Rab7 palmitoylation is required for efficient endosome-to-TGN trafficking

Retromer is an evolutionarily conserved multimeric complex first identified in the yeast Saccharomyces cerevisiae as responsible for the retrieval of a vacuolar protein sorting receptor, Vps10p, from the prevacuolar compartment to the trans-Golgi network (TGN) (Seaman et al., 1997, 1998). In yeast, retromer is a pentameric complex composed of a conserved trimer, Vps26p, Vps29p and Vps35p, which is responsible for cargo recognition (referred to as cargo selective complex, CSC), and a dimer of sorting nexins (SNXs), Vps5p and Vps17p (Horazdovsky et al., 1997). Yeast strains deleted of each of the retromer subunits are unable to correctly traffic the lysosomal enzyme carboxypeptidase Y (CPY) to the vacuole and secrete it into the extracellular space (Bonangelino et al., 2002). In humans, the retromer CSC is conserved (Haft et al., 2000), and can associate with a varying combination of SNXs, thereby expanding the range of cargos that can be recognized and trafficked (Cullen, 2008; Gallon and Cullen, 2015). In human cells, retromer is responsible for the retrieval of cargo receptors from the endosome to the TGN (Arighi et al., 2004; Seaman, 2004), and is required for the recycling of transmembrane proteins from the endosomal compartment to the plasma membrane (Burd and Cullen, 2014; Feinstein et al., 2011; Steinberg et al., 2013; Temkin et al., 2011). Increasing evidence has underlined how alterations in retromer expression or mutations in its subunits are associated with several neurodegenerative diseases, including Alzheimer's disease and Parkinson's disease (Follett et al., 2017; Small and Petsko, 2015).

Rab7 (also known as RAB7A) and its yeast homolog Ypt7 are responsible for the recruitment and stabilization of the retromer CSC at the endosome (Balderhaar et al., 2010; Burd and Cullen, 2014; Rojas et al., 2008; Seaman et al., 2009). Deletion of Ypt7 in yeast is associated with CPY secretion (Bonangelino et al., 2002) and cargo receptor accumulation at the endosome (Liu et al., 2012). In human cells, siRNA-mediated downregulation of Rab7 expression is associated with displacement of the CSC from the endosomal membrane to the cytosol, and with secretion of the lysosomal hydrolase cathepsin D (Rojas et al., 2008; Seaman et al., 2009).

In addition to its role in retromer recruitment, Rab7 also regulates endosome/lysosome fusion events via its interaction with Rab-interacting lysosomal protein (RILP) (Cantalupo et al., 2001). RILP, in turn, interacts with the endosomal sorting complex required for transport (ESCRT) system (Progida et al., 2006), mediating the internalization of mono-ubiquitinated transmembrane proteins into multivesicular bodies, leading to their degradation in lysosomes. Furthermore, Rab7 can regulate autophagy by modulating the interaction of lysosomes with autophagosomes (McEwan et al., 2015). How cells regulate the function of Rab7 in these various pathways is not well understood.

A RNAi screen designed to identify integral membrane proteins implicated in endosome-to-TGN trafficking found the palmitoyltransferase (PAT) DHHC5 (Breusegem and Seaman, 2014). PATs are transmembrane proteins characterized by a conserved catalytic domain containing a tetrapeptide of aspartic acid, histidine, histidine and cysteine (DHHC) (Mitchell et al., 2006), and mediate palmitoylation, a reversible attachment of a palmitate chain onto a cysteine of a target protein. Palmitoylation can serve as a reversible membrane anchor, promote localization of proteins to detergent resistant membranes, or modulate protein-protein interaction (Linder and Deschenes, 2007). Furthermore, two yeast PATs, Akr1 and Swf1, were identified in a screen for novel proteins implicated in CPY trafficking, suggesting that these enzymes play a critical role in lysosomal trafficking (Bonangelino et al., 2002).

Here, we show that Rab7 is palmitoylated on two cysteine residues. The nonpalmitoylatable Rab7 mutant, Rab7_C83,84S, localizes to membranes, suggesting that this modification is not required for membrane anchoring. Lack of Rab7 palmitoylation results in reduced interaction with retromer, leading to less efficient endosome-to-TGN cargo receptor trafficking, which results in the missorting of the lysosomal protein cathepsin D. Interestingly, Rab7 palmitoylation is not required for the degradation of the epidermal growth factor receptor (EGFR) or for the interaction between Rab7 and RILP, suggesting that this post-translational modification on Rab7 is required for its role in endosome-to-TGN trafficking.

To determine the role of palmitoylation in modulating endosome-to-TGN trafficking, we first asked if any retromer subunits are palmitoylated as this post-translational modification could act as a reversible membrane anchor for retromer CSC. Owing to the lack of an effective antibody for Vps29, we transfected HEK-293 cells with a FLAG-tagged form of Vps29 [Vps29-(FLAG)₃] and performed acyl resin-assisted capture (Acyl-RAC) analysis, a well-characterized method to test for protein palmitoylation (Forrester et al., 2011). None of the retromer CSC subunits were captured in our assay, but we recovered the well-known palmitoylated protein TRAPPC3 (Kummel et al., 2006), suggesting that none of the retromer CSC components are palmitoylated (Fig. 1A).

