab32 were associated with the effector VARP (Fig. 6D,E and Fig. S6A,B). Together, these findings suggest that Rab32 can associate with mTOR regardless of its nucleotide bound status.

As mTOR trafficking to lysosomes is crucial for its subsequent activation and signaling (Sancak et al., 2010), we reasoned that Rab32 may regulate mTOR association with lysosomes. To test this, lysosomes were biochemically isolated using an immunoprecipitation-based pulldown of the HA-tagged transmembrane lysosome protein TMEM192 as described previously (Abu-Remaileh et al., 2017). Hep3B cells were depleted of endogenous Rab32 using siRNA treatment and transfected with TMEM192-HA followed by cell homogenization and pulldown of intact lysosomes using magnetic anti-HA beads. Importantly, lysosome-associated mTOR levels were significantly reduced (∼50%) in Lamp1-enriched lysosome fractions gleaned from the Rab32-depleted cells compared to control cells (Fig. 7A,B). In addition, the mTORC1 subunits Raptor, RagC and Lamtor1 were also present on lysosomes, although at levels that were lower than in control cells. Similar Lyso-IP results were observed in HeLa cells (Fig. S6C). Together, these data suggest that loss of Rab32 impairs mTORC1 association with lysosomes.

Given that Rab32 knockdown appears to inhibit mTORC1 signaling and lysosomal association, we reasoned that mTORC1 activity can be rescued by overexpression of active Rag heterodimers that drive mTORC1 to the lysosome (Sancak et al., 2010, 2008). As seen in Fig. 7C,D, expression of RagB^GTP/RagC^GDP heterodimers in Rab32-depleted Hep3B cells was sufficient to rescue phosphorylation of S6K and S6 to that of control levels after stimulation with amino acids (Fig. 7C,D). Similar effects were observed in HeLa cells (Fig. S6D,E). Overall, our data support a model whereby loss of Rab32 reduces lysosomal mTORC1 levels and thus attenuates mTORC1 activity (Fig. 7E).

In this study we report that Rab32 localizes to lysosomes (Fig. 1) and supports robust basal and amino acid-induced mTORC1 phosphorylation of S6K and S6 substrates (Figs 2 and 3, Figs S1, S2). Loss of Rab32 inhibits mTORC1 signaling and reduces cell proliferation, viability and cell size (Fig. 2, Fig. S2). In agreement with mTORC1 inhibition, loss of Rab32 also promotes nuclear TFEB localization and lysosome biogenesis (Fig. 4, Fig. S4). We made two unexpected observations in this study. First, autophagy is also impaired in Rab32-depleted cells, perhaps due to reduced autophagosome biogenesis (Fig. 5, Fig. S5). Second, Rab32 interacts with mTOR in a GTP/GDP-independent manner and promotes mTOR association with lysosomes (Figs 6 and 7, Fig. S6). Taken together, these findings highlight a novel role for the small GTPase Rab32 in regulating growth and metabolism of epithelial cells via the mTORC1 signaling pathway.

The results presented here support a model whereby mTORC1 activity and association with lysosomes is dependent upon functional Rab32 (Fig. 7E). Indeed, other studies have demonstrated that activity of mTORC1 depends on its recruitment to lysosomes (by Rag GTPases), where it is subsequently activated (by Rheb GTPase) in response to nutrients and growth factors (Sancak et al., 2010, 2008). The finding that mTORC1 activity is restored in Rab32 knockdown cells upon overexpression of active Rag GTPases (Fig. 7C,D and Fig. S6D,E) is consistent with the premise that loss of Rab32 inhibits mTORC1 activation due to reduced mTORC1-lysosomal recruitment. It will be important to identify the binding motifs that enable binding between Rab32 and mTOR to better understand whether Rab32 directly regulates recruitment of mTOR and its attachment to the lysosome surface. It is worth noting that no differences were observed in the distribution of Rab32 to lysosomes in response to stimulation with amino acids – as was observed for mTORC1 (data not shown). Therefore, Rab32 might function as a ‘docking site’ for mTORC1 on the lysosomal surface or might affect the lysosomal proteome more generally. Indeed, Rab32 knockdown decreased lysosomal association of RagC and Lamtor1, which could also partially explain reduced lysosomal mTORC1 levels by Lyso-IP (Fig. 7A,B and Fig. S6C). Future studies will be required to assess how Rab32 depletion impacts on the lysosome proteome versus specific mTORC1-related components.

