Vps41 functions as a molecular ruler for HOPS tethering complex-mediated membrane fusion

Organelles of the endomembrane system in eukaryotic cells are connected by vesicles and carriers, which form at a donor membrane and fuse with the acceptor organelle. They contain specific membrane lipids and peripheral as well as integral membrane proteins, coding for the identity of each organelle (Borchers et al., 2021; Gomez-Navarro and Miller, 2016; Posor et al., 2022; Settembre and Perera, 2023). Among organelle-specific proteins, Rab GTPases guide membrane fusion events through interaction with different effector proteins (Bagde and Fromme, 2023; Borchers et al., 2021; Hutagalung and Novick, 2011; Müller and Goody, 2018). Rabs function as molecular switches. In their GDP-bound form, Rabs are substrates of the guanine nucleotide dissociation inhibitor (GDI) chaperone, which binds both the N-terminal GTPase domain and the C-terminal prenyl anchor of Rabs. GDI can thus extract the Rab–GDP from membranes and keep the protein soluble in the cytosol. On membranes, guanine nucleotide exchange factors trigger GTP loading of Rabs, which can bind specific effector proteins, thus facilitating their cellular function. For inactivation, specific GTPase-activating proteins bind to the Rab and trigger GTP hydrolysis. The Rab–GDP is then extracted by GDI from the membrane (Barr, 2013; Borchers et al., 2021; Hutagalung and Novick, 2011).

For membrane fusion, membrane-bound Rab–GTP interacts with a specific tethering factor or tethering complex, which binds both donor and target membranes (van der Beek et al., 2019; Lepore et al., 2018; Spang, 2016; Ungermann and Kümmel, 2019; Zhang and Hughson, 2021). Along the secretory pathway, α-helical CATCHR complexes function at the endoplasmic reticulum, Golgi and plasma membrane (Baker and Hughson, 2016; Khakurel and Lupashin, 2023; Zhang and Hughson, 2021). Within the endolysosomal pathway, two large hexameric complexes, CORVET and HOPS, are required for membrane tethering and fusion (van der Beek et al., 2019; Spang, 2016; Ungermann and Kümmel, 2019; Wickner and Rizo, 2017). In the yeast Saccharomyces cerevisiae, CORVET binds to the Rab5-like GTPase Vps21 on early endosomes, whereas HOPS binds similarly to the Rab7-like GTPase Ypt7 at late endosomes and vacuoles (Markgraf et al., 2009; Peplowska et al., 2007; Perini et al., 2014; Seals et al., 2000; Wurmser et al., 2000).

We recently solved the structures of the CORVET and HOPS complexes (Shvarev et al., 2022, 2024). Both complexes harbor five subunits with N-terminal β-propeller and C-terminal α-solenoid domains, and Vps33 as a shared Sec1/Munc18 (SM) protein, which binds to soluble N-ethylmaleimide-sensitive factor-attachment protein receptors (SNAREs) during fusion (Baker et al., 2015; Ungermann and Kümmel, 2019; Zhang and Hughson, 2021). They share a core of four subunits (Vps11, Vps18, Vps16 and Vps33) but differ in their Rab-specific subunits. In yeast, the CORVET subunits Vps8 and Vps3 both bind to Vps21 on early endosomes, whereas HOPS interacts with Ypt7 through Vps41 and Vps39 (Peplowska et al., 2007; Seals et al., 2000; Wurmser et al., 2000). In both complexes, Vps11 and Vps18 interact in an antiparallel manner via their α-solenoid domains, forming a central backbone. The SNARE-binding domains are attached laterally to the Vps18–Vps11 backbone, and the Rab-binding domains emerge on opposing ends of the complex. Whereas in CORVET, Rab-binding Vps8 interacts tightly with Vps11 of the backbone, the corresponding subunit Vps41 in HOPS is stretched out, exhibiting strong flexibility.

At physiological concentrations, HOPS and Ypt7 are essential for fusion, suggesting that HOPS is both a tether and fusion catalyst (Langemeyer et al., 2020; Zick and Wickner, 2016). Although fusion assays revealed the importance of HOPS within the fusion cascade, its precise molecular function during fusion remains enigmatic. N-terminal deletions of the β-propeller domains in several subunits rendered HOPS inactive in an in vitro fusion assay (Behrmann et al., 2014; Lürick et al., 2017; Shvarev et al., 2022). For Vps41, such a deletion also abolished tethering due to the removal of the Ypt7-binding site (Lürick et al., 2017; Shvarev et al., 2022).

