Structure of the HOPS tethering complex, a lysosomal membrane fusion machinery

  1. Dmitry Shvarev
  2. Jannis Schoppe
  3. Caroline König
  4. Angela Perz
  5. Nadia Füllbrunn
  6. Stephan Kiontke
  7. Lars Langemeyer
  8. Dovile Januliene
  9. Kilian Schnelle
  10. Daniel Kümmel
  11. Florian Fröhlich
  12. Arne Moeller  Is a corresponding author
  13. Christian Ungermann  Is a corresponding author
  1. Department of Biology/Chemistry, Structural Biology section, Osnabrück University, Germany
  2. Department of Biology/Chemistry, Biochemistry section, Osnabrück University, Germany
  3. Department of Plant Physiology and Photo Biology, University of Marburg, Germany
  4. Department of Chemistry and Pharmacy, Institute of Biochemistry, University of Münster, Germany
  5. Department of Biology/Chemistry, Molecular Membrane Biology group, Osnabrück University, Germany
  6. Center of Cellular Nanoanalytic Osnabrück (CellNanOs), Osnabrück University, Germany

Abstract

Lysosomes are essential for cellular recycling, nutrient signaling, autophagy, and pathogenic bacteria and viruses invasion. Lysosomal fusion is fundamental to cell survival and requires HOPS, a conserved heterohexameric tethering complex. On the membranes to be fused, HOPS binds small membrane-associated GTPases and assembles SNAREs for fusion, but how the complex fulfills its function remained speculative. Here, we used cryo-electron microscopy to reveal the structure of HOPS. Unlike previously reported, significant flexibility of HOPS is confined to its extremities, where GTPase binding occurs. The SNARE-binding module is firmly attached to the core, therefore, ideally positioned between the membranes to catalyze fusion. Our data suggest a model for how HOPS fulfills its dual functionality of tethering and fusion and indicate why it is an essential part of the membrane fusion machinery.

Editor's evaluation

This landmark study reports the cryo-EM structure of HOPS, a heterohexameric tether that participates in the fusion of late endosomes, autophagosomes, and AP-3 vesicles with lysosomes. The structure provides a convincing update of earlier, lower-resolution models. Interestingly, the SNARE-binding module is attached to the core of the complex. These results suggest possible mechanisms by which HOPS could catalyze SNARE-dependent fusion.

https://doi.org/10.7554/eLife.80901.sa0

eLife digest

Our cells break down the nutrients that they receive from the body to create the building blocks needed to keep us alive. This is done by compartments called lysosomes that are filled with a cocktail of proteins called enzymes, which speed up the breakdown process. Lysosomes are surrounded by a membrane, a barrier of fatty molecules that protects the rest of the cell from being digested. When new nutrients reach the cell, they travel to the lysosome packaged in vesicles, which have their own fatty membrane. To allow the nutrients to enter the lysosome without creating a leak, the membranes of the vesicles and the lysosome must fuse.

The mechanism through which these membranes fuse is not fully clear. It is known that both fusing membranes must contain proteins called SNAREs, which wind around each other when they interact. However, this alone is not enough. Other proteins are also required to tether the membranes together before they fuse. To understand how these tethers play a role, Shvarev, Schoppe, König et al. studied the structure of the HOPS complex from yeast. This assembly of six proteins is vital for lysosomal fusion and, has a composition similar to the equivalent complex in humans.

Using cryo-electron microscopy, a technique that relies on freezing purified proteins to image them with an electron microscope and reveal their structure, allowed Shvarev, Schoppe, König et al. to provide a model for how HOPS interacts with SNAREs and membranes. In addition to HOPS acting as a tether to bring the membranes together, it can also bind directly to SNAREs. This creates a bridge that allows the proteins to wrap around each other, driving the membranes to fuse.

HOPS is a crucial component in the cellular machinery, and mutations in the complex can cause devastating neurological defects. The complex is also targeted by viruses – such as SARS-CoV-2 – that manipulate HOPS to reduce its activity. Shvarev, Schoppe, König et al.’s findings could help researchers to develop drugs to maintain or recover the activity of HOPS. However, this will require additional information about its structure and how the complex acts in the biological environment of the cell.

