LRRK1 phosphorylation of Rab7 at S72 links trafficking of EGFR-containing endosomes to its effector RILP

EGFR is activated by EGF at the plasma membrane and transduces signals important for cellular responses, such as growth, differentiation, proliferation and motility (Albeck et al., 2013; Ceresa and Peterson, 2014; Schlessinger, 2000). Activation of EGFR also initiates events leading to its own endocytosis. Internalized receptors are first associated with early endosomes, which then mature into late endosomes (Goh and Sorkin, 2013; Huotari and Helenius, 2011; Scott et al., 2014). During this maturation, EGFR is transported to lysosomes for degradation by the dynein–dynactin motor protein complex (Driskell et al., 2007). Recent studies have shown that EGFR signaling occurs not only at the plasma membrane but also in endosomes after internalization (Irannejad et al., 2015; Sorkin and von Zastrow, 2009). Thus, endosomal trafficking of EGFR determines the spatiotemporal regulation of EGFR signaling (Bakker et al., 2017; Miaczynska, 2013; Tomas et al., 2014).

Small GTPases of the Rab family are critical regulators of membrane trafficking (Hutagalung and Novick, 2011; Stenmark, 2009). Rabs switch between a membrane-associated, GTP-bound, active form and a cytosolic, GDP-bound, inactive form. The active, GTP-bound Rab protein binds to various effectors to regulate membrane trafficking (Hutagalung and Novick, 2011; Stenmark, 2009). Rab7 (herein referring to Rab7a) is a member of the Rab family that has been demonstrated to play a crucial role in regulating endo-lysosomal membrane traffic (Guerra and Bucci, 2016). During the early-to-late endosome transition, Rab7 is recruited to the subdomains of early endosomes bearing Rab5, followed by Rab5 displacement from the same endosome and the acquirement of Rab7-mediated transport capacity (Pfeffer, 2013; Poteryaev et al., 2010; Rink et al., 2005). Rab7 regulates the movement of endosomes along microtubules in a bi-directional manner by interacting with either of two effectors: the effector RILP, which recruits the dynein–dynactin motor complex driving minus-end transport (Johansson et al., 2007; Jordens et al., 2001), or the effector FYCO1, which recruits the kinesin motor driving plus-end transport (Pankiv et al., 2010). However, EGFR-containing endosomes have been shown to be preferentially transported by the Rab7–RILP complex, moving along microtubules toward the perinuclear region (Progida et al., 2007; Vanlandingham and Ceresa, 2009). It remains largely unknown how Rab7 selectively interacts with RILP to facilitate this endosomal trafficking of EGFR.

Recently, we have demonstrated that the endocytic trafficking of EGFR destined for lysosomal degradation is regulated by LRRK1 (Hanafusa et al., 2011; Ishikawa et al., 2012; Kedashiro et al., 2015). LRRK1 is related to the familial Parkinsonism gene product LRRK2 (also known as Park8) and belongs to the ROCO family of proteins, which contain a Ras of complex proteins (ROC) GTPase domain and a MAPKKK-like kinase domain (Bosgraaf and Van Haastert, 2003). LRRK1 forms a complex with activated EGFR and is involved in the initiation and maintenance of the dynein-mediated transport of early endosomes containing EGFR. This involvement of LRRK1 is dependent on its intrinsic kinase activity (Hanafusa et al., 2011; Ishikawa et al., 2012; Kedashiro et al., 2015). In the initiation step, LRRK1 phosphorylates CLIP-170 (also known as CLIP1), a microtubule plus-end protein, which facilitates its interaction with a subunit of dynactin p150^Glued (also known as DCTN1). This, in turn, stimulates the dynein–dynactin complex-driven transport of EGFR-containing endosomes (Kedashiro et al., 2015). Furthermore, we have shown that EGFR regulates LRRK1 kinase activity via tyrosine phosphorylation, which is required for the proper endosomal trafficking of EGFR (Ishikawa et al., 2012). Phosphorylation of LRRK1 at Y944 results in reduced LRRK1 kinase activity. Accordingly, mutation of LRRK1 Y944 into a phenylalanine residue (Y944F) abolishes EGF-stimulated tyrosine phosphorylation, resulting in the hyper-activation of LRRK1 kinase activity and enhanced motility of EGF-containing endosomes towards the perinuclear region. However, the downstream target of LRRK1 that mediates this long-range movement of EGFR-containing endosomes had not been identified.

In this study, we examined the relationship between LRRK1 and Rab7 in mediating long-range EGFR movement. We show that LRRK1 interacts with and phosphorylates Rab7 at a conserved S72 residue located in its switch II region. We demonstrate that LRRK1-mediated phosphorylation of Rab7 increases its interaction with its effector RILP, but not with FYCO1, and promotes the long-range minus-end-directed transport of EGFR-containing endosomes. Thus, our findings reveal a mechanism by which LRRK1 determines the selective interaction of Rab7 with its effector in a cargo-dependent manner.