As described in Figs 2, 3 and Fig. S2, it was somewhat surprising to find that phosphorylation of S6K and S6 is strongly attenuated in the absence of Rab32, with little or no inhibition observed over phosphorylated ULK1 or 4EBP1 substrates, respectively. One possibility is that Rab32 regulates mTORC1 substrate specificity. However, it is perhaps more likely that different mTORC1 substrates (S6K, ULK1 and 4EBP1) exhibit different sensitivities to changes in mTORC1 kinase activity. For example, treatment with rapamycin was shown to readily inhibit S6K phosphorylation in most cell types, whereas phosphorylation of 4EBP1 was not affected (Choo et al., 2008; Qin et al., 2016). In line with this, Kang et al. identified that S6K is a weak mTORC1 substrate, whereas both ULK1 and 4EBP1 are phosphorylated with much higher affinity by mTORC1 (Kang et al., 2013). The authors proposed that the mTORC1 affinity to its substrate explains why treatment with rapamycin inhibits phosphorylation of S6K, while mTORC1-mediated phosphorylation of ULK1 or 4EBP1 is not readily inhibited under the same conditions. Hence, the effect of Rab32 knockdown on S6K and S6 appears to mimic effects that are similar to those seen after treatment with rapamycin.

The results presented here suggest that, whereas WT and GTP-bound Rab32 appear to localize to lysosomes at a higher frequency than the GDP-bound form (Fig. 1), the GTPase activity of Rab32 is not required for mTOR binding. The findings from the GST-pulldown experiments indicated that all Rab32 forms display equal binding capacity to endogenous mTOR, whereas the known Rab32 effector VARP only interacts with WT and GTP-bound forms (Fig. 6D,E and Fig. S6A,B). These observations suggest that Rab32 associates with mTOR kinase independent of its nucleotide-bound status, a contrast compared with many traditional Rab-effector interactions (Zhen and Stenmark, 2015). It should be noted, however, that some GTPase-independent functions have been described previously. For example, the Rho GTPase Rac1 acts independently of its GTPase activity as an adaptor protein for mTOR and regulates cellular membrane associations of both mTORC1 and mTORC2 complexes (Saci et al., 2011). Rab21 has also been implicated in binding to integrin receptors regardless of its GTP/GDP nucleotide status, whereas other Rab proteins, such as Rab27a and Rab11, have been reported to associate with interacting partners specifically in their GDP-bound states (Kimura et al., 2008; Mai et al., 2011; Shirane and Nakayama, 2006). Since the GDP-bound T39N form of Rab32 is not frequently observed at the lysosome surface, it is possible that it can interact with mTOR on different organelles. Indeed, both mTOR complexes and their substrates have been suggested to reside in additional organelles besides the lysosome (Betz and Hall, 2013; Betz et al., 2013; Thomas et al., 2014).

Results from the current study suggest that Rab32 knockdown increases lysosome biogenesis, perhaps by inhibiting mTORC1. Indeed, loss of Rab32 increased nuclear localization of TFEB, lysosome numbers and lysosome protein levels, which is consistent with inhibition of mTORC1 signaling (Fig. 4, Fig. S4). This observation is interesting, given that Rab32 and its closest family member Rab38 have been previously reported to play tissue-specific and functionally redundant roles in the biogenesis of melanosomes (lysosome-related organelles) (Bultema et al., 2014; Tamura et al., 2009; Wasmeier et al., 2006). In fact, Rab32 and Rab38, and their effector VARP are crucial for cargo sorting during melanosome maturation, as disruption of Rab32, Rab38 or VARP activity causes mistrafficking of melanin-synthesizing enzymes as well as defective melanogenesis (Marks et al., 2013). Thus, whereas Rab32 knockdown in melanocytes inhibits melanosome biogenesis, Rab32 knockdown in other epithelial cells appears to promote lysosome biogenesis through inhibition of mTORC1.