Introduction

Lysosomal fusion underlies a plethora of cellular processes. It is essential in the maintenance and upkeep of eukaryotic membranes and fundamental to secretion, endocytosis, and autophagy (Saftig and Puertollano, 2021). Macromolecules from different trafficking pathways end up in lysosomes where they are degraded (Klionsky et al., 2021; Saftig and Puertollano, 2021). This process relies on multiple fusion events within the endomembrane system. In general, fusion depends on SNAREs, which are present on opposite membranes and zipper into four-helix bundles with the help of Sec1/Munc1 (SM) proteins (Jahn and Fasshauer, 2012; Südhof and Rothman, 2009; Wickner and Rizo, 2017). Each assembled SNARE complex contains three Q-SNAREs (Qa, Qb, and Qc) and one R-SNARE, which are categorized according to the interaction of the glutamine and arginine residues in the central hydrophilic layer of the otherwise hydrophobic interfaces within the SNARE complex (Jahn and Fasshauer, 2012; Südhof and Rothman, 2009; Wickner and Rizo, 2017). Prior to fusion, specialized tethering complexes establish tight links between organelles and interact with SM proteins to promote fusion (Baker and Hughson, 2016; Kuhlee et al., 2015; Ungermann and Kümmel, 2019). Despite their central position in trafficking, the underlying mechanics of tethering complexes and how they catalyze membrane fusion remain unresolved.

The heterohexameric HOPS complex mediates fusion of late endosomes, autophagosomes, and AP-3 vesicles with mammalian lysosomes or vacuoles in yeast (van der Beek et al., 2019; Spang, 2016; Wickner and Rizo, 2017), and is probably the best-studied tethering complex. Fusion assays using yeast vacuoles or reconstituted SNARE-bearing proteoliposomes showed that HOPS is essential for membrane fusion at physiological SNARE concentrations (D’Agostino et al., 2017; Mima et al., 2008; Zick and Wickner, 2016). HOPS is the target of viruses such as SARS-CoV2 (Miao et al., 2021), and its inactivation blocks Ebola infections (Carette et al., 2011). Furthermore, multiple HOPS mutations can cause severe diseases ranging from Parkinson’s to lysosomal disorders (van der Beek et al., 2019; Sanderson et al., 2021; van der Welle et al., 2021).

Five out of six HOPS subunits (Vps11, Vps16, Vps18, Vps39, and Vps41) are predicted to share a similar architecture, comprising an N-terminal β-propeller and a C-terminal α-solenoid domain (Figure 1A). Vps11 and Vps18 form the core and carry conserved C-terminal RING finger domains (Rieder and Emr, 1997), which are essential for HOPS formation (Hunter et al., 2017), but also show E3 ligase activity on their own (Segala et al., 2019). At the opposite sites, Vps41 and Vps39 bind to membrane-anchored small GTPases (the Rab7-like Ypt7 in yeast) (Bröcker et al., 2012; Lürick et al., 2017; Ostrowicz et al., 2010), while Vps16 and the SM protein Vps33 establish the SNARE-binding module (Baker et al., 2015; Graham et al., 2013). Low-resolution negative-stain electron microscopy (EM) analyses revealed the overall arrangement of HOPS (Bröcker et al., 2012; Chou et al., 2016), yet were insufficient to localize the exact position of its subunits and suggested significant flexibility within the particle.

Figure 1 with 4 supplements see all
Composition and architecture of the yeast HOPS complex.