We have recently reported that LRRK1 regulates the dynein-mediated transport of EGFR-containing endosomes in a manner dependent on its kinase activity (Ishikawa et al., 2012; Kedashiro et al., 2015). Rab7 also plays an important role in the transport of EGFR-containing endosomes toward the nucleus (Ceresa and Bahr, 2006; Vanlandingham and Ceresa, 2009). We therefore examined the possible relationship between LRRK1 and Rab7. We first asked whether LRRK1 and Rab7 physically interact, and found that Flag-tagged LRRK1 co-precipitated with GFP-tagged Rab7 (Fig. 1A, lane 2). Since Rab7 cycles between the GDP-bound inactive and GTP-bound active conformations, we next asked whether LRRK1 preferentially binds to one of these two forms. However, we found that LRRK1 interacted similarly with both Rab7(Q67L), a mutant of Rab7 to which GTP is bound, and Rab7(T22N), to which GDP is bound (Fig. 1A, lane 3,4 compared to lane 2). Thus, LRRK1 interacts with Rab7 independently of its guanine nucleotide-binding state.

Next, we asked whether LRRK1 phosphorylates Rab7 in an in vitro kinase assay with purified recombinant GST–Rab7. We found that a hyper-active LRRK1 mutant, LRRK1(Y944F) (Ishikawa et al., 2012), could phosphorylate GST–Rab7, whereas the kinase-inactive mutant LRRK1(K1243M) could not (Fig. 1B, lanes 2,3). These results show that Rab7 is a kinase substrate of LRRK1. To identify the site(s) within Rab7 that are phosphorylated by LRRK1, we incubated GST–Rab7 with LRRK1(Y944F) or LRRK1(K1243M) under in vitro kinase conditions and subjected the products to analysis by liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS). Four phosphorylation sites, S34, T40, S72 and T168, were found in Rab7 (Table S1). To determine which Rab7 site(s) are phosphorylated by LRRK1 in vivo, we analyzed Rab7 phosphorylation by performing phosphate-affinity (Phos-tag) polyacrylamide gel electrophoresis (PAGE), which detects phosphorylated Rab7 through its slower migration. When LRRK1(Y944F) was co-expressed with wild-type Flag–Rab7 in HEK293 cells, Rab7 proteins appeared in the upper band of the Phos-tag PAGE gel (Fig. 1C, lane 2). In contrast, co-expression of LRRK1(K1243M) did not induce a mobility shift (Fig. 1C, lane 1). Thus, LRRK1 kinase activity catalyzes the phosphorylation of Rab7 in vivo. When co-expressed with LRRK1(Y944F), Flag-Rab7(S72A) did not appear in the upper band (Fig. 1C, lane 8). However, other Rab7 mutants, namely, Flag–Rab7(S34A), (T40A) and (T168A), still gave a band shift in the Phos-tag PAGE analysis (Fig. 1C, lanes 4,6 and 10). These results suggest that Rab7 S72 is phosphorylated by LRRK1 in vivo.

To confirm that Rab7 S72 is the major site of LRRK1 phosphorylation, we produced GST–Rab7(S72A) and performed an in vitro kinase assay. We found that, in contrast to wild-type GST–Rab7, phosphorylation of GST–Rab7(S72A) by LRRK1(Y944F) was undetectable (Fig. 1B, lane 3 compared to lane 6). These results suggest that within Rab7, S72 is the major phosphorylation site for LRRK1. We further confirmed that the S72 site in Rab7 can be phosphorylated by LRRK1 in vivo. For this purpose, we generated an antibody specific for Rab7 phosphorylated at S72 (pS72-Rab7). In immunoblots, we detected a single band recognized by the anti-pS72-Rab7 antibody when Flag–Rab7 was co-expressed with GFP-LRRK1(Y944F) but not with GFP–LRRK1(K1243M) (Fig. S1A). Furthermore, the S72A mutation abolished the pS72-Rab7 signal even in cells co-expressing LRRK1(Y944F) (Fig. S1A). In addition, we found that anti-pS72-Rab7 antibody recognized only the upper band of Rab7 in a Phos-tag PAGE gel (Fig. S1B). Taken together, these results indicate that LRRK1 indeed phosphorylates Rab7 S72 in vivo.