In our structure of HOPS, the Rab-interacting β-propeller of Vps41 extends far away from the core, and the connecting α-solenoid is highly flexible (Shvarev et al., 2022). In addition, the identified amphipathic helix in the Vps41 β-propeller (Cabrera et al., 2010) pointed away from the membrane in our model, in contrast to its predicted membrane interaction during tethering events. As this is distinct from the homologous region in CORVET, we asked whether the flexible connection of the Vps41 β-propeller to the core of the complex is required for function, and if the extension essentially acts as a molecular ruler for accurate tethering.

Here, we show that HOPS function in fusion requires a defined distance for the correct positioning of the Vps41 β-propeller relative to the core. HOPS with a shortened or extended α-solenoid region in Vps41 is still capable of tethering Ypt7-decorated liposomes, but it is unable to facilitate fusion. Our data thus suggest that the length rather than the architecture of the Vps41 α-solenoid is essential for HOPS-mediated membrane fusion.

Our recent structure revealed that the Vps41 subunit and, in particular, its N-terminal β-propeller, which binds the Rab7-like Ypt7, extends far away from the core of the HOPS complex (Fig. 1A). We first sought to determine whether flexibility and extension of Vps41 is critical for maintaining the architecture of HOPS. We generated a Vps41 mutant lacking 113 amino acids in the α-solenoid (Vps41^Δ527–640; schematic of the mutation is shown in Fig. 1A, right panel), which corresponds to a shortening by 33 Å, and inserted the respective plasmid into the HOPS overexpression strain. We then isolated the HOPS complex, determined its composition and mass, and analyzed its structure by negative stain electron microscopy (Fig. 1B–E). This displayed the truncated α-solenoid of Vps41, while the overall architecture of HOPS was retained (Fig. 1F). As expected, in contrast to the wild-type complex, a compact density was visible close to the HOPS core and Vps11 β-propeller in the mutant complex, indicating that the rest of the Vps41 α-solenoid and β-propeller were now brought closer to the HOPS core (Fig. 1F, violet arrowheads).

We next asked whether the relative positioning of the Vps41 β-propeller to the core is critical for HOPS function in vivo. In addition to Vps41^Δ527–640, two other mutants were designed to either shorten the α-solenoid domain ∼4–5 nm while maintaining the orientation of the β-propeller (Vps41^Δ527–710, similarly to Vps41^Δ527–640), or to extend the distance up to 6 nm by inserting a flexible 20 amino acid GSGS linker (Vps41^GS-linker) (Fig. 2A). All constructs were expressed under the endogenous VPS41 promoter as the only copy of Vps41 in cells.

As a determinant of the in vivo function of Vps41, we initially analyzed the vacuolar morphology using the lipophilic dye FM4-64 to stain the vacuolar rim. In the absence of Vps41, vacuoles were highly fragmented with several dot-like structures, whereas wild-type cells had one to three vacuoles (Fig. 2B,C). In contrast, the two truncation mutants (Vps41^Δ527–640 and Vps41^Δ527–710) had five or more vacuoles, whereas the Vps41^GS-linker mutant showed an intermediate phenotype with on average three vacuoles. To further test for physiological defects, we grew cells on plates containing increasing concentrations of ZnCl₂, a stressor of the endolysosomal pathway (Montoro et al., 2021). Cells lacking Vps41 grew normally on control plates, but poorly in the presence of 3 mM ZnCl₂ (Fig. 2D). Strikingly, both Vps41 truncation mutants (Vps41^Δ527–640, Vps41^Δ527–710) showed much stronger growth defects than the vps41 deletion, already on control plates. In contrast, the Vps41^GS-linker mutant exhibited better growth than the deletion (Fig. 2D). To verify that the observed growth phenotype on control plates (Fig. 2D) is due to the impaired HOPS complex, we repeated this experiment using a VPS39-deletion (vps39Δ) background strain to test each of the generated mutants. In the vps39Δ background, the Vps41^Δ527–710 and the Vps41^GS-linker mutants behaved like the vps41 deletion (Fig. 2E), suggesting that the observed growth defect can indeed be attributed to a defect in the HOPS complex itself. The most severe Vps41^Δ527–640 mutant also recovered, particularly in the presence of ZnCl₂, although it still showed some growth defects on control medium (Fig. 2D,E). This suggests that the Vps41^Δ527–640 mutant protein causes additional cellular defects even in the absence of HOPS, possibly by sequestering Ypt7 from its other effectors (kleine Balderhaar et al., 2010; Liu et al., 2012; Numrich et al., 2015).