As a master regulator of cell growth and metabolism, mTORC1 also integrates lysosomal nutrient sensing to monitor and balance anabolic and catabolic processes. Under nutrient-replete conditions, autophagy progression is inhibited due to phosphorylation of the ULK1 complex by mTORC1. However, attenuation of mTORC1 signaling during periods of nutrient starvation or stress can lead to autophagosome formation and autophagy activation (Kim et al., 2011; Rabanal-Ruiz et al., 2017). We were surprised to find that depletion of endogenous Rab32 perturbed autophagosome formation even though mTORC1 signaling was also inhibited (Fig. 5 and Fig. S5). However, the effect Rab32 knockdown has on ULK1 phosphorylation was modest compared to its effect on phosphorylation of S6K and S6. Thus, residual ULK1 phosphorylation by mTORC1 might contribute to autophagy inhibition in Rab32-depleted cells, especially given the apparent increase in total ULK1 protein levels (∼2.5-fold, data not shown). In addition, Rab32 might affect autophagosome biogenesis directly, consistent with previous studies (Hirota and Tanaka, 2009; Matsui and Fukuda, 2013). Similar effects on mTORC1 and autophagy have also been reported for other Rab GTPases. For example, Rab1A has been proposed to activate both mTORC1 signaling in response to amino acid signaling at the Golgi as well as the formation of autophagosome vesicles (Huang et al., 2011; Thomas et al., 2014; Winslow et al., 2010).

In summary, we have identified Rab32 as a lysosome component which acts as a novel regulator of cellular growth and metabolism through the regulation of mTORC1 signaling. Importantly, it appears that Rab32 is necessary for both basal and amino acid-induced activation of mTORC1 and phosphorylation of S6K and S6, as well as the regulation of nuclear TFEB localization and lysosome biogenesis. Supportive of these findings is that Rab32 interacts with mTOR kinase and facilitates mTOR association with lysosomes, which appears to be important for downstream mTORC1 signaling. Ultimately, it would be of interest to understand the physiological role of Rab32 regulation and perturbation of mTORC1, as they pertain to cellular metabolic processes, such as hepatic steatosis as well as regeneration and neoplastic growth.

Antibodies were used at 1:1000 for western blotting (WB) or 1:100 for immunofluorescence (IF) analysis unless otherwise specified. Antibodies against CoxIV (#4850), cathepsin D (#2284), EGFR (#2232), FLAG (#8146 for IF and #14793 for WB), GAPDH (#5174 at 1:5000), GβL/mLST8 (#3274), HA (#3724 at 1:10,000), Lamtor1 (#8975), LC3 (#2775), mTOR (#2983 at 1:200 for IF), p62 (#5114), p70 S6K (Thr389) (#9234), p70 S6K (#9202), RagC (#3360), Raptor (#2280), Rictor (#2114), S6 (Ser235/236) (#4858 at 1:5000), S6 (#2217 at 1:3000), TSC2 (#4308), ULK1 (Ser757) (#6888), ULK1 (#8054), 4EBP1 (Thr37/46) (#2855) and 4EBP1 (#9644) were from Cell Signaling (Danvers, MA). Antibodies against GFP (#sc-9996 at 1:2000), GST (#sc138 at 1:2000), Lamp1 (#sc-20011) and Rab32 (#sc-390206 at 1:500) were from Santa Cruz (Santa Cruz, CA). The anti-Raptor antibody (#42-4000) was from ThermoFisher (Rockford, IL). Antibodies against LAMP2A (#ab18528) and VARP (#108216) antibodies were from Abcam (San Francisco, CA). The anti-actin (#A2066 at 1:2000) antibody was from Sigma (St Louis, MO). For western blot analysis, HRP-conjugated goat anti-rabbit and anti-mouse antibodies (at 1:5000) were purchased from Invitrogen (Carlsbad, CA), which were detected using SuperSignal Pico or Femto substrates from ThermoFisher. The goat anti-rabbit and anti-mouse secondary antibodies conjugated to either Alexa-Fluor-488, Alexa-Fluor-594 or CY5 (at 1:500) and ProLong Antifade reagent used for IF staining were all obtained from ThermoFisher. Other reagents used were: LysoTracker Deep Red (#L12492) and LysoSensor Green DND-189 (#L7535) from ThermoFisher; Torin1 (#4247) from Tocris (Bristol, United Kingdom); Chaps (#220201) from EMD Millipore (Temecula, CA); all other reagents were obtained from Sigma unless otherwise specified.