Active Rab7 predominantly localizes to late endosomes and lysosomes (Bucci et al., 2000). We previously reported that a hyper-active LRRK1(Y944F) mutant enhances the long-range movement of EGFR-containing endosomes toward the nucleus, which results in their perinuclear clustering (Ishikawa et al., 2012). We therefore used confocal fluorescence microscopy to investigate the effect of LRRK1(Y944F) on the subcellular localization of Rab7 after EGF stimulation. The movement of EGFR-containing endosomes was followed in HeLa S3 cells after cells were treated with fluorescently labeled Alexa Fluor 647-conjugated EGF (A647–EGF). At 10 min after EGF stimulation of control cells, A647–EGF and endogenous Rab7 were distributed in a fine (small dots) punctate pattern, but did not colocalize (Fig. 2A,D). However, at 10 min after EGF stimulation of cells expressing GFP–LRRK1(Y944F), a significant fraction of endogenous Rab7 was colocalized with A647–EGF in the punctate structures (Fig. 2B,D). We have previously demonstrated that EGFR and LRRK1 are endocytosed together to the endosomes (Hanafusa et al., 2011). Consistent with this, almost all the GFP–LRRK1(Y944F) colocalized with A647–EGF (Fig. 2B). Since most of the EGF is present in early endosomes soon after EGF stimulation, it would be expected that LRRK1(Y944F) induces Rab7 localization in early endosomes. To test this possibility, we co-stained the cells with markers for different endocytic vesicles. We found that at 10 min after A647–EGF stimulation of GFP–LRRK1(Y944F)-expressing cells, a fraction of the Rab7 that colocalized with GFP–LRRK1(Y944F) also overlapped with the early endosomal marker EEA1, but not with the late endosomal marker CD63 or the lysosomal marker LAMP1 (Fig. S2A–C). Thus, LRRK1(Y944F) induces the localization of Rab7 in early endosomes. In contrast, expression of the kinase-inactive GFP–LRRK1(K1243M) failed to induce localization of Rab7 to early endosomes (Fig. 2C,D), suggesting that the ability of LRRK1(Y944F) to induce Rab7 localization is due to the hyper-activation of its kinase activity. Furthermore, we found that when LRRK1(Y944F) was expressed, phosphorylated Rab7 at S72 was localized in GFP-LRRK1(Y944F)-positive EGFR-containing endosomes (Fig. 2E,F). These results suggest that LRRK1(Y944F) phosphorylates Rab7 S72 in EGFR-containing early endosomes, leading to the stabilization of its endosomal localization.

Recently, Shinde and Maddika have demonstrated that PTEN dephosphorylates Rab7 at S72, a modification that is required for the recruitment of Rab7 to late endosomes induced by the GDP dissociation inhibitor (GDI) protein (Shinde and Maddika, 2016). Indeed, they showed that in PTEN-depleted cells, Rab7 was localized in a diffused cytoplasmic pattern, suggesting that phosphorylation of Rab7 S72 prevents its membrane localization by inhibiting the interaction of Rab7 with GDI (Shinde and Maddika, 2016). This contradicts our results showing that phosphorylation of Rab7 S72 stimulates its endosomal localization. However, in the Shinde and Maddika study, the effect of PTEN depletion on localization of Rab7 was examined in the absence of stimulation. Consistent with their results (Shinde and Maddika, 2016), we confirmed that colocalization of GFP–Rab7 with the late endosomal marker CD63 was reduced in PTEN-depleted cells in the absence of EGF stimulation (Fig. S3A,B).

Therefore, we examined the effect of PTEN knockdown on the intracellular distribution of Rab7 after EGF stimulation. When control siRNA-treated HeLa S3 cells were stimulated with Rhodamine-conjugated EGF (Rh–EGF) for 10 min, GFP–Rab7 did not localize to Rh–EGF-positive endosomes (Fig. S4A). In contrast, we found that, in PTEN-depleted cells, colocalization of GFP–Rab7 with Rh–EGF increased at 10 min after EGF stimulation (Fig. S4B). This is similar to the results observed in cells expressing LRRK1(Y944F) (Fig. 2B,D). These results suggest that depletion of PTEN promotes the localization of Rab7 to EGFR-containing endosomes.

We further examined the subcellular localization of endogenous Rab7 in PTEN-depleted cells. In control cells, the majority of Rab7 was absent from A647–EGF-positive endosomes at 8 min after EGF stimulation (Fig. 3A,D). In contrast, in PTEN-depleted cells, colocalization of Rab7 with A647–EGF was significantly increased (Fig. 3B,D). These results confirm that PTEN negatively regulates the localization of Rab7 to EGFR-containing endosomes. Importantly, simultaneous depletion of both PTEN and LRRK1 completely abrogated Rab7 localization to EGFR-containing endosomes (Fig. 3C,D; Fig. S3A). Thus, LRRK1 and PTEN target S72 in Rab7 as a kinase and a phosphatase, respectively. Taken together, these results suggest that LRRK1 promotes the localization of Rab7 to EGFR-containing endosomes, while PTEN prevents its premature localization.