To test whether the mutants are dominant, we mated all mutant strains with wild-type yeast and recovered normal vacuole morphology, concluding that the phenotype we observed is not dominant if one wild-type copy of Vps41 is present in cells (Fig. 3A,B). We then asked whether the mutant Vps41 proteins would affect HOPS function completely, if they were present as the only copy in haploid cells (Fig. 2B). A simple test for this is the exposure of cells to hypotonic conditions such as water, which triggers rapid fusion of vacuoles (Bonangelino et al., 2002; LaGrassa and Ungermann, 2005; Zieger and Mayer, 2012). Under these conditions, wild-type cells formed one to two large vacuoles, which fragmented if cells were again subjected to hypertonic medium containing 0.9 M NaCl. Mutant cells behaved similarly to wild-type cells, even though the overall number of vacuoles remained larger (Fig. 3C,D). This indicates that the mutant Vps41 proteins can still support HOPS-dependent fusion in vivo to some extent.

As HOPS functions not only in vacuole–vacuole fusion, but in all fusion events at the vacuole, we reasoned that an impairment of HOPS function could result in an altered intracellular localization of the complex. We therefore used 2×mNeonGreen to tag the HOPS-specific Vps39 or Vps41 subunits C-terminally in wild-type and vps41 mutant strains and analyzed their localization relative to FM4-64-stained vacuoles. In all mutants, more Vps41- and Vps39-positive dots were present on FM4-64-labeled structures. These observed numbers of dots correlated with the increased number of vacuoles in these mutants (Fig. 4A–C).

To test whether the Vps41 mutations caused a defect in the endosomal system, we followed mNeonGreen-tagged Ivy1 and Vps21 as partially endosomally localized proteins, and Sec7 as a Golgi marker. Similar numbers of dots were observed in wild-type and Vps41 mutants (Fig. 4D; Fig. S2C,D), indicating that the Vps41 mutations did not cause a defect of the late Golgi or endosomes but had a specific fusion deficiency at the vacuole. In agreement, N-terminal tagging of Ypt7 with mNeonGreen, which has no effect on vacuole morphology in wild-type cells (Füllbrunn et al., 2024) but can respond to HOPS defects (Behrmann et al., 2014), resulted in vacuole fragmentation in all vps41 mutant strains (Fig. 4D; Fig. S4A). We thus conclude that the Vps41 mutations affect HOPS function, resulting in altered vacuole morphology but not affecting other endosomal proteins.