Rab7 is required to recruit retromer to endosomes (Rojas et al., 2008; Seaman et al., 2009), and was identified in a screen for palmitoylated neuronal proteins (Kang et al., 2008). We hypothesized that Rab7 palmitoylation could affect retromer endosomal recruitment by modulating the activity of this protein. We first tested endogenous Rab7 in HEK-293 cells for palmitoylation using Acyl-RAC and found that this small GTPase is palmitoylated (Fig. 1A), as it was captured in our assay when cells were treated with hydroxylamine (NH₂OH), which cleaves palmitate groups from proteins exposing a free cysteine(s) to interact with the thiopropyl Sepharose beads. Palmitoylation occurs on cysteine residues of proteins and we identified four evolutionarily conserved cysteine residues in Rab7 and Ypt7. The two C-terminal cysteines, 205 and 207 for mammalian Rab7, and 206 and 208 for yeast Ypt7, are prenylated (Magee and Newman, 1992), a post-translational modification that is required for membrane binding. No post-translational modifications on the other pairs of cysteine residues, 83 and 84 for Rab7, and 84 and 85 for Ypt7, have been described.

To identify the residues targeted by this post-translational modification, we used Myc-tagged Rab7 (Myc-Rab7) and mutated the various cysteines using site-directed mutagenesis. HEK-293 cells were transfected with wild-type Myc-Rab7 and the various mutants, and an Acyl-RAC assay was performed (Fig. 1B). We found that the single cysteine-to-serine mutation of residues 83 or 84 resulted in a decrease in palmitoylation compared to wild-type Rab7, while the double mutant Rab7_C83,84S was not recaptured, indicating that it is not palmitoylated (Fig. 1B). Mutation of cysteine 143, not present in Ypt7, did not affect Rab7 palmitoylation (Fig. 1B). Mutation of the two C-terminal cysteines (Rab7_C205,207S) resulted in a reduction in the palmitoylation level of Rab7 (Fig. 1B). Prenylation of C-terminal cysteines in Rab GTPases serves as a membrane anchor (Chavrier et al., 1991), so the reduction in palmitoylation observed in the prenylation mutants may be caused by the inability of the mutant protein to be membrane associated. Since all known PATs are integral membrane proteins, prenylation and membrane anchoring of Rab7 may be a prerequisite for palmitoylation.

To determine how palmitoylation modulates the function of Rab7, we first analyzed the membrane distribution of wild-type and mutant proteins. We transfected HEK-293 cells with wild-type Myc-Rab7 and the various mutants, and performed a membrane separation experiment to determine the amount of Rab7 on membranes. Our separation was successful as the cytosolic protein α-tubulin and the integral membrane protein Lamp-2 were found in the soluble and pellet fraction, respectively (Fig. 1C). As expected, wild-type Rab7 is mostly membrane bound while Rab7_C205,207S is localized to the soluble fraction (Fig. 1C). The membrane localization of the Rab7_C83S, Rab7_C84S and the nonpalmitoylatable mutant Rab7_C83,84S is comparable to that of wild-type Rab7, suggesting that palmitoylation is not regulating the ability of the protein to bind membranes (Fig. 1C). Statistical analysis of four separate experiments found no significant changes in membrane distribution between Myc-Rab7, Myc-Rab7_C83S, Myc-Rab7_C84S and the nonpalmitoylatable mutant (Myc-Rab7_C83,84S), while the distribution of the prenylation mutant (Myc-Rab7_C205,207S) was significantly different from that of the other proteins as it was almost exclusively cytosolic (Fig. 1D).

Palmitoylation can serve as a membrane anchor, or it can serve to localize proteins to specific membrane compartments (Rocks et al., 2005). To test if the palmitoylation of Rab7 modulates its intracellular distribution, we evaluated the localization of wild-type Myc-Rab7, Myc-Rab7_C83,84S and Myc-Rab7_C205,207S with the early endosomal marker Rab5 (RFP-Rab5) (Fig. 2A-I), as well as their localization with the late endosomal marker RILP (Fig. 2K-S). We found that Myc-Rab7 and Myc-Rab7_C83,84S partially colocalized with RFP-Rab5 (R_wt=0.2194, R_C83,84S=0.2554) (Fig. 2J) and more strongly with RILP (R_wt=0.4167, R_C83,84S=0.4832) (Fig. 2T), while Myc-Rab7_C205,207S had a negative Pearson's correlation with both RFP-Rab5 (R_205,207S=−0.02875) (Fig. 2J) and RILP (R_205,207S=−0.1222) (Fig. 2T) suggesting that this mutant was not found in early or late endosomes, as expected. Overall, our results indicate that palmitoylation is not required for the localization of Rab7 to endosomal membranes.

To test the function of Rab7 palmitoylation, we engineered a Rab7 knockout HEK-293 cell line (Rab7-KO) using CRISPR-Cas9 (Fig. 3A). The absence of endogenous Rab7 allows us to study the function of Rab7 palmitoylation without the background of any residual expression of endogenous wild-type protein (Fig. 3A). To determine whether Rab7 palmitoylation is required for the recruitment of retromer to endosomal membranes, we used a membrane separation experiment. Compared to HEK-293 or HEK-293 cells expressing Myc-Rab7, retromer CSC is displaced to the soluble fraction (cytosol) in Rab7-KO cells (Fig. 3B), as shown by western blotting (WB) for Vps26A and Vps35. This phenotype is rescued by the expression of wild-type Myc-Rab7, but not by Myc-Rab7_C205,207S expression. The expression of Myc-Rab7_C83,84S did not efficiently restore Vps26A membrane localization (Fig. 3B).