The LRRK1 phosphorylation site in Rab7, S72, is located in the conserved switch II region (Fig. 4A). It has been recently reported that another member of the ROCO family, LRRK2, phosphorylates a subset of Rab proteins, including Rab3, Rab8, Rab10, Rab12, Rab35 and Rab43, at conserved serine and/or threonine residues in the switch II region (Fig. 4A) (Ito et al., 2016; Steger et al., 2016, 2017). The switch II region is known to be relatively disordered in the inactive GDP-bound state compared with that in the active GTP-bound state (Lee et al., 2009; Pfeffer, 2005; Pylypenko et al., 2018). Indeed, LRRK2 preferentially phosphorylates GTP-bound Rab proteins (Liu et al., 2018). Therefore, we examined whether LRRK1 phosphorylation of Rab7 is also specific for the GTP-bound form. Phos-tag analysis revealed that when expressed in HEK293 cells, LRRK1(Y944F) induced phosphorylation of wild-type Rab7 and the GTP-bound form Rab7(Q67L), but not of the GDP-bound form Rab7(T22N) (Fig. 4B, lanes 2–4). These results indicate that LRRK1 also specifically phosphorylates GTP-bound Rab7.

The fact that GTP-bound Rab7 localizes to membranes raised the possibility that LRRK1 phosphorylation of Rab7 GTP-bound form also occurs at the membrane. To test this possibility, we constructed Rab7(CS), where two C-terminal cysteine residues (C205 and C207), prenylation of which is known to be required for Rab7 membrane localization (Seabra, 1998; Wu et al., 2009), are replaced by serine residues. Since Rab7(CS) is not prenylated, it not expected to localize to membranes (Seabra, 1998; Wu et al., 2009). Consistent with this expectation, Flag–Rab7(CS) and Flag–Rab7(Q67L/CS) exhibited a diffuse distribution and did not localize to CD63-positive late endosomes (Fig. S5). We found that expression of LRRK1(Y944F) was unable to induce phosphorylation of Rab7(CS) or Rab7(Q67L/CS) in Phos-tag analysis (Fig. 4B, lanes 5,6). Thus, membrane localization of Rab7 is required for its phosphorylation by LRRK1. These results suggest that LRRK1 selectively phosphorylates GTP-bound membrane-localized Rab7. Consistent with this, a significant but relatively low proportion of wild-type and GTP-bound Rab7 was phosphorylated by LRRK1(Y944F) (Fig. 4C).

Rab GTPases cycle between the cytosol, where they are GDP-bound and inactive, and specific membrane compartments, where they are GTP-bound and active (Hutagalung and Novick, 2011; Stenmark, 2009). Rab7 crystal structure studies have revealed that the switch II region regulates hydrolysis of GTP and coordinates the binding to effector proteins (Wu et al., 2005). We therefore tested whether Rab7 protein interactions are modulated by LRRK1 phosphorylation. RILP is one of the Rab7 effector proteins that mediate downstream vesicular trafficking events (Cantalupo et al., 2001; Jordens et al., 2001). To investigate the effect of LRRK1 phosphorylation of Rab7 S72 on its interaction with RILP, we performed pulldown experiments using Halo-tagged RILP. Lysates prepared from Cos7 cells expressing Halo–RILP were first bound to Magne-HaloTag beads. Then, lysates of HEK293 cells expressing GFP–Rab7 with or without GFP-LRRK1(Y944F) were mixed with the beads. These pulldown experiments revealed that the Rab7–RILP interaction was enhanced by LRRK1(Y944F) (Fig. 5A). This effect was ablated in the phosphorylation-incompetent GFP-Rab7(S72A) mutant (Fig. 5A). We next examined whether the effect of LRRK1 phosphorylation on Rab7 interaction was effector specific. FYCO1 is another Rab7 effector that binds to the switch II domain of Rab7 (Pankiv et al., 2010). However, we found that in contrast to RILP, LRRK1(Y944F) had no effect on Rab7–FYCO1 interaction (Fig. 5B). These results suggest that LRRK1(Y944F)-induced phosphorylation of Rab7 specifically increases its interaction with RILP.

Next, we examined the effect of LRRK1(Y944F) on the intracellular distribution of RILP after EGF stimulation. RILP functions to maintain Rab7 in the active and membrane-bound state, and RILP expression has been reported to cause an accumulation of late endosomes around the nucleus (Cantalupo et al., 2001; Jordens et al., 2001). We expressed Halo–RILP in HeLa S3 cells and stimulated them with A647–EGF. At 10 min after EGF stimulation, Halo–RILP accumulated near the nucleus and failed to localize to A647–EGF-positive endosomes (Fig. 5C,E). Interestingly, when GFP–LRRK1(Y944F) and Halo–RILP were co-expressed, colocalization of A637–EGF with Halo–RILP increased (Fig. 5D,E). These results suggest that LRRK1 promotes the recruitment of RILP to EGFR-containing endosomes by enhancing the interaction between Rab7 and RILP.