siRNA oligonucleotides targeting human Rab32 (siGenome SMARTpool (#M-009920-02-0005) and individual oligonucleotides (#D-009920-01-0002, #D-009920-02-0002, #D-009920-03-0002, #D-009920-05-0002) or control non-targeting oligonucleotides were purchased from Dharmacon/Horizon Discovery Group (Lafayette, CO). siRab32 pool oligonucleotides were used throughout the paper unless otherwise specified.

The human WT, Q85L and T39N FLAG-Rab32 constructs were a kind gift from Dr John D. Scott (University of Washington, Seattle, WA). GFP-Rab32 constructs were cloned by amplifying Rab32 sequence from FLAG-Rab32 constructs using the following primers: 5′-ATGCCTCGAGTCATGGCGGGCGGAG-3′ and 5′-GCATGAATTCGCTCAGCAACACTGGGATTTG-3′, followed by cloning into pEGFP-C1 vector (Clonetech, Palo Alto, CA) using XhoI and EcoRI enzymes (New England BioLabs, Ipswich, MA). GST-Rab32 constructs were cloned by amplifying Rab32 sequence from GFP-Rab32 constructs using the following primers: 5′-GGATCCATGGCGGGCGGAGGAGCC-3′ and 5′-CTCGAGTCAGCAACACTGGGATTTC-3′, followed by cloning into TA-pCR2.1 vector (Invitrogen) using BamHI and XhoI enzymes and subcloned into pGEX-4T-1 (GE Healthcare, Chicago, IL) using the same enzymes.

Plasmids obtained from Addgene (Watertown, MA): pLJC5-Tmem192-3xHA (#102930) (Abu-Remaileh et al., 2017), pRK5-HA GST RagBQ99L (RagB^GTP) (#19303) and pRK5-HA GST-RagCS75L (RagC^GDP) (#19305) (Sancak et al., 2008) were a gift from David Sabatini; fatty acid synthase (FAS) promoter luciferase was a gift from Bruce Spiegelment (#8890) (Kim et al., 1998), RFP-EGFP-LC3 was a gift from Tamotsu Yoshimori (#21074) (Kimura et al., 2007), pEGFP-N1-TFEB was a gift from Shawn Ferguson (#38119) (Roczniak-Ferguson et al., 2012). pRL-TK Renilla control reporter vector (#E2241; Promega Madison, WI) was a kind gift from Dr Gregory J. Gores (Mayo Clinic, Rochester, MN). Lamp1 constructs were previously described (Schulze et al., 2013).

Primary rat hepatocyte isolation and culturing was performed as previously described (Schott et al., 2017). Hep3B2.1-7 (Hep3B), hepatocellular carcinoma from human liver (ATCC HB-8064, Manassas, VA), was maintained in minimum Eagle's medium (MEM) (Corning, Corning NY) with 10% fetal bovine serum (FBS), 100 U/ml penicillin, 100 µg/ml streptomycin (ThermoFisher) and supplemented with 1 mM sodium pyruvate, nonessential amino acids, and 0.075% (wt/vol) sodium bicarbonate (Corning). HeLa cell line, an adenocarcinoma from human cervix (ATCC CCL-2), was maintained in Dulbecco's modified Eagle's medium (Corning) with 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin. Cells are routinely tested for contamination. All cells were grown at 37°C in 5% CO₂ on acid-washed coverslips for fluorescence microscopy and in plastic culture dishes for biochemical analysis. Amino acid-free medium was formulated according to manufacturer's instructions (US Biological Life Sciences, Salem, MA; #D9800-13). To make amino acid-rich medium, amino acid-free medium was supplemented with MEM amino acids solution (Sigma; #M5550) and L-Glutamine (Gibco/ThermoFisher; #A2916801). For serum starvation and re-stimulation experiments we used Hanks’ Balanced Salt Solution (HBSS) (Gibco; #24020117) followed by re-stimulation with 10% FBS MEM medium. Cells were transfected using Lipofectamine 2000 for DNA constructs or RNAiMAX for siRNA reagents (both from ThermoFisher) according to the manufacturer's protocol.