To support our membrane separation assay, we transfected Rab7-KO cells with Myc-Rab7, the nonpalmitoylatable mutant Myc-Rab7_C83,84S and the prenylation defective mutant Myc-Rab7_C205,207S, and examined the intracellular localization of the CSC subunit of retromer via Vps35 immunofluorescence (IF) (Fig. 4). In HEK-293 cells, Vps35 appears in punctate structures characteristic of endosomes (Fig. 4A, arrowheads). In cells not expressing Rab7 (Rab7-KO cells), we observed an almost complete loss of Vps35 endosomal staining (Fig. 4B, asterisks), consistent with previously published results (Rojas et al., 2008; Seaman et al., 2009). Expression of Myc-Rab7 in Rab7-KO cells rescued Vps35 localization to punctate structures (Fig. 4B, arrowheads). Quantification of the fluorescence intensity of Vps35 puncta revealed a significant increase in Rab7-KO cells expressing wild-type Myc-Rab7 compared to nontransfected cells. Interestingly, the expression of Myc-Rab7_C83,84S in Rab7-KO cells (Fig. 4C, arrowheads) partially rescued the membrane localization of Vps35, but quantification showed that the increase compared to Rab7-KO cells was not significant. Finally, expression of Myc-Rab7_C205,207S failed to rescue Vps35 localization, as expected, since this mutant is not membrane bound (Fig. 4D).

To better understand how Rab7 palmitoylation is required to recruit retromer, we tested how this post-translational modification affects the interaction between Rab7 and its effectors using bioluminescence resonance energy transfer (BRET), a method that detects molecular proximity in live cells. Wild-type Rab7 and Rab7_C83,84S were fused at the N-terminus to the energy donor Renilla luciferase II (RLucII) to obtain RLucII-Rab7 and RLucII-Rab7_C83,84S, while Vps26A and RILP were fused at the C-terminus with the energy acceptor green fluorescence protein 10 to obtain Vps26A-GFP10 and RILP-GFP10. To ensure that tagging Rab7 with RLucII (RLucII-Rab7) did not affect its membrane localization and function, we performed rescue experiments using membrane separation and IF microscopy to determine if RLucII-Rab7 could restore retromer CSC recruitment in Rab7-KO cells (Fig. S1). Recruitment to membranes was not impaired by the RLucII tag as RLucII-Rab7 was membrane bound (Fig. S1A). Furthermore, RLucII-Rab7 and Myc-Rab7 both rescued retromer CSC membrane recruitment in Rab7-KO cells, as shown by blotting for Vps26A and Vps35 following a membrane separation assay (Fig. S1A). To confirm these results, we expressed Myc-Rab7 or RLucII-Rab7 in Rab7-KO cells immunostained for Vps35 (Fig. S1B) and quantified the fluorescence intensity of Vps35 puncta in Rab7-KO cells (Fig. S1C, black bar), Rab7-KO cells expressing Myc-Rab7 (Fig. S1C, white bar) and Rab7-KO cells expressing RLucII-Rab7 (Fig. S1C, gray bar). Both Myc-Rab7 and RLucII-Rab7 were able to recruit retromer CSC to membranes as statistical analysis of the fluorescence intensity of Vps35 puncta showed a significant increase in both Myc-Rab7- and RLucII-Rab7-expressing cells compared to untransfected cells (Fig. S1C). The recruitment efficiency of Myc-Rab7 and RLucII-Rab7 were comparable as Vps35 intensity in the two conditions was not significantly different (Fig. S1C).

To determine whether palmitoylation plays a role in modulating interaction between Rab7 and retromer, HEK-293 cells were cotransfected with a constant amount of RLucII-Rab7 or RLucII-Rab7_C83,84S, and increasing concentrations of Vps26A-GFP10 or RILP-GFP10, to generate BRET titration curves (Kobayashi et al., 2009; Mercier et al., 2002). The BRET signal between RLucII-Rab7 and Vps26A-GFP10 rapidly increased with increasing amounts of expressed Vps26A-GFP10 until it reached saturation (Fig. 5A, black line), suggestive of a specific interaction (Mercier et al., 2002). By contrast, RLucII-Rab7_C83,84S displayed a lower BRET value with Vps26A-GFP10 that never fully reaches saturation (Fig. 5A, blue line). As a control, we tested the interaction of Rab1a (RLucII-Rab1a), which localizes to the Golgi complex, with retromer. As expected, we found no interaction, as demonstrated by a linear curve (Fig. 5A, red line). Conversely, we tested the interaction between Rab7 and the µ1 (µ1-GFP10) subunit of the AP-1 family of clathrin adaptors. Once again, we did not detect an interaction (Fig. 5A, green line), suggesting that the BRET curves between Rab7 and retromer represent a specific interaction. BRET₅₀ is the value at which the concentration of the acceptor is required to obtain 50% of the maximal BRET signal (BRET_MAX) and is indicative of the propensity for the donor and acceptor to interact (Kobayashi et al., 2009; Mercier et al., 2002). We calculated BRET₅₀ for the interaction between wild-type Rab7 or Rab7_C83,84S and Vps26A, and found a fourfold increase in the BRET₅₀ value for Rab7_C83,84S binding to Vps26A (Fig. 5A, black bar) compared to wild-type Rab7 (Fig. 5A, gray bar), suggesting a much weaker interaction. Interestingly, palmitoylation of Rab7 did not appear to modulate the Rab7/RILP interaction, as we found no significant changes in the BRET titration curves between Rab7 and RILP (Fig. 5B, black line), or Rab7_C83,84S and RILP (Fig. 5B, blue line), or the BRET₅₀ of these interactions (Fig. 5B, gray and black bars). As expected, Rab1a did not interact with RILP (Fig. 5B, red line). These data show that even if the palmitoylation-deficient mutant is still membrane bound, its interaction with Vps26A is significantly less efficient than that of wild-type Rab7. However, it would appear that not all interactions between Rab7 and effectors are modulated by palmitoylation, suggesting that this post-translational modification can mediate specific Rab7 functions.