Expression of the hyperactive LRRK1(Y944F) mutant enhances the long-range movement of EGFR-containing endosomes, resulting in the accumulation of clustered and enlarged endosomes (Ishikawa et al., 2012). To examine this phenomenon further, we asked whether the effect of LRRK1(Y944F) on the localization of EGFR-containing endosomes is dependent on the dynein motor. Stimulation of LRRK1(Y944F)-expressing cells with Rh–EGF for 30 min induced the perinuclear clustering of Rh–EGF-positive endosomes (Fig. 6A,E). In contrast, pre-treatment of these cells with ciliobrevin, a specific inhibitor of dynein (Firestone et al., 2012), resulted in the scattered distribution of GFP–LRRK1(Y944F) and Rh–EGF-positive endosomes in the cytosol (Fig. 6B,E). This result confirmed that LRRK1(Y944F)-mediated clustering of EGFR-containing endosomes is dependent on dynein.

We next examined whether Rab7 or RILP is required for this process. We found that depletion of Rab7 by means of siRNA completely eliminated the LRRK1(Y944F)-induced clustering of Rh–EGF-positive endosomes (Fig. 6C,E; Fig. S6A,B). Furthermore, depletion of RILP also significantly inhibited this clustering (Fig. 6D,E; Fig. S6A). These results were also observed using different Rab7 and RILP siRNAs (Fig. S6C–E). Thus, both Rab7 and RILP are required for LRRK1(Y944F)-induced perinuclear clustering of EGFR-containing endosomes. Taken together, these results suggest that the Rab7–RILP–dynein complex is necessary for LRRK1(Y944F)-mediated long-range transport of EGFR-containing endosomes toward the nucleus.

In this study, we show that LRRK1 phosphorylates the conserved S72 residue within the switch II region of Rab7. This in turn increases the association of Rab7 with its effector RILP, resulting in enhanced minus-end transport of EGFR-containing endosomes through a dynein-dependent mechanism. Recent studies have shown that the endocytic trafficking of EGFR depends on the concentration of EGF to which cells are exposed (Sigismund et al., 2005). When cells are stimulated with low EGF concentrations, EGFR is typically subjected to clathrin-mediated endocytosis and a large fraction of the receptor is recycled to the cell surface, allowing for prolonged EGFR signaling (Sigismund et al., 2008). In contrast, at higher concentrations of EGF, the receptor is preferentially internalized by a clathrin-independent pathway that involves its ubiquitylation, sorting into the intraluminal vesicles (ILVs) of the multivesicular body and transfer to the lysosomes for degradation (Futter, 1996; Sigismund et al., 2005, 2013). ILV sorting of the receptor also involves the physical removal of the signaling tail of EGFR from the cytosol, which effectively terminates downstream signaling prior to lysosomal degradation (Eden et al., 2009; Katzmann et al., 2002). These systems are thought to allow cells to both amplify weak physiological inputs and to effectively cope with overstimulation. We have previously demonstrated that LRRK1 is required for the efficient sorting of ubiquitylated EGFR into the ILVs by interacting with STAM1 and Hrs, subunits of the endosomal sorting complexes required for transport (ESCRT)-0 complex, and facilitates the dynein-driven transport of EGFR-containing endosomes for lysosomal degradation (Hanafusa et al., 2011; Ishikawa et al., 2012; Kedashiro et al., 2015). Therefore, LRRK1 might function in the latter system to efficiently terminate EGFR signaling, thus protecting the cell from overstimulation.