To directly test whether HOPS with a truncated or more extended Vps41 linker is still able to support fusion, we generated the corresponding overexpression strains and isolated mutant HOPS complexes with Vps41^Δ527–710 or Vps41^GS-linker, similarly to HOPS with Vps41^Δ527–640 (Fig. 1B,D). All complexes had the same stoichiometry as wild-type HOPS (Fig. 5A). Initially, we tested whether mutant HOPS complexes can support tethering. For this, fluorescently labeled liposomes without or with GTP-loaded and prenylated Ypt7 were incubated with wild-type or mutant HOPS (D'Agostino et al., 2017; Lürick et al., 2017; Shvarev et al., 2022). The reaction mixture was then centrifuged after incubation to pellet tethered liposomes, and the remaining fluorescence in the supernatant was measured (Shvarev et al., 2022, 2024). Mutant HOPS complexes were able to tether liposomes in vitro, although at reduced efficiency compared to the wild-type complex (Fig. 5B). To assess fusion efficiency, we isolated vacuoles lacking either the major vacuolar protease Pep4 (pep4Δ) or the vacuolar alkaline phosphatase (pho8Δ), which requires proteolytic cleavage by Pep4 in the vacuole for activity, from a vps11-1 temperature-sensitive mutant yeast strain. When vacuoles from the two tester strains fuse, the immature alkaline phosphatase of pep4Δ is processed by the protease present in vacuoles isolated from pho8Δ cells. The activity of mature alkaline phosphatase is measured by a colorimetric assay as a measure of fusion (Haas, 1995) (see Materials and Methods). Vacuoles from the vps11-1 mutant require HOPS addition for fusion (Bröcker et al., 2012; Lürick et al., 2017; Ostrowicz et al., 2010; Stroupe et al., 2006). The addition of increasing concentrations of wild-type HOPS to the isolated vacuoles reconstituted the fusion activity of vps11-1 vacuoles as expected, whereas mutant complexes only poorly rescued fusion activity (Fig. 5C). As the Vps41^Δ527–640 mutant cells had the strongest growth defect (Fig. 2C), we wondered whether the corresponding HOPS complex might inhibit the fusion of wild-type vacuoles by competing with endogenous HOPS. To test this, we compared the activity of isolated Vps41^Δ527–640 mutant HOPS with wild-type HOPS in vitro (Fig. 5D). When wild-type HOPS was added to the fusion assay, fusion activity initially increased at HOPS concentrations up to 30 nM. Further addition of HOPS inhibited fusion, as too much HOPS likely titrated out Ypt7 on one of the membranes to be tethered. In contrast, HOPS with Vps41^Δ527–640 strongly inhibited fusion at any concentration. We conclude that HOPS with truncated Vps41 can support tethering but is unable to promote productive fusion in vitro and could indeed compete with endogenous HOPS growth experiments (Fig. 2C).

Our data suggest that HOPS activity depends on a defined length of Vps41. If this were the case, the introduction of a linker of exactly the missing length should be sufficient to restore HOPS function. To restore the original linker length, we added a 10 amino acid (GGGGS)₂ linker to the Vps41^Δ527–640 mutant and tested whether this construct restores HOPS function (Fig. 6A,B). We compared the restored and non-restored mutants of Vps41 in three assays. First, we analyzed their function in endolysosomal transport by tracing the uptake of the lipophilic dye FM4-64 from the cell surface to the vacuolar surface over 60 min (Fig. 6A,D). In wild-type cells, the dye accumulated in dots and arrived at the vacuole membrane after 10 min, whereas it remained in dots throughout the measurement in vps41Δ cells as expected. In cells expressing the two Vps41 truncation mutants (Vps41^Δ527–640 and Vps41^Δ527–710), and less so in cells expressing Vps41^GS-linker, delivery of FM4-64 was delayed compared to that in wild-type cells (Fig. 6A,D), in agreement with the partial morphology defect observed before (Fig. 2B,C). However, cells expressing the Vps41^Δ527–640 mutant with the restoring (GGGGS)₂ linker showed strongly improved FM4-64 trafficking and restored vacuole morphology (Fig. 6A,D; Fig. S2A,B). This suggests that the linker recovered most of Vps41 functionality in the endolysosomal pathway.

As a second assay for HOPS functionality, we analyzed the transport of mCherry-tagged Atg8, an autophagy-specific marker, to the vacuole lumen (Guimaraes et al., 2015). Autophagy is induced by starvation of amino acids (‘N-starved’). Under these conditions, Atg8 is conjugated on the phagophore and ends up in the lumen of the autophagosome after closure. Upon fusion of the autophagosome with the vacuole, Atg8 is degraded in the vacuole lumen, as indicated by the luminal mCherry signal, which is clearly visible in wild-type cells (Fig. 6C). In contrast, vps41Δ as well as Vps41^Δ527–640 mutant cells accumulated multiple Atg8-positive dots and no luminal signal in the vacuole, indicative of an autophagy defect (Fig. 6C,E,F). The other two mutants (Vps41^Δ527–710 and Vps41^GS-linker) showed luminal Atg8 signals (Fig. 6C,E) as well as some non-vacuolar Atg8-positive dots (Fig. 6C,F), indicating that autophagosome–vacuole fusion is still partially functional. Strikingly, cells expressing the rescued Vps41^Δ527–640 mutant with the (GGGGS)₂ linker clearly showed Atg8 in the lumen, similarly to wild-type cells, indicating functional complementation by the extension of the length of Vps41, the former α-solenoid region.