To further explore how palmitoylation mediates the Rab7-retromer interaction, we performed a co-immunoprecipitation (co-IP) experiment by expressing wild-type Myc-Rab7, the nonpalmitoylatable mutant Myc-Rab7_C83,84S and (HA)₃-Vps35 separately in HEK-293 cells. We then mixed lysates from each of the Rab7-expressing cells with lysate from cells expressing (HA)₃-Vps35 and performed co-IP using anti-HA antibody. If palmitoylation modulates the conformation of Rab7 to enable an interaction with retromer, Myc-Rab7_C83,84S would not bind (HA)₃-Vps35 in this assay. On the other hand, if palmitoylation is required to localize Rab7 to facilitate an interaction with retromer, we would expect an interaction in this assay since the proteins are in a lysate and not membrane localized. We found an interaction between Myc-Rab7 and (HA)₃-Vps35, and between Myc-Rab7_C83,84S and (HA)₃-Vps35 (Fig. S2), suggesting that the role of palmitoylation is to localize Rab7 to a specific endosomal domain leading to an optimal interaction with retromer. As expected, (HA)₃-Vps35 also interacted with another retromer subunit, Vps26A (Fig. S2).

To determine whether palmitoylation is required to localize Rab7 to endosomal domains, thereby modulating the Rab7-retromer interaction, we transfected U2OS cells with Myc-Rab7 (Fig. 5C,D, green) or Myc-Rab7_C83,84S (Fig. 5E,F, green), to evaluate colocalization between these proteins and endogenous Vps26A (Fig. 5A-D, red) using structured illumination microscopy (SIM). Colocalization between Myc-Rab7 or Rab7_C83,84S and Vps26A was determined using Mander's coefficient. We determined the amount of Vps26A colocalizing with wild-type Rab7 or Rab7_C83,84S puncta, and observed reduced colocalization between Vps26A and Rab7_C83,84S (Fig. 5E,F,G, black bars) compared to Myc-Rab7 (Fig. 5C,D,G, white bars). We then compared the amount of Rab7 localized to Vps26A puncta between wild-type Rab7 and Rab7_C83,84S (Fig. 5G, white bars), and found a slight, but statistically significant, decrease in the degree of colocalization observed in the Rab7_C83,84S mutant. Indeed, Rab7 is required to recruit Vps26A to endosomes (Rojas et al., 2008), thus it is expected that Rab7 will be present in Vps26A-positive puncta on endosomes. However, these data suggest that loss of palmitoylation does not completely prevent interaction between Rab7 and retromer but, due to reduced interaction ability, nonpalmitoylatable Rab7 is not as efficient at recruiting retromer to endosomal membranes.

Since Rab7 palmitoylation is required for an optimal interaction with retromer CSC, we hypothesized that the absence of palmitoylation would result in decreased retromer function. In humans and yeast, altered retromer function is associated with impaired trafficking of lysosomal/vacuolar enzymes that results in their secretion into the extracellular space (Bonangelino et al., 2002; Follett et al., 2016; Rojas et al., 2008). We asked if Rab7 palmitoylation is required for the correct trafficking of the lysosomal enzyme cathepsin D. To test the effect of Rab7 palmitoylation on cathepsin D trafficking, we compared the secretion of cathepsin D in HEK-293 cells, Rab7-KO cells and Rab7-KO cells transiently expressing wild-type Myc-Rab7 and the various mutants (Fig. 6A). As expected, in HEK-293 cells, cathepsin D is correctly trafficked to the lysosome with almost no secretion of the protein into the medium (Fig. 6A). Rab7-KO cells are not able to efficiently recruit retromer to endosomes, resulting in inefficient trafficking of the lysosomal sorting receptors and the secretion of cathepsin D (Fig. 6A). The expression of Myc-Rab7 in Rab7-KO cells rescues cathepsin D trafficking; therefore, secretion is reduced compared to Rab7-KO cells (Fig. 6A). The expression of the prenylation mutant (Rab7_C205,207S) does not rescue Rab7-KO cells, resulting in secretion of cathepsin D (Fig. 6A). Expression of Myc-Rab7_C83,84S does not efficiently rescue cathepsin D secretion, suggesting that efficient lysosomal localization of this protease is dependent on Rab7 palmitoylation (Fig. 6A). Functionally, palmitoylation of Rab7 plays a significant role in modulating efficient endosome-to-TGN trafficking of the lysosomal sorting receptors and the proper localization of lysosomal proteins.