A previous study has shown that PTEN dephosphorylation of Rab7 S72 is necessary for the GDI-dependent recruitment of Rab7 into late endosomes; this study also showed that a constitutive phosphomimetic mutant Rab7(S72E) fails to localize to the endosomal membrane and does not interact with RILP (Shinde and Maddika, 2016). These results are apparently inconsistent with our present findings. However, while PTEN depletion reduces the localization of Rab7 into late endosomes in unstimulated cells, PTEN depletion actually enhances Rab7 localization to EGF-positive endosomes in EGF-stimulated cells. Furthermore, it is difficult to elucidate the role of Rab7 S72 phosphorylation by using Rab7(S72E). It is known that newly synthesized GDP-bound Rab proteins interact with Rab escort proteins (REP1 and REP2), which interacts with geranylgeranyl transferase type II (GGTII) and facilitates Rab protein prenylation and membrane anchoring (Hutagalung and Novick, 2011; Pylypenko et al., 2018; Seabra and Wasmeier, 2004). In the case of Rab8, phosphomimetic substitution of the residue corresponding to Rab7 S72 interferes with its interaction with REP1, REP2 and GGTII, resulting in defective prenylation and failure to localize to the membrane (Steger et al., 2016). Similarly, the nascent Rab7(S72E) protein may be defective in its interaction with REP1, REP2 or GGTII, resulting in the accumulation of the hypoprenylated protein in the cytosol. Indeed, Rab7(S72E) exhibits a diffused cytoplasmic distribution (Shinde and Maddika, 2016). In addition, because Rab7(S72E) fails to interact with its GEF CCZ1, it exists predominantly in the GDP-bound inactive form (Shinde and Maddika, 2016). This shift to the inactive form would explain why the S72 phosphomimetic substitution interferes with its interaction with RILP. The cycling of Rab7 between the cytosolic GDP-bound inactive and membrane-associated GTP-bound active conformations is mediated by interactions with a number of regulatory proteins. Our results indicate that LRRK1 phosphorylates the active GTP-bound form of Rab7 at the membrane subsequent to its endosomal delivery by GDI. A recent study by Steger et al. has shown that LRRK2 phosphorylates a subset of Rab proteins on an evolutionarily conserved residue in the switch II domain, which, in the case of Rab7, corresponds to S72 (Steger et al., 2016, 2017). This region undergoes a conformational change upon GTP binding, and this conformational change is required for Rab interaction with its binding partners (Pfeffer, 2005). Indeed, phosphorylation of Rabs on this residue inhibits their association with a number of regulatory proteins, including REP1, REP2, GGTII, GDI1, GDI2 and GEFs (Steger et al., 2016, 2017). In contrast, the two effector proteins RILP-L1 and RILP-L2 have been shown to bind preferentially to the LRRK2-phosphorylated Rab8, Rab10 and Rab12 and function in the formation of primary cilia (Dhekne et al., 2018; Steger et al., 2017). These interactions are mediated by highly conserved basic residues in the RILP homology (RH) domain of RILP-L1 and RILP-L2, which are also conserved in RILP. Interestingly, we found that LRRK1 phosphorylation on S72 enhanced the binding of Rab7 to RILP. In contrast, LRRK1 phosphorylation had no effect on the Rab7–FYCO1 interaction. Thus, the effect of Rab7 S72 phosphorylation on the effector interaction is specific for RILP. These results raise the possibility that LRRK1-mediated phosphorylation of Rab7 recruits the RILP–dynein–dynactin complex to EGFR-positive endosomes, and this would explain why EGFR-containing endosomes are largely restricted to migrate toward the nucleus. Our findings thus provide the first evidence for cargo-dependent regulation of endosomal trafficking.

We have previously reported that LRRK1 facilitates the initiation of the dynein-driven transport of EGFR-containing endosomes through the recruitment of p150^Glued at microtubule plus-ends by phosphorylating CLIP-170 (Kedashiro et al., 2015). Thus, the hyperactivation of LRRK1 kinase enhances the motility of EGFR-containing endosomes toward the nucleus and causes the immature perinuclear clustering of EGFR-containing endosomes, which cannot fuse with lysosomes (Ishikawa et al., 2012). However, the downstream target of LRRK1 in this process has not been identified. Here, we identified Rab7 as a substrate for LRRK1 that maintains the dynein-mediated transport of EGFR-containing endosomes along microtubules. These results suggest that LRRK1 plays an important role in the initiation and maintenance of the dynein-mediated transport of EGFR-containing endosomes by phosphorylating CLIP-170 and Rab7, respectively.

LRRK2 has been reported to be present at multiple distinct membrane compartments and to function in vesicle trafficking through the endolysosomal and autophagic pathways (Alegre-Abarrategui et al., 2009; Biskup et al., 2006). Interestingly, LRRK2 also phosphorylates the GTP-bound membrane-associated form of Rabs (Liu et al., 2018). LRRK1 and LRRK2 might regulate different aspects of membrane trafficking by phosphorylating target Rabs in different contexts. Since active GTP-bound Rabs function by recruiting distinct sets of effector proteins, it is critical to understand how they recognize their effectors specifically. Thus, the binding of effectors to the Rabs is regulated not only by guanine nucleotide cycling, but also by LRRK1- and LRRK2-mediated phosphorylation.

HeLa S3, Cos7 and HEK293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum. These cell lines were obtained from the Japanese Collection of Research Bioresources (JCRB) or the American Type Culture Collection (ATCC) and were regularly tested for mycoplasma contamination. Antibody against pS72-Rab7 was produced in rabbit by injection with the synthetic phosphopolypeptide AGQERFQpSLGVAF (where p stands for phosphorylated residue), coupled to keyhole limpet haemocyanin and affinity purified (Sigma). Antibodies and their suppliers were: anti-Rab7 (ab137029, Abcam or D95F2, Cell Signaling), anti-Flag (M2, Sigma or FLA-1, MBL), anti-GFP (JL-8, Clontech or 598, MBL), anti-PTEN (138G6, Cell Signaling), anti-Halo (G9281 or G9211, Promega), anti-EEA1 (clone 14, BD Transduction Laboratories), anti-CD63 (sc5275, Santa Cruz Biotechnology), anti-LAMP1 (H4A3, BD Transduction Laboratories) and anti-RILP (SAB2107831, Sigma). Affinity-purified rabbit antibodies against LRRK1 have been described previously (Hanafusa et al., 2011). Rh–EGF was from Ciliobrevin (Calbiochem) and A647–EGF from Invitrogen.