Cells were processed for immunofluorescence as previously described (Henley and McNiven, 1996). In brief, cells were rinsed in PBS, fixed in 2.5% formaldehyde, permeabilized in 0.1% Triton X-100 for 2 min and blocked in buffer containing 5% goat serum. Fixed samples were then incubated with primary and secondary antibodies diluted in blocking buffer at 37°C. For images of live cells, cells were plated on glass-bottomed imaging dishes (Cell E&G LLC, San Diego, CA) following transfection and maintained at 37°C and 5% CO₂ for the duration of imaging. LysoTracker and LysoSensor staining was performed according to manufacturer's protocol. Images of both fixed and live cells were acquired using Zeiss LSM 780 confocal microscope (Carl Zeiss, Oberkochen, Germany). Images of TFEB-GFP localization were acquired using a Zeiss AxioObserver epifluorescence microscope (Carl Zeiss, Thornwood, NY) equipped with a Colibri 7 LED light source. Image analysis was assessed using ImageJ software (NIH).

For immunoprecipitation of GFP-tagged proteins, GFP-Trap agarose beads (Chromotek, Planegg-Martinsried, Germany) were used following manufacturer's protocol. For immunoprecipitation of endogenous mTOR, cells were collected in 0.3% CHAPS lysis buffer (40 mM HEPES pH 7.4, 120 mM NaCl, 1 mM EDTA, 10 mM Pyrophosphate, 10 mM glycerophosphate, 0.3% CHAPS) containing complete protease (Roche, Basel, Switzerland) and Halt phosphatase (ThermoFisher; #78440) inhibitors, lysed on ice for 30 min and centrifuged to remove nuclei and cell debris. Protein concentrations were determined by BCA assay (Pierce/ThermoFisher; #23225). Cell lysates were precleared with prewashed Protein A-Sepharose or Protein G PLUS-Agarose beads (Sigma, #P3391 or Santa Cruz, #sc-2002, respectively) and incubated with primary antibody against mTOR (Bethyl Laboratories, Montgomery, TX; #A301-143A), IgG control or beads alone as an additional control. The samples were then incubated with beads followed by four to six washes in lysis buffer (increased salt concentration to 150 mM NaCl). All protein samples subjected to western blot analysis were resolved by SDS-PAGE and transferred onto PVDF membrane (EMD Millipore; #IPVH00010), following immunoblotting and detection using autoradiography film (HyBlot CL from Denville Scientific, Holliston, MA) and Kodak X-OMAT automatic processor (Rochester, NY).

GST-fusion proteins were expressed in Escherichia coli BL21(DE3) pLysS-competent cells (Invitrogen) and purified using glutathione-coated beads (GE Healthcare) according to manufacturer's instructions. Cells were lysed in lysis buffer [6:1 ratio of TCMN100 buffer (20 mM Tris pH 7.4, 1 mM CaCl₂, 1 mM MgCl₂, 100 mM NaCl) and NTCMN300 buffer (0.5% NP40, 20 mM Tris pH 7.4, 1 mM CaCl₂, 1 mM MgCl₂ and 300 mM NaCl)] and incubated with GST-proteins for 1.5 h followed by four washes in NTCMN300 buffer.

Lysosomes were isolated as previously described with minor modifications (Abu-Remaileh et al., 2017). Briefly, cells were transfected with control or Rab32 siRNA followed by expression of TMEM192-HA lysosomal protein. Cells seeded into three 15 cm plates (∼80% confluency) were used for each Lyso-IP pulldown. Cells were washed twice with PBS (Corning; #21-040-CV), scraped in 1 ml of KPBS buffer (136 mM KCl, 10 mM KH₂PO₄, 2 mM EDTA pH 7.25) containing protease inhibitors and centrifuged at 1000 g for 2 min at 4°C. Pelleted cells were resuspended in 350 µl of KBPS and homogenized with 30-50 strokes in a 2 ml homogenizer. The homogenate was centrifuged again at 1000 g for 2 min and equal amounts of supernatant were incubated with 20 µl of prewashed anti-HA magnetic beads (ThermoFisher; #88837) on a rotor for 30 min at 4°C to pulldown HA-tagged lysosomes. Isolated lysosomes were then washed five times in KBPS buffer followed by two gentle washes on the vortex. Samples were boiled at 100°C for 5 min and subjected to SDS-PAGE western blot analysis.