As a third assay, we analyzed growth on plates as a measure of physiology, which was also corrected by the (GGGGS)₂ linker added to Vps41^Δ527–640 (Fig. 6G). This indicates that the Vps41 α-solenoid region needs to have a defined length, but not necessarily an α-solenoid structure, to support HOPS function.

We previously hypothesized that HOPS functions as a fusion catalyst by not only facilitating SNARE assembly via Vps33, but also by providing a stable backbone and thus an additional force towards the membrane when SNARE assembly occurs (Shvarev et al., 2022). The data presented here suggest that shortening of the α-solenoid domain of Vps41 prevents this function of HOPS as a fusion catalyst. HOPS binds to Ypt7-positive membranes via its Rab-interacting subunits Vps41 and Vps39. The Vps41 subunit especially reaches out prominently from the core structure, which is distinct from the structure of the related CORVET complex (Shvarev et al., 2022, 2024). Here, we show that, despite being flexible, this structural arrangement functions as a molecular ruler, which is essential for vacuole fusion. In our assay, we compared different truncation and extension mutants of Vps41, all of which had impaired functionality. However, the insertion of a flexible GS linker to match the extension rescued the most severely impaired Vps41^Δ527–640 mutant. This shows that the Vps41 α-solenoid domain functions as a molecular ruler to bridge the distance between the Ypt7-binding site in the β-propeller and the core of HOPS.

Why is it important that HOPS has such a molecular ruler? The Vps41 β-propeller, which binds Ypt7, is localized in our structural model far away from the core of HOPS (Figs 1A and 6H, panel 1) (Shvarev et al., 2022). As the structure was obtained from the soluble HOPS complex, binding to Ypt7 and the membrane likely causes a reorientation of Vps41 relative to the HOPS core (Fig. 6H, panel 2). This is likely as the membrane-interacting ALPS motif of the Vps41 β-propeller (Cabrera et al., 2010) pointed in our model away from the membrane (Shvarev et al., 2022). We therefore hypothesize that the exact distance of the α-solenoid ruler promotes the correct orientation of the Vps41 β-propeller towards the membrane. This then allows for coincidental binding of Vps41 to both Ypt7 and membranes via the ALPS motif (Fig. 6H, panel 3). The β-propeller of Vps41 could then be oriented next to the β-propeller of Vps11 with Ypt7 in the middle, which might strengthen the overall stability of HOPS and thus cause a tighter interaction between opposing membranes. Consequently, a more compact HOPS might support SNARE assembly and thus fusion more efficiently. We proposed a similar orientation of Ypt7 at the opposite end between the Vps18 β-propeller and the N-terminal part of Vps39 (Shvarev et al., 2022). Future structural analyses will be needed to determine the exact positioning of Ypt7 within the HOPS complex during tethering.

The interpretation that distance matters is confirmed by our mutant analysis. Whereas HOPS carrying the Vps41^Δ527–640, Vps41^Δ527–710 and Vps41^GS-linker mutant proteins can tether membranes but not promote fusion in vitro, the corresponding Vps41 mutants behave differently in cells. The most severe mutant, Vps41^Δ527–640, shows the most striking defects in growth, FM4-64 trafficking and autophagy, followed by the even more truncated Vps41^Δ527–710 and longer Vps41^GS-linker mutants. A likely explanation for the discrepancy between the in vitro and in vivo observations is the lack of the physiological context of native membranes. Apparently, a partially defective HOPS (with Vps41^Δ527–710 or Vps41^GS-linker) can still support membrane fusion with lower efficiency, which can be sufficient for endolysosomal transport and autophagy. We do not yet understand why the even shorter α-solenoid domain of the Vps41^Δ527–710 mutant causes a less severe phenotype. One possibility is that the orientation of the truncated α-solenoid of Vps41^Δ527–710 in relation to the remaining HOPS complex results in a more advantageous positioning of the Vps41 β-propeller at the membrane than that in cells expressing Vps41^Δ527–640 to still support fusion. Importantly, restoring the length of the linker rescues all defects of the most severe Vps41^Δ527–640 mutant, showing that distance of the Vps41 α-solenoid region relative to the β-propeller, rather than the α-solenoid structure, matters for HOPS function (Fig. 6).