Human LRRK1 was cloned from a cDNA library by RT-PCR (our clone lacks 27 amino acids at the N-terminus compared with NM_024652), and GFP–LRRK1, GFP–LRRK1(K1243M) and GFP–LRRK1(Y944F) were generated as described previously (Ishikawa et al., 2012). Rab7(Q67L), Rab7(T22N), Rab7(S34A), Rab7(T40A), Rab7(S72A), Rab7(T168A), Rab7(CS) and Rab7(Q67L/CS) were generated by using the QuikChange site-directed mutagenesis kit according to the manufacturer's protocol (Stratagene, La Jolla, CA) and subcloned into the pEGFP-C1 (Clontech) or pCMV-Flag vectors. Halo–RILP and Halo–FYCO1 were obtained from Kazusa DNA Res. Inst. (Promega). siRNA targeting human LRRK1 and negative control siRNAs were obtained as previously described (Hanafusa et al., 2011). Pre-validated PTEN siRNA was purchased from Qiagen (catalog no. SI00301504) and siRNA for human Rab7 #1 [target sequence: CGGTTCCAGTCTCTCGGTd(TT)], human Rab7 #2 [target sequence, GGATGACCTCTAGGAAGAAd(TT)], human RILP #1 [target sequence, GATCAAGGCCAAGATGTTAd(TT)] and human RILP #2 [target sequence, GCAGCGGAAGAAGATCAAGd(TT)] were purchased from JBioS. Annealed siRNAs were transfected using RNAiMAX (Invitrogen). The transfected cells were analyzed 72 h after transfection. It was confirmed using FITC-labeled oligonucleotides that almost all HeLa S3 cells had been transfected with the siRNA oligonucleotides.

The recombinant proteins GST–Rab7 and GST–Rab7(S72A) were each expressed in the E. coli strain BL21-CodonPlus (DE3)-RIPL and purified using glutathione–Sepharose 4B (GE Healthcare) following the manufacturer's guidelines. All GFP–LRRK1 proteins were expressed in Cos7 cells and immunopurified with anti-GFP antibody (4 μl/sample; 598, MBL) (Kedashiro et al., 2015). Kinase reactions were performed in a final volume of 20 µl buffer consisting of 50 mM HEPES pH 7.4, 5 mM MgCl₂, 5 mM MnCl₂, 0.5 mM DTT, 5 µCi of [γ-³²P]ATP and 100 µM ATP. Samples were incubated for 20 min at 30°C and the reactions terminated by addition of Laemmli sample buffer and boiling. Samples were resolved by SDS-PAGE and analyzed by autoradiography.

For LRRK1 phosphorylation site analysis, immunoprecipitated Rab7 proteins were subjected to a non-radioactive in vitro kinase assay and then eluted with guanidine solution (50 mM NH₄HCO₃, 7 M guanidine-HCl), followed by reduction, alkylation, demineralization and concentration as described previously (Kedashiro et al., 2015). Rab7 proteins were digested with trypsin for 16 h at 37°C. From these peptide samples, phosphopeptides were enriched and captured using the Titansphere Phos-TiO Kit according to the manufacturer's instructions. Nano-electrospray tandem mass analysis was performed using a Q Exactive mass spectrometer (ThermoFisher Scientific Inc.) system combined with a Paradigm MS4 HPLC system (Michrom BioResources Inc.). Samples were injected into the Advance nanoflow UHPLC/HTS-PAL system equipped with a MonoCap C18 Nano-flow column 0.1 mm×150 mm (GL Sciences). Reversed-phase chromatography was performed with a linear gradient (0 min, 5% B; 70 min, 40% B) of solvent A (H₂O with 0.1% formic acid) and solvent B (acetonitrile) at an estimated flow rate of 400 nl/min. Ionization was performed with an ADVANCE CaptiveSpray Source (Michrom BioResources Inc.). A precursor ion scan was carried out using a 380–1900 mass to charge ratio (m/z) prior to MS/MS analysis. Raw data were analyzed using Proteome Discoverer™ software with the Sequest™ algorithm at 15 ppm precursor mass accuracy and 0.02 Da MS/MS tolerance. The peptide search was performed against UniProtKB Homo sapiens reference proteome dataset (release 2012_10) with a 1% false discovery rate cut-off. The most likely localization of a phosphorylation site was determined by using the PhosphoRS algorithm within the Proteome Discoverer software.