Although Rab7 is required for the spatiotemporal recruitment of retromer and therefore efficient endosome-to-TGN trafficking of CI-MPR (also known as IGF2R) and sortilin, it also regulates the degradation of integral membrane proteins such as EGFR (Shinde and Maddika, 2016; Vanlandingham and Ceresa, 2009). Upon stimulation with epidermal growth factor (EGF), EGFR is internalized and degraded in lysosomes (Futter et al., 1996). We stimulated cells that had been previously incubated for 1 h with cycloheximide and 100 ng/ml EGF for either 15 or 120 min and compared the degradation of EGFR in HEK-293, Rab7-KO and Rab7-KO cells expressing wild-type Myc-Rab7, Myc-Rab7_C83,84S or Myc-Rab7_C205,207S. Nonstimulated (NS) cells served as controls. As expected, HEK-293 cells and HEK-293 cells overexpressing Myc-Rab7 efficiently degrade EGFR following stimulation with EGF (Fig. 6B). However, compared to HEK-293 cells, Rab7-KO cells are not able to degrade EGFR as efficiently (Fig. 6B), while Rab7-KO cells expressing Myc-Rab7 can. Rab7-KO cells expressing Myc-Rab7_C83,84S appear to degrade some EGFR but with slower degradation kinetics (at 15 min chase) compared to HEK-293 cells, while Rab7-KO cells expressing Myc-Rab7_C205,207S also presented significant delay.

Although EGFR degradation was partially restored in Rab7-KO cells expressing Myc-Rab7_C83,84S, the degradation assay determines the amount of EGFR remaining in all cells, not just in cells that have been rescued. To support our EGFR degradation data, we tested the degradation of EGF labeled with Alexa Fluor 488 (EGF-488) so that we could evaluate degradation in rescued cells specifically. Cells were incubated with 300 ng EGF-488 for 30 min, washed and then chased for 0, 15 and 60 min. In HEK-293 cells, we found very little EGF-488 as it was efficiently degraded (Fig. 7A,F,K). Quantification of 20 cells revealed that, on average, HEK-293 cells contained 4.15, 4.35 and 3.2 EGF-488 puncta at 0, 15 and 60 min, respectively (Fig. 7P, white bars). In Rab7-KO cells, we observed a significant increase in the amount of EGF-488 puncta (Fig. 7B,G,L, asterisks), and quantification revealed 10.65, 10.85 and 9.8 EGF-488 puncta at 0, 15 and 60 min, respectively (Fig. 7P, black bars). Expressing wild-type Myc-Rab7 reduced the number of EGF-488 puncta in Rab7-KO cells (Fig. 7C,H,M) to levels comparable to HEK-293 cells (Fig. 7P, gray bars). Supporting our EGFR degradation data (Fig. 6B), expressing Myc-Rab7_C83,84S in Rab7-KO cells also decreased the number of EGF-488 puncta to levels comparable to HEK-293 cells (Fig. 7D,I,N). Quantification of 20 cells revealed no significant difference in the number of puncta (4.4, 5.65, 4.65 at 0, 15 and 60 min, respectively) in Rab7-KO expressing Myc-Rab7_C83,84S compared to HEK-293 cells (Fig. 7P, diagonal striped bars). Cells that do not express Myc-Rab7_C83,84S in the same field of view accumulated EGF-488 puncta (Fig. 7D,N, asterisks). As expected, expression of Myc-Rab7_C205,207S in Rab7-KO cells (Fig. 7E,J,O) had no effect on the number of EGF-488 puncta (Fig. 7P, horizontal striped bars) as these cells contained an average of 8.7, 9.5 and 10.7 puncta at 0, 15 and 60 min, respectively.

Rabs are key regulators of the formation, trafficking, and fusion of transport vesicles at the endoplasmic reticulum (ER), Golgi complex, and early and late endosomes (Hutagalung and Novick, 2011). Rab7 functions in many pathways at endosomes, including the recruitment of retromer (Rojas et al., 2008; Seaman et al., 2009) and the recruitment of RILP (Cantalupo et al., 2001), and mediates the fusion of lysosomes with autophagosomes via the homotypic fusion and protein sorting (HOPS) complex (McEwan et al., 2015). The molecular mechanisms that regulate and determine these various functions of Rab7 are not well understood. Here, we demonstrate that Rab7 is palmitoylated, and that this post-translational modification plays a role in mediating the function of Rab7 in retromer recruitment and therefore in regulating endosome-to-TGN traffic.

Rab7 and Ypt7 have four conserved cysteine residues. Two of these are the C-terminal cysteines that are prenylated, C205 and C207 in mammalian Rab7, and C206 and C208 in yeast Ypt7. As we have shown (Fig. 1C), these terminal residues are required for membrane anchoring. The other conserved cysteines are found between Switch I and Switch II, C83 and C84 in mammalian Rab7, and C84 and C85 in yeast Ypt7. We now demonstrate that these residues in Rab7 are palmitoylated. Rab7_C205,207S is not palmitoylated and we suggest that membrane anchoring is a prerequisite for palmitoylation. The palmitoyltransferases (PATs), of which 24 have been found in humans and seven in yeast, are all multispan integral membrane proteins (Linder and Deschenes, 2007). We propose that prenylation enables Rab7 to anchor into endosomal membranes. Subsequently, a subpopulation could be palmitoylated by one or more PATs. Further work will be required to identify the mechanism of Rab7 palmitoylation. The PAT DHHC5 is an excellent candidate as endosome-to-TGN trafficking was disrupted in cells depleted of DHHC5 by RNAi (Breusegem and Seaman, 2014). Furthermore, DHHC5 localizes to endosomes and could serve as a local mechanism for Rab7 palmitoylation.