(A) Domain architecture and size of HOPS subunits. Predicted domains and structural features are indicated. (B) Size exclusion chromatography (SEC) of the affinity-purified HOPS. Purification was done as described in Materials and methods. (C) Mass photometry analysis. Peak fractions from SEC were analyzed for size. (D) Purified HOPS. Proteins from affinity purification (eluate) and SEC (red dashes in (B)) were analyzed by SDS-PAGE. (E) Overall architecture of the HOPS complex. Composite map from local refinement maps (Figure S1-3) was colored by assigned subunits. One of the consensus maps used for local refinements was low-pass-filtered and is shown as a transparent envelope. (F) Atomic model of the HOPS complex. For the N-terminal fragments of Vps41 and Vps39, which were not resolved to high resolution by local refinements, AlphaFold models are used and manually fitted into the densities of consensus maps (Figure 1—figure supplements 1 and 2). The triangular shape of the complex is highlighted with the approximate distance between the β-propellers of Vps41 and Vps39. (G) Schematic representation of the HOPS complex indicates central features.

The way HOPS fulfills its function remains speculative, and multiple mechanisms have been proposed, including a role as a bulky membrane stressor (D’Agostino et al., 2017) or, conversely, as a highly flexible membrane tether (Bröcker et al., 2012; Chou et al., 2016). In the absence of detailed structural data, it remains obscure how HOPS facilitates lysosomal fusion.

Results

Structure of the HOPS complex

Previously, structural studies were hampered by the low stability and flexibility of the complex, which required fixation through mild crosslinking for sample preparation and confined structural studies to negative stain analyses (Bröcker et al., 2012; Chou et al., 2016). To enable high-resolution cryo-EM of non-crosslinked HOPS, we vastly improved and accelerated our purification protocols and removed any delays during the sample preparation procedure (Figure 1B–D). Single-particle analysis including extensive classifications followed by local refinements led to a composite structure with resolutions between 3.6 and 5 Å (Figure 1E and Figure 1—figure supplements 13).

HOPS forms a largely extended, slender structure extending approximately 430 Å in height and 130 Å in width, resembling a triangular shape (Figure 1E–G, Video 1). In the center of the modular complex, Vps11 and Vps18 align antiparallel through their elongated α-solenoids, establishing a large interface area of 1972 Å2 (Figure 1E-G, Figure 2, Figure 2—figure supplement 4), which resulted in the highest resolution obtained within HOPS (Figure 1—figure supplement 3) and is comparable with protein interfaces in other complexes with similar structural elements as in HOPS (e.g. Kschonsak et al., 2022). Interestingly, AlphaFold predicts a long unstructured region within Vps11 (Q760 to D784), resulting in an upper and lower part of the subunit. However, this region is clearly resolved and organized in our density. In our model, the two core subunits create a central assembly hub for the four other subunits (Vps39, Vps41, Vps16, and Vps33) that fulfill specific functions and localize to the periphery of the complex. The N-termini of Vps11 and Vps18 are located distally from the core and each carries a β-propeller, which can be deleted without affecting the complex formation (Figure 3A and D, Figure 1—figure supplement 4). At the C-termini, both Vps11 and Vps18 have long α-helices which are followed by RING finger domains (Figures 1E–G2A and B, Figure 2—figure supplement 1). Both features are key elements for the stability of the modular architecture and serve as anchor points for the additional subunits. In agreement, HOPS, carrying mutations in the RING finger domains of Vps11 (vps11-1) (Peterson and Emr, 2001), selectively lost Vps39, whereas only Vps41 was obtained from a similar Vps18 mutant.

Video 1
Overall architecture of HOPS tethering complex: transition between ribbon and molecular surface representation.
Figure 2 with 4 supplements see all
Vps11 and Vps18 C-termini as central interaction hubs for all other subunits.

Atomic model of HOPS with highlighted interaction sites between subunits. (A, B) Coiled-coil motifs followed by the RING finger domains (violet) are the key structural features of HOPS. (A) The Vps18 C-terminal hub. Vps18 and Vps41 interact via the coiled coil and the Vps18 RING finger domain (displayed as non-transparent cartoons). (B) The Vps11 C-terminal hub. Vps11 interacts via its RING finger domain and the coiled coil with Vps39 (displayed as non-transparent cartoons). (C) Connection of the SNARE binding module (Vps33 and Vps16) to the backbone of HOPS via interactions with the structured loop at the RING finger domain and the C-terminus of Vps18 (displayed as non-transparent cartoons). (D) Vps39 connects by its C-terminal helix the β-propeller of Vps18, which provides additional stability in this part of HOPS.

