An Interorganellar Bidding War: Vps13 Localization by Competitive Organelle-Specific Adaptors

Abstract

Membrane contact sites are regulated through the controlled recruitment of constituent proteins. Yeast vacuolar protein sorting 13 (Vps13) dynamically localizes to membrane contact sites at endosomes, vacuoles, mitochondria, and the endoplasmic reticulum under different cellular conditions and is recruited to the prospore membrane during meiosis. Prior to our recent work, the mechanism for localization at contact sites was largely unknown. We identified Ypt35 as a novel Vps13 adaptor for endosomes and the nucleus-vacuole junction. Furthermore, we discovered a conserved recruitment motif in Ypt35 and found related motifs in the prospore membrane and mitochondrial adaptors, Spo71 and Mcp1, respectively. All three adaptors compete for binding to a six-repeat region of Vps13, suggesting adaptor competition regulates Vps13 localization. Here, we summarize and discuss the implications of our work, highlighting key outstanding questions.

Membrane contact sites (MCSs) are critical for nonvesicular transport of lipids and ions and are tightly regulated through controlled recruitment of tethers and lipid transport proteins. Crosstalk between some MCSs, such as the endoplasmic reticulum (ER)-mitochondrial encounter structure (ERMES) and vacuole-mitochondrial patches (vCLAMPs), allows one site to compensate for loss of another (Elbaz-Alon et al., 2014). Furthermore, some MCS proteins have been observed at multiple different contact sites, suggesting yet another mode of crosstalk. How these shared proteins are selectively targeted to diverse membranes has been unclear.

Yeast vacuolar protein sorting 13 (Vps13) is a large, widely conserved protein that has a suggested lipid transfer function (Kumar et al., 2018) and dynamically localizes to vCLAMPs, nucleus-vacuole junctions (NVJs), endosomes, and prospore membranes. Vps13 localization is regulated by nutrient availability. When glucose is abundant, localization is predominantly endosomal, while under glucose-limiting conditions, Vps13 relocalizes to the NVJ (Lang, Peter, Walter, & Kornmann, 2015; Park et al., 2016). Additionally, Vps13 gain-of-function mutants that localize to vCLAMPs rescue defects of ERMES mutants, and loss of both VPS13 and ERMES is synthetic lethal. This suggests that Vps13 may compensate for loss of ERMES by acting at vCLAMPs (Lang et al., 2015). Our recent work uncovered how Vps13 is targeted to different membranes and suggests a regulatory mechanism that may be applicable to other MCS proteins (Bean et al., 2018).

Pursuing a hit from a genomic screen, we identified the sorting nexin Ypt35 as an adaptor that recruits Vps13 to endosomes and the NVJ. The NVJ localization of Ypt35 and Vps13 was mutually dependent, suggesting Ypt35 recruits Vps13 to the vacuole membrane, while Vps13 may bind another factor at the ER. We mapped the Vps13-interacting region of Ypt35 to a hydrophobic “PxP” motif centered on two conserved prolines (Figure 1(a)). This motif, which was both necessary and sufficient for Vps13 recruitment, bound a central region of Vps13.

Figure 1.

Vps13 localization by organelle-specific adaptors. (a) WebLogo of the PxP interaction motif from Saccharomycetales-level homologs of Ypt35, Spo71, and Mcp1. The consensus is shown with Φ indicating hydrophobic residues. (b) Ypt35, Spo71, and Mcp1 use PxP motifs to bind the VAB domain and compete to recruit Vps13 from a cytosolic pool to endosomes/vacuole, the prospore membrane, and mitochondria, respectively. Yeast Vps13 may interact with the ER through an additional unidentified adaptor, enabling a lipid transfer function (gray arrows) similar to human VPS13A and VPS13C (Kumar et al., 2018). The presence of Vps13 at contact sites that link mitochondria or endosomes with the ER (dotted lines) is speculative. (c) Schematic of Saccharomyces cerevisiae Vps13 highlighting the six-repeat VAB domain and conserved asparagines. ER = endoplasmic reticulum; Vps13 = vacuolar protein sorting 13.