For immunoprecipitation, cells were lysed in RIPA buffer [50 mM Tris-HCl pH 7.4, 0.15 M NaCl, 0.25% deoxycholic acid, 1% NP-40, 1 mM EDTA, 1 mM dithiothreitol, phosphatase inhibitor cocktail 2 (Sigma) and protease inhibitor cocktail (Sigma)], followed by centrifugation at 15,000 g for 12 min. The supernatant was added to 50 µl (1.5 mg) of Dynabeads Protein G (Invitrogen) with the indicated antibodies (each antibody was used at 5 μg/sample) and rotated for 2 h at 4°C. The beads were then washed three times with ice-cold phosphate-buffered saline (PBS) and subjected to immunoblotting. For Halo-tag pulldown assays, Cos7 cells expressing Halo–RILP or Halo–FYCO1 were lysed in mammalian lysis buffer (G938A, Promega) supplemented with 1 mM dithiothreitol, phosphatase inhibitor cocktail 2 (Sigma) and protease inhibitor cocktail (Promega), followed by centrifugation at 15,000 g for 12 min. The supernatant was added to 100 µl (20% slurry) of Magne-HaloTag beads (Promega), rotated for overnight at 4°C, and then washed three times with ice-cold lysis buffer. HEK293 cells expressing GFP–Rab7 (WT or S72A) with or without GFP–LRRK1(Y944F) were lysed in RIPA buffer [50 mM Tris-HCl pH 7.4, 0.15 M NaCl, 0.25% deoxycholic acid, 1% NP-40, 1 mM EDTA, 1 mM dithiothreitol, phosphatase inhibitor cocktail 2 (Sigma) and protease inhibitor cocktail (Sigma)] supplemented with 10 mM MgCl₂ and 0.1 mM GTPγS, followed by centrifugation at 15,000 g for 12 min. The supernatants were mixed with Halo–RILP or Halo–FYCO1-bound beads and rotated for 2 h at 4°C, and then washed three times with ice-cold lysis buffer and once with cold PBS. The assays were analyzed by immunoblotting with the indicated antibodies. Primary antibodies were mouse anti-GFP at 1:500 or mouse anti-Halo at 1:1000.

For analysis of phosphorylation status of Rab7, 15% Phos-tag precast gels (SuperSep Phos-tag, Wako) were used. After electrophoresis, the Phos-tag acrylamide gel was washed three times by gentle shaking in transfer buffer (Fast buffer, ATTO) containing 0.01% SDS and 10 mM EDTA for 10 min and then incubated in transfer buffer containing 0.01% SDS without EDTA for 10 min according to the manufacturer's protocol. Proteins were transferred to polyvinylidene difluoride (PVDE) membranes and analyzed by immunoblotting with anti-Flag (1:500) or anti-pS72-Rab7 (1:200) antibody. Immunoreactive proteins were visualized and quantified using the FUSION system (VILBER).

For immunofluorescence microscopy, cells were grown on coverslips, treated as indicated and then fixed in 4% paraformaldehyde for 15 min at 37°C, permeabilized in 0.5% Triton X-100 for 5 min, and incubated with primary and secondary antibodies. Primary antibodies were rabbit anti-Rab7 at 1:200, rabbit anti-pS72-Rab7 at 1:250, rabbit anti-Halo at 1:500, anti-EEA1 at 1:100, anti-CD63 at 1:250, anti-LAMP1 at 1:500. Secondary antibodies were Alexa Fluor 488-, 555- or 647-goat anti-mouse-IgG or anti-rabbit-IgG antibodies (Invitrogen). Confocal microscopy was performed using an Olympus FV1000 microscope. Confocal images were captured at 0.5 µm intervals and z-stacks were processed with FV1000 software. Quantification of the colocalization of Rab7 or Halo–RILP with A647–EGF was carried out using an ImageJ plug-in (NIH) to generate a binary image of the pixels in each image. The ImageJ algorithm generated an automated threshold, and colocalization was quantified for pixels whose intensities were higher than the threshold. The number of both A647–EGF-positive vesicles and Rab7- or Halo–RILP-labeled A647–EGF double-positive vesicles per cell were counted. Data were plotted as the values obtained by dividing Rab7- or Halo–RILP-labeled A647–EGF-positive vesicles by the total number of A647–EGF-positive vesicles for each cell. For each series of experiments, the microscope settings were optimized for the brightest unsaturated images and remained unaltered during analysis. For comparisons among samples, we chose cells expressing relatively similar levels of GFP–LRRK1 (Y944F or K1243M mutants), GFP–Rab7, Flag–Rab7 (WT, CS, Q67L or Q67L/CS mutants) or Halo–RILP, based on the fluorescence intensities of GFP, Flag or Halo, respectively.

We thank M. Fukuda for helpful discussion.