Previous work on Vps33 showed that HOPS combines tethering with efficient SNARE chaperoning (Baker et al., 2015). If Vps33 is mutated, the corresponding HOPS complex fails to support proteoliposome fusion at physiological SNARE concentrations, but works like wild-type HOPS if the SNARE concentration on membranes is raised fourfold (Baker et al., 2015), suggesting that the specific activity of HOPS is impaired. We show here that a specific length of HOPS is required for fusion of the HOPS core with Ypt7 and the membrane. A HOPS complex with an impaired Vps41 ruler can still tether membranes and has a functional SNARE-binding site, but a strongly reduced specific activity. This strengthens our argument that HOPS is not just a tether with an SM protein but a membrane fusion catalyst with a dedicated architecture (Shvarev et al., 2022).

HOPS is predominantly upright on Ypt7-coated membranes (Füllbrunn et al., 2021), similar to CORVET when bound to Vps21-decorated liposomes (Shvarev et al., 2024). In our model, HOPS binds Ypt7 and brings membranes together and thus determines the position where fusion occurs. We hypothesize that in cells expressing wild-type Vps41, Ypt7 binding is combined with a molecular ruler function to promote the most favorable orientation of HOPS at membranes, which then favors catalysis (Fig. 6D). Intriguingly, Vps8 as the equivalent subunit to Vps41 in CORVET is tightly bound to Vps11 and has a longer α-solenoid domain (Shvarev et al., 2024). CORVET might thus not depend on a mechanism similar to that of Vps41 in HOPS, but is already in a more compact conformation to start with. Further comparative structural and functional analysis of HOPS and CORVET in Rab-dependent tethering and fusion assays will be required to test this model.

S. cerevisiae strains are listed in Table S1. For purification, HOPS subunits were expressed under the control of the GAL1 promotor. The 3× FLAG tag was attached to Vps41, except for the Vps41^Δ527–710 mutant, in which Vps18 carried the tag. Proteins for localization studies were tagged according to Janke et al. (2004). For generation of vps41 mutants, pRS404 plasmids (Sikorski and Hieter, 1989) were generated coding for the VPS41 promoter and the Vps41-coding region using Phusion mutagenesis PCR (Thermo Fisher Scientific). Plasmids were linearized using Bsu36I (New England Biolabs) and inserted into the TRP locus in respective vps41Δ strains.

Cells were grown overnight in yeast peptone (YP) medium containing 2% glucose (v/v) (YPD), diluted to an optical density (OD) at 600 nm (OD₆₀₀) of 0.2 the next day and grown to logarithmic phase. A total of 0.25 OD units was taken and serial dilutions (1:10, 1:100 and 1:1000) were spotted onto YPD plates containing the indicated amounts of ZnCl₂. Plates were incubated for up to 4 days, and images were acquired every 24 h with a ChemiDoc XRS+ system (Bio-Rad Laboratories). Growth tests were performed in triplicates.

Cells were grown in 10 ml YPD medium until the logarithmic growth phase. A total of 1 OD₆₀₀ unit was harvested and resuspended in 50 µl medium complemented with 30 µM FM4-64 (Molecular Probes). Samples were incubated for 30 min at 30°C in the dark, washed once with 1 ml YPD, and then incubated for another 30 min at 30°C. Before imaging, cells were centrifuged at room temperature (2 min, 500 g) and resuspended in synthetic complete medium supplemented with all amino acids and glucose (SDC+all) or SD-N medium for nitrogen starvation. For salt shock experiments, NaCl was added to a final concentration of 0.9 M, followed by a 5 min incubation before imaging. For hypotonic shock, cells were centrifuged at room temperature (2 min, 500 g), resuspended in water, and incubated for 5 min while shaking in a thermoshaker at 30°C. Samples were analyzed by fluorescence microscopy using a DeltaVision Elite microscope. The imaging system is based on an Olympus IX-71 inverted microscope equipped with a 100× NA 1.49 objective, a sCMOS camera (PCO), an InsightSSI illumination system 4ʹ,6-diamidino-2-phenylindole, GFP, mCherry and Cy5 filters, and SOftWoRx software (Applied Precision). Deconvolution was performed with the SOftWoRx software. Images were analyzed and processed with ImageJ (Schneider et al., 2012).