Cells were incubated in serum-free medium for 4 h followed by stimulation with 50 ng/ml of EGF for 0, 30 or 60 min prior to cell lysis and western blot analysis. The treatments were performed in the presence of 50 µg/ml cycloheximide throughout the duration of experiment.

Reverse phase protein array (RPPA) assay was performed by the Antibody-Based Proteomics Core located in Baylor College of Medicine, Houston, TX as previously described (Chang et al., 2015). Briefly, Hep3B cells were transfected with either control or Rab32 siRNA for 72 h and reseeded at 70-80% confluency on 10 cm plates in quadruplicates. Cells were lysed in 150 µl of Tissue Protein Extraction Reagent (TPER) (Pierce; #7850) with 150 mM of NaCl and protease and phosphatase inhibitors (Roche; #11836153001 and #04906837001, respectively). Protein concentration was measured using BCA assay and 0.5 mg/ml of protein samples were denatured in 4XSDS sample buffer and boiled at 100°C for 8 min. Lysates were then shipped on dry ice for RPPA analysis to Baylor College of Medicine. We determined significant changes between control and Rab32 knockdown samples using Student's t-test (with P<0.05 considered to be significant).

Cell proliferation and viability was determined using a standard crystal violet assay. In brief, cells were plated in 96-well plates at 4000 cells per well followed by siRNA knockdown for 96 h. Cells were then fixed in 25% glutaraldehyde and stained with crystal violet dye. Crystal violet staining was solubilized in 100 nM sodium citrate solution and measured at 550 nM wavelength using a plate reader. Colorimetric proliferation and viability assay of control and Rab32-depleted cells was performed using CellTiter 96 AQ_ueous Non-Radioactive Cell Proliferation Assay (MTS) kit following manufacturer's protocol (Promega; #G5421). The average 2D cell area was measured using cytosolic marker staining for reference using ImageJ software. The average 3D cell diameter of trypsinized cells in suspension was measured using a Countess automated cell counter (ThermoFisher).

Cells were subjected to siRNA knockdown for 24 h followed by co-transfection with FASN firefly and control pRL-TK Renilla luciferase vectors for additional 24 h. After re-seeding on 24-well plates, cells were then harvested in passive lysis buffer and assayed using Dual-Luciferase Reporter Assay System reagents (Promega; #E1910) using Synergy H1 Hybrid Multi-Mode Reader (BioTek, Vinooski, VT). Control Renilla luciferase vector values were used as an internal control to normalize transfection efficiency.

Data are represented as mean ±s.e.m.; two-tailed paired Student's t-test was used to assess the statistical significance between two conditions (unless otherwise specified) with P<0.05 considered to be significant. Experiments were repeated at least three times and representative data are shown.

We thank the members of the McNiven laboratory, especially Ryan J. Schulze and Gina L. Razidlo for helpful discussions. The authors also acknowledge the Antibody-based Proteomics Core, the core director Dr Shixia Huang, Dr Hsin-Yi Cincy Lu and Carlos Ramos for their excellent technical assistance in performing the RPPA screens, and Dr Kimal Rajapakshe, Dr Cristian Coarfa and Dimuthu Perera for RPPA data processing and normalization [Baylor College of Medicine, Houston, TX with funding support from Cancer Prevention & Research Institute of Texas Proteomics & Metabolomics Core Facility Support Award (RP170005) and National Cancer Institute Cancer Center Support Grant (P30CA125123)]. We also thank Kirsten Aspros and Dr Petra Hirsova for technical assistance with the Crystal Violet and luciferase assays, respectively (Mayo Clinic, Rochester, MN). We appreciate the gifts of FLAG-Rab32 and pRL-TK Renilla luciferase plasmids from Dr John D. Scott (University of Washington, Seattle, WA) and Dr Gregory J. Gores (Mayo Clinic, Rochester, MN), respectively.