We found that the prospore membrane adaptor Spo71 (Park, Okumura, Tachikawa, & Neiman, 2013) also bound this region of Vps13 suggesting it may contain a similar Vps13 recruitment motif. We identified putative PxP motifs in the yeast proteome including in Spo71 and Mcp1, the adaptor that recruits Vps13 to mitochondria (Peter et al., 2017) and demonstrated that both hits represent bona fide PxP motifs required for Vps13 recruitment.

All three adaptors bind the same region of Vps13 in a PxP motif-dependent manner. We propose this region, which consists of a six-repeat sequence, be called the Vps13 adaptor binding (VAB) domain. As all three adaptors bind this domain, we tested if they compete for Vps13 recruitment. Overexpression of any motif blocked the ability of Ypt35 or Mcp1 to recruit Vps13 to endosomes or mitochondria, respectively. Furthermore, high levels of Ypt35 or Spo71 reduced the ability of overexpressed Mcp1 to rescue the mitochondrial defects of ERMES mutants. These findings indicate that competing organelle-specific adaptors regulate Vps13 localization (Figure 1(b)).

The precise manner in which this competition regulates localization is unclear. The different PxP motifs appear to possess different binding strengths with Spo71 being the strongest and Mcp1 the weakest. This may explain why Vps13 localization matches that of Ypt35 in mitotic cells but during sporulation, when Spo71 is expressed, Vps13 relocalizes to the prospore membrane. Regulation of motif exposure could further control Vps13 localization. The degree to which each mechanism influences the activity of a given adaptor in response to external stimuli remains to be tested.

Deletion of a specific adaptor is predicted to cause a subset of the phenotypes found in vps13 knockouts. Indeed, deletion of SPO71 or VPS13 results in sporulation defects (Park et al., 2013) and deletion of MCP1 or VPS13 causes synthetic lethality with ERMES mutants (Peter et al., 2017). Interestingly loss of VPS13, but not YPT35, causes carboxypeptidase Y (CPY) missorting, which may reflect altered Golgi-endosome recycling of its receptor Vps10 (Brickner & Fuller, 1997). This suggests an as-yet unidentified adaptor may function in CPY sorting by recruiting Vps13 to Golgi or endosomal membranes.

The VAB domain is present in the four human VPS13 proteins (VPS13A–D), but it is unclear if its adaptor binding role is conserved. Recent work identified VPS13A and VPS13C as lipid droplet-ER MCS proteins that also localize to contact sites between the ER and mitochondria or endolysosomes, respectively (Kumar et al., 2018). The VPS13A-mitochondrial interaction is independent of the VAB domain and instead relies on amphipathic helices that also drive lipid droplet interactions. In contrast, the VAB domain is required for VPS13C recruitment to endolysosomes, but its binding partner is unknown. The adaptor binding role of some VAB domains may have been lost as human VPS13 proteins diverged to allow targeting to distinct membranes. Kumar et al. further demonstrated that VPS13A and VPS13C have FFAT motifs that bind VAMP-associated (VAP) proteins to mediate interactions with the ER. Whether yeast Vps13 is targeted to the ER through a related FFAT motif, or by binding to lipids or other proteins at the ER or NVJ remains to be determined.

Mutations in each of the human proteins cause distinct neurological diseases (Kumar et al., 2018). Each approximately 100-residue repeat within the VAB domain contains an invariant asparagine (Figure 1(c)). Of importance, mutations of these conserved asparagines in VPS13B (N2993S) and VPS13D (N3521S) are causative for Cohen syndrome and spastic ataxia, respectively (Gauthier et al., 2018; Kolehmainen et al., 2003). These mutations may mislocalize VPS13 by reducing adaptor binding or cause global folding defects; our ongoing studies seek to clarify the effect of these mutations.

The recent work opens several avenues for investigation. Further characterization of the adaptor binding region(s) of yeast Vps13 could determine if there is any discrimination between PxP motifs. Controlling localization of yeast Vps13 by ablating competing PxP motifs could identify which Vps13 functions are associated with given membranes. Additional adaptors might be found by screening PxP motif-containing proteins for interaction with Vps13. Finally, it will be interesting to test if the competition-based model for Vps13 localization applies to other MCS proteins.