Purification of HOPS wild-type and Vps41 mutant complexes was performed as described (Shvarev et al., 2022) with minor changes. For this, 2 l of yeast peptone (YP) medium containing 2% galactose (v/v) was inoculated with 6 ml of an overnight preculture, followed by 24 h incubation at 30°C. Cultures were harvested by centrifugation at 4800 g and 10 min at 4°C, and washed once with ice-cold HOPS purification buffer [HPB; 300 mM NaCl, 20 mM HEPES/NaOH, pH 7.4, 1.5 mM MgCl₂ and 10% (v/v) glycerol]. The pellet was resuspended in a 1:1 ratio (w/v) in HPB supplemented with 1 mM phenylmethylsulfonylfluoride (PMSF), 1× FY protease inhibitor mix (Serva) and 1 mM dithiothreitol (DTT; AppliChem). Resuspended cells were dropwise frozen in liquid nitrogen before cell lysis using a freezer mill cooled with liquid nitrogen (SPEX SamplePrep). For purification, the powder was thawed on ice and resuspended in HPB supplemented with 1 mM PMSF, 1×FY and 1 mM DTT using a glass pipette. Resuspended cells were centrifuged two times at 5000 and 15,000 g at 4°C for 10 and 20 min, respectively. The supernatant was added to 1.5 ml of prewashed anti-FLAG M2 affinity gel (Sigma-Aldrich) and incubated for 45 min at 4°C on a nutator. Beads were centrifuged for 1 min at 500 g at 4°C to separate the supernatant and transferred to a 2.5 ml MoBiCol column (MoBiTec). The sample was washed with 25 ml HPB before the addition of FLAG peptide and incubated for 40 min at 4°C. The eluate was collected by centrifugation (500 g, 30 s, 4°C). For concentrating, a Vivaspin 100 kDa MWCO concentrator (Sartorius), which was previously incubated for 45 min with HPB containing 1% Triton X-100, was used. The final eluate was applied to a Superose 6 Increase 15/150 column (Cytiva) for size-exclusion chromatography (SEC) on an ÄKTA Go purification system. The peak fraction was used for further analysis.

Mass photometry analysis was performed using a TwoMP mass photometer (Refeyn). Data were obtained using the AcquireMP software (Refeyn) and analyzed with DiscoverMP (Refeyn). For samples, glass coverslips were used on which perforated silicone gaskets were placed to form wells for every sample to be measured. Samples were analyzed at a final concentration of 10 nM in a total volume of 20 μl. Here, SEC buffer (300 mM NaCl, 20 mM HEPES/NaOH, pH 7.4, 1.5 mM MgCl₂) was used for dilutions (see above).

3 µl of the purified protein sample was applied onto a glow-discharged carbon-coated copper grid with plastic support. The sample was blotted and stained with 2% (w/v) uranyl formate solution (Januliene and Moeller, 2021). Micrographs were collected manually on a JEM-2100Plus transmission electron microscope (JEOL) operated at 200 kV and equipped with a XAROSA CMOS 20-megapixel camera (EMSIS) at a nominal magnification of 30,000 (3.12 Å per pixel). Data analysis was done with ImageJ (Schneider et al., 2012) and cryoSPARC (Punjani et al., 2017).

HOPS-mediated tethering of liposomes was analyzed as described previously (Füllbrunn et al., 2021). 42.6% (mol/mol) 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC; Avanti Polar Lipids), 18% (mol/mol) 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE; Avanti Polar Lipids), 1% (mol/mol) 1,2-dipalmitoyl-sn-glycero-3-(cytidine diphosphate) (DAG; Avanti Polar Lipids), 8% (mol/mol) ergosterol (Sigma-Aldrich) and 1% (mol/mol) ATTO488-1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (ATTO488, ATTO-TEC) was used as the lipid composition. Liposomes labeled with fluorescent dye were generated and preloaded with prenylated Ypt7 (pYpt7; Langemeyer et al., 2018). For this, 50 nmol liposomes were added to 50 pmol pYpt7:GDI and GTP and incubated for 30 min at 27°C. 0.17 mM of the preloaded liposomes was added to 250 nM wild-type or mutant HOPS complex in a final volume of 15 μl before incubation for 10 min at 27°C. Tethered liposomes were sedimented by centrifugation for 5 min at 1000 g at 4°C. Tethering efficiency was analyzed in a SpectraMaxM3 fluorescence plate reader (Molecular Devices) by comparing the ATTO488 fluorescence signal of the supernatant before and after sedimentation.