Palmitoylation does not act as a membrane anchor for Rab7, but rather it modulates interactions. Using BRET, we found that Rab7_C83,84S bound significantly less retromer than wild-type Rab7. However, RILP bound equally well to both wild-type Rab7 and Rab7_C83,84S. It will be interesting to determine how palmitoylation of Rab7 mediates interaction with other effectors such as HOPS, Rubicon and PLEKHM1. Based on our data, we find that palmitoylation modulates Rab7 function and its interaction with retromer by localizing the protein into specific endosomal subdomains, and probably not by regulating its GTP-bound status.

It is becoming more apparent that the function of Rab7 is modified and regulated by post-translational modifications. We have demonstrated that Rab7 palmitoylation is required for the efficient endosome-to-TGN trafficking of the lysosomal sorting receptors. Recent studies have also shown how phosphorylation (Francavilla et al., 2016; Shinde and Maddika, 2016) and ubiquitination (Song et al., 2016) can regulate Rab7 function. Rab7 has at least two phosphorylation sites (serine 72 and tyrosine 183), which can act as a negative regulator or modulate function. Indeed, phosphorylation at serine 72 acts as an ‘off switch’ for Rab7 as the phosphomimetic mutant was completely cytosolic (Shinde and Maddika, 2016), while phosphorylation at tyrosine 183 is required to mediate the degradation of the EGFR (Francavilla et al., 2016). Here, we demonstrate that palmitoylation also regulates the function of Rab7. As far as we know, we are the first to confirm this post-translational modification on a mammalian Rab protein and the first to functionally characterize the role of palmitoylation on the function of a Rab. It will be interesting to confirm palmitoylation of other Rabs and elucidate how this modification would affect their function. Our work adds another regulatory mechanism to Rab7 biology. By combining phosphorylation, ubiquitination and palmitoylation, cells can modulate Rab7 function to various pathways and modulating these mechanisms could prove to be valuable targets in diseases in which Rab7 is implicated.

Unless otherwise stated, all reagents were purchased from Fisher Scientific. The following mouse monoclonal antibodies were used: anti-Myc 9E10 (MMS-150P, Cedarlane Laboratories), anti-Vps35 (ab57632, Abcam), anti-actin Ab5 (612657, BD Biosciences), anti-α-tubulin (T9026, Sigma-Aldrich), anti-Lamp-2 (ab25631, Abcam), anti-Renilla luciferase (MAB4400, Fisher Scientific) and anti-HA (16B12, Cedarlane Laboratories). The following rabbit polyclonal antibodies were used: anti-Vps26A (ab23892, Abcam), anti-TRAPPC3 (15555-1-AP, Cedarlane Laboratories), anti-RILP (ab128616, Abcam), anti-Myc (2272, Cell Signaling Technology) and anti-FLAG (F7425, Sigma-Aldrich). The following rabbit monoclonal antibodies were used: anti-Rab7 (D95F2, Cell Signaling Technology), and anti-EGFR (ab52894, Abcam). All antibodies were used at a dilution of 1:1000 for WB and IF except anti-Myc, which was used at 1:500 for IF, and Lamp-2, which was used at 1:500 for WB.

All restriction enzymes used for cloning were purchased from New England Biolabs. Vps29-(FLAG)₃ was generated by amplifying Vps29 cDNA from Vps29-YFP via PCR and cloned into BstbI/NotI of pEZ-M14 (GeneCopoeia). For Myc-Rab7, the cDNA of Rab7 was amplified from RFP-Rab7 (a gift from Ari Helenius, plasmid #14436, Addgene) via PCR and cloned into the XbaI/NheI sites of pKMyc (a gift from Ian Macara, plasmid #19400, Addgene). All the mutants described were generated via PCR mutagenesis using cloned PFU polymerase (Agilent Technologies). RLucII-Rab7 and RLucII-Rab7_C83,84S were generated by amplifying Rab7 cDNA from Myc-Rab7 and Myc-Rab7_C83,84S and cloned into the EcoRV/XhoI site of pcDNA3.1 Hygro (+) RLucII-GFP10-st2 plasmid. To obtain Vps26A-GFP10, Vps26A cDNA was amplified from Vps26A-YFP and cloned into the NheI/BamHI sites of pcDNA3.1 Hygro (+)-GFP10-RLucII. RLucII-Rab1a was generated by amplifying Rab1a cDNA from Myc-Rab1a (a gift from Dr Terry Hebert, McGill University, Montreal, Canada) and cloned into the EcoRV/XhoI sites of pcDNA3.1 Hygro (+) RLucII-GFP10-st2 plasmid. pcDNA3.1 Hygro (+) RLucII-GFP10-st2 and pcDNA3.1 Hygro (+) GFP10-RLucII-st2 plasmids were generous gifts from Dr Michel Bouvier (IRIC, Université de Montreal, Montreal, Canada). RILP-GFP10 was generated by amplifying RILP cDNA from HA-RILP (a generous gift from Cecilia Bucci, University of Salento, Lecce, Italy) and cloned into the NheI/KpnI sites of pcDNA3.1 Hygro (+)-GFP10-RLucII.

Cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and penicillin-streptomycin (Thermo Fisher Scientific). All cells were originally obtained from ATCC and periodically checked for contamination. All transfections, unless otherwise noted, were performed with polyethylenimine (PEI) (Thermo Fisher Scientific). Briefly, solution A was prepared diluting plasmid into Opti-MEM (Thermo Fisher Scientific). Solution B was prepared by diluting PEI (1 µg/µl) in Opti-MEM in a 1:3 ratio with the DNA to transfect. After 5 min incubation, the two solutions were mixed, vortexed for 3 s, incubated at room temperature (RT) for 15 min and added to the cells.

The protocol to detect palmitoylated protein was adapted from a published protocol (Ren et al., 2013). Briefly, protein lysates were incubated overnight at RT with 0.5% methyl methanethiosulfonate (MMTS) (Sigma-Aldrich) to block free cysteine residues. Proteins were then precipitated by adding two volumes of cold acetone and incubated at −20°C for 2 h. After washing with cold acetone, the pellet was resuspended in binding buffer (100 mM HEPES, 1 mM EDTA, 1% SDS). Water-swollen thiopropyl Sepharose 6B (GE Healthcare Life Sciences) was added and samples were divided into two equal parts. One part was treated with hydroxylamine (Sigma-Aldrich), pH7.5, to a final concentration of 0.2 M to cleave palmitate residues from proteins; the other part was treated with an equal amount of NaCl as a control. After 3 h incubation at RT, beads were washed five times with binding buffer and captured proteins were eluted with 75 mM DTT.

24 h post-transfection, cells were harvested, snap frozen with liquid nitrogen and allowed to thaw at room temperature. Samples were resuspended in buffer 1 (0.1 M Mes-NaOH, pH 6.5, 1 mM MgAc, 0.5 mM EGTA, 200 µM sodium orthovanadate, 0.2 M sucrose) and centrifuged at 10,000 g for 5 min at 4°C. The supernatant containing cytosolic proteins was collected (S, soluble fraction) and the pellet was resuspended in buffer 2 (50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100) and spun at 10,000 g for 5 min at 4°C. The supernatant containing membrane proteins was collected (P, pellet fraction) for further analysis. Equal volumes of each fraction were loaded for WB. The intensity of the bands on blots was determined using Fiji (Schindelin et al., 2012). Analysis of the distribution of proteins between the two fractions was performed by dividing the intensity of each fraction by the total in each condition and expressed as a percentage.

HEK-293 were transfected with an all-in-one CRISPR plasmid for Rab7 (plasmid #HTN218819, GeneCopoeia). 72 h post-transfection, cells were treated with 1 mg/ml Geneticin (Thermo Fisher Scientific) for 1 week. Limiting dilution was performed to isolate single cells. Single clones were allowed to grow for approximately 2 weeks prior to testing via immunoblotting to identify Rab7-KO clones.

HEK-293 or U2OS cell IF was performed as previously described (Dumaresq-Doiron et al., 2010). Briefly, cells were seeded on glass coverslips overnight and subsequently transiently transfected with indicated plasmids. 48 h post-transfection, cells were washed once in PBS, fixed with 4% paraformaldehyde in PBS for 15 min at RT and washed twice with PBS. Cells were permeabilized with 0.1% Triton X-100, 1% bovine serum albumin (BSA) (Fisher Scientific) in PBS for 5 min at RT, washed once with PBS, and incubated for 2 h with primary antibodies in a solution of 0.1% BSA in PBS. Cells were then washed three times in PBS and incubated for 1 h at RT with appropriate secondary antibodies conjugated to either Alexa Fluor 594 (Thermo Fisher Scientific) or Alexa Fluor 488 in a solution of 0.1% BSA in PBS. Cells were washed once with PBS and incubated with DAPI (Thermo Fisher Scientific) in PBS for 5 min at RT. Cells were then washed three times with PBS and mounted on glass slides with Fluoromount G (Fisher Scientific). Quantification of fluorescence intensity was performed using Fiji software by manually outlining nonrescued or rescued Rab7-KO cells. Quantification of colocalization was performed using the Fiji plugin Coloc2 by manually outlining the double-labeled cells.

BRET assay was performed as previously described (Gales et al., 2006; Kobayashi et al., 2009). Briefly, HEK-293 cells were seeded in 12-well plates (Fisher Scientific) and transfected with the indicated plasmids. 48 h post-transfection, cells were washed once with PBS, detached with 5 mM EDTA in PBS and resuspended in 500 µl PBS. Cell suspensions were distributed into 96-well plates (Fisher Scientific); each sample was plated in triplicate. BRET signal was measured after the addition of the RLucII substrate DeepBlueC coelenterazine (Cedarlane Laboratories) to a final concentration of 5 µM. Readings were performed using an Infinite M1000Pro (Tecan). The BRET signal was calculated as a ratio between GFP10 emission (500–535 nm) over RLucII emission (370–450 nm). The BRET net signal was calculated as the difference between the total BRET signals and the one obtained from experiments in which only the RLucII is expressed. The expression level of Vps26A-GFP10 was measured by reading the emission of fluorescence at 510 nm after exciting the fluorophore at 400 nm. The expression level of RLucII was measured by reading the emission at 370–450 nm after the addition of the substrate. BRET signal was plotted as function of GFP10 fluorescence over RLucII luminescence (GFP10/RLuc).