Yeast cultures from wild-type or vps11-1 strains with either pep4Δ or pho8Δ were grown in 1 l YP medium containing 2% glucose (v/v) to an OD₆₀₀ of 1 at 30°C (wild-type) or 23°C (vps11-1). Cultures were harvested (4800 g, 10 min, 4°C) and resuspended in 50 ml buffer (0.1 M Tris/HCl, pH 9.4, and 10 mM DTT) followed by a 10 min incubation at 30°C. Probes were centrifuged as before and cells were resuspended in 25 ml spheroplasting buffer [0.2× YPD, 0.6 M sorbitol and 50 mM KPi, pH 7.4; KPi is from 1 M KPi buffer, pH 7.4 (generated from 2.64 g potassium dihydrogen phosphate + 11.02 g dipotassium hydrogen phosphate per 100 ml)]. Then, 0.3 mg/ml lyticase (made in house) was added and cells were incubated for another 30 min at 30°C. Spheroplasts were collected after centrifugation at 1000 g for 5 min at 4°C before resuspension in 2.5 ml of 15% (w/v) Ficoll (Cytiva; made in 0.2 M sorbitol and 10 mM PIPES/KOH, pH 6.8) and 0.02 mg/ml DEAE Dextran (Sigma-Aldrich). Samples were incubated on ice for 5 min, then heat shocked for 2 min at 30°C. Probes were transferred into a SW40 tube (Seton) and sequentially layered with 3 ml 8% (w/v) Ficoll (in 0.2 M sorbitol and 10 mM PIPES/KOH, pH 6.8) and 3 ml 4% (w/v) Ficoll (in 0.2 M sorbitol and 10 mM PIPES/KOH, pH 6.8), and then filled up with 0% Ficoll (0.2 M sorbitol and 10 mM PIPES/KOH, pH 6.8) and centrifuged at 100,000 g for 90 min at 4°C in a SW40 rotor (Beckman Coulter). Vacuoles at the 0–4% interface were harvested by pipetting, followed by protein concentration determination using Bradford analysis (ROTIQuant solution, Roth). Vacuoles were diluted to a concentration of 0.3 mg/ml in 0% Ficoll before the fusion reaction. For the fusion reaction, 3 µg vacuoles of each strain were incubated with 3 µl 10× fusion buffer (125 mM KCl, 5 mM MgCl₂, 20 mM sorbitol, 1 mM PIPES/KOH, pH 6.8), 3 µl 10× ATP buffer (5 mM ATP, 1 mg/ml creatine kinase, 400 mM creatine phosphatase, 10 mM PIPES/KOH, pH 6.8, 0.2 M sorbitol), 10 µM coenzyme A (Sigma-Aldrich, dissolved in 0 % Ficoll), 0.01 µM Sec18 and 2 µM Vam7 (90 min, 26°C). Purification of proteins (Sec18, Vam7) was performed as described previously (Langemeyer et al., 2018). Wild-type and mutant HOPS complexes were titrated into the assay. The total reaction volume was 35 µl. p-nitrophenolphosphate (PNPP; Sigma-Aldrich, dissolved in 0% Ficoll) as the substrate of the phosphatase Pho8 was added (10 mM MgCl₂, 0.25 M Tris, pH 8.5, 0.4% Triton X-100, 1.5 mM PNPP) and quenched after 5 min by adding 500 µl stop solution (1 M glycine, pH 11.5). Vacuole fusion activity was determined by measuring the absorbance of the generated nitrophenol at 400 nm (Haas, 1995).

The statistical analyses of quantifications were performed as indicated. For this, two tailed unpaired t-tests were used. Non-significant (n.s.) differences (P>0.05) are not indicated in the plots. *P<0.05; **P<0.01; ***P<0.001.

We wish to acknowledge the access to microscopy platforms at the CellNanOs center at Osnabrück University.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.263788.reviewer-comments.pdf