U2OS cells were seeded in eight-well LabTek #1.5 borosilicate chambered coverglass imaging dishes (Fisher Scientific) to 50–60% confluency. Wild-type Myc-Rab7 and Myc-Rab7_C83,84S were transiently expressed for 36 h in separate wells. Cells were then fixed with 2% paraformaldehyde (PFA) at 22°C for 20 min. Fixed cells were washed three times with 25 mM ammonium chloride (NH₄Cl) and 50 mM glycine to remove excess PFA and reduce autofluorescence, and then three times with PBS. The cells were permeabilized with 0.1% (v/v) Triton X-100 in PBS and blocked with 10% FBS (v/v) in PBS. Immunostaining was performed using anti-Myc and anti-Vps26A antibodies. An anti-mouse Alexa Fluor 488-conjugated secondary antibody (Thermo Fisher Scientific) and an anti-rabbit Alexa Fluor 647-conjugated secondary antibody (Thermo Fisher Scientific) were then used to indirectly label proteins with green and far-red fluorescence, respectively. After immunostaining, cells were washed three times with PBS prior to imaging. Imaging of samples was performed using SIM on a Zeiss Elyra PS.1 superresolution inverted microscope. Samples were imaged at an effective magnification of 101× (63× objective+1.6× optovar tube lens) on an oil immersion objective. Typically, 10–20 slices (0.110 μm) were captured for each field of view for an imaging volume of ∼1.1–2.2 μm. A 647 nm (and subsequently 488 nm) laser line was directed into the microscope optical train via a multimode fiber coupler. The laser was passed through a diffraction grating, and a series of diffraction orders (−1, 0, +1) were projected onto the back focal plane of the objective. These wavefronts were collimated in the objective to create a 3D sinusoidal illumination pattern on the sample. The diffraction grating was then rotated and translated throughout the acquisition to create patterned offset images containing higher spatial frequency information compared to widefield imaging. Five lateral positions were acquired at each of three diffraction grating rotations (120°) for a total of 15 raw images per slice. SIM imaging with both the 488 nm and 647 nm laser was performed at 50 ms exposures with laser power varying between 3 and 10%, and a gain level of 60–80. Raw SIM image stacks were processed with the Structured Illumination toolbar of the Zeiss Zen software. A series of parameters were set to generate an optical transfer function (OTF) used for 3D reconstruction. The noise filter for Wiener de-convolution was set to a value of −6.0, and the maximum isotropy option was left unselected to recover all available frequency information at exactly the 120° rotation angles. Super-resolution frequency weighting was set at 1.0. Negative values arising as an artifact of the Wiener filter were clipped to zero using the Baseline Cut option. Sectioning filters used to remove the 0 frequencies from the 0th order and nonshifted 1st order (+1, −1) were set to 100 and 83, respectively. Processed SIM images were then aligned via an affine transformation matrix of predefined values obtained using 100 nm multicolor Tetraspeck fluorescent microspheres (Thermo Fisher Scientific). Dual channel-aligned images were analyzed using the colocalization function in Zen. Regions of interest were drawn in the cytoplasm (omitting the nucleus) of 10–20 cells for each of the wild-type and mutant Rab7 cell types. The average Manders colocalization coefficients over the imaging volume were calculated for both channels. The average Manders colocalization coefficients and standard deviations were calculated across the sample size for wild-type and all mutants.

A cathepsin D secretion assay was performed as previously described (Follett et al., 2016). Briefly, HEK-293 and Rab7-KO cells were transfected with the indicated plasmids. 48 h post-transfection, cells were incubated in serum-free medium (Opti-MEM) supplemented with 100 µg/ml cycloheximide and chased for 4.5 h. The medium was collected, precipitated with trichloroacetic acid, and the pellet was resuspended in Laemmli sample buffer. Cells were collected, and lysed in TNE (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 2 mM EDTA, 0.5% Triton X-100 and protease inhibitor cocktail) for 30 min on ice as previously described (Dumaresq-Doiron et al., 2013). The precipitated proteins collected from the medium and the cell lysate were then analyzed by WB.

HEK-293 or Rab7-KO cells were transfected with Myc-Rab7 and the various Rab7 mutants. 48 h post-transfection, the serum-starved cells were treated with 50 μg/ml cycloheximide for 1 h to prevent de novo synthesis of EGFR during EGF stimulation. Cells were subsequently stimulated with 100 ng/ml EGF for the indicated times, harvested and lysed in TNE as described above. The levels of EGFR were determined by WB.

Cells were seeded on coverslips. 24 h later, the cells were transfected with wild-type Myc-Rab7 or mutant constructs. 24 h post-transfection, cells were serum-starved in Opti-MEM for 1 h followed by a 30-min pulse of EGF-488 (ThermoFisher) at a concentration of 300 ng/ml. Cells were then washed with PBS and fixed in 4% paraformaldehyde at the following chase time points: 0, 15 or 60 min. Cells were immunostained with anti-Myc primary and Alexa Fluor 594-conjugated secondary antibodies. Cells were imaged using a Zeiss LSM 780 confocal microscope. The number of puncta per cell was counted manually (20 cells per condition for each time point).

Statistical analysis was performed using GraphPad Prism Version 7. Statistical tests used are described in the corresponding figure legends.

We thank Dr Peter J. McCormick (University of Surrey, UK) for critical reading of the manuscript and helpful discussions.