Figure 3 with 1 supplement see all
HOPS couples tethering and fusion activities.

(A) Schematic representations of HOPS wild-type (as in Figure 1G) and mutants lacking N-terminal β-propeller domains (indicated by pink asterisks). (B) Tethering assay. Fluorescently labeled liposomes loaded with prenylated Ypt7-GTP or none were incubated with HOPS and mutant complexes. Tethering was determined as described in Materials and methods. Data shown from three biological replicates, bars indicate standard deviation. (C) Fusion assay. Fusion of proteoliposomes carrying vacuolar SNAREs were preincubated with Ypt7-GDI, GTP, and Mon1-Ccz1. For fusion, HOPS wild-type or mutant and the soluble Vam7 SNARE were added (Langemeyer et al., 2020; Langemeyer et al., 2018). Analysis was done as described (Zick and Wickner, 2016). See Materials and methods. Data shown from three biological replicates, bars indicate standard deviation. (D) Representative 2D class averages obtained from negative-stain analyses of wild-type HOPS and mutants. Pink asterisks indicate missing densities in the mutants.

Vps41 and Vps39 provide the Ypt7-interaction sites at their peripheral N-terminal regions. Their extended C-terminal helices, similar to those of Vps11 and Vps18, are tightly interlocked through coiled-coil motifs with the long C-terminal α-helices and RING finger domains from the respective core subunits (Vps41 with Vps18, and Vps39 with Vps11) (Figure 2A and B). Additional stability of Vps39 within the complex is provided by the interaction of the long α-helix at the C-terminus of Vps39 with the β-propeller of Vps18 (Figure 2D). In our density, peripheral portions of both Vps41 and Vps39 are least well resolved indicating their considerable flexibility. Multi-modular classification analyses revealed angular re-orientations of about 9° for Vps41 and 20° for Vps39 (Figure 2—figure supplement 3A-D) relative to the core, resulting in variable positions of the N-terminal β-propellers. At the top, Vps41 reaches out by approximately 100 Å in length and, similarly, Vps39 forms an elongated arch at the bottom (Figure 1E–G), positioning both Ypt7-interacting units at the farthest ends of the complex.

The SNARE-binding element, composed of Vps16 and the SM protein Vps33, branches out to the lateral side of the complex approximately at the center of the structure (Figures 1E–G2). Vps16 shares a large interface with the coiled-coil motif formed by Vps18 and Vps41 and the N-terminus of Vps18, which is stabilized through interactions between hydrophobic and charged residues (Figure 2C). Vps33 is in immediate contact with the structured loop of Vps18 (residues 824–831) that connects the elongated helix with the RING finger domain (Figure 2C). This, as well as the role of RING finger domains in the interlocking of other subunits, explains, why mutations at RING domains result in devastating human diseases and HOPS dysfunction (van der Beek et al., 2019; Edvardson et al., 2015; Robinson et al., 1991; van der Welle et al., 2021; Zhang et al., 2016) and cause failure of correct HOPS assembly (Figure 2—figure supplement 2). Overall, the SNARE binding module appears to be stably connected to the central core, while only the short C-terminal section of Vps16 α-solenoid (residues 739–798) displays high variability and is not resolved in our structure.

HOPS couples tethering to fusion activity

Tethering complexes bridge membranes by binding small GTPases, but also harbor or bind SM proteins (Ungermann and Kümmel, 2019). Reconstituted vacuole fusion is strictly HOPS and Ypt7-dependent at physiological SNARE concentrations (Langemeyer et al., 2018; Zick and Wickner, 2016), suggesting that HOPS is not just a tether, but part of the fusion machinery (Baker et al., 2015; Wickner and Rizo, 2017). However, so far, it was unknown how tethering and fusion activities of HOPS may be linked mechanistically. To address this, we first analyzed the N-terminal β-propellers of Vps41 and Vps39, the likely binding sites with Ypt7 (Lürick et al., 2017; Ostrowicz et al., 2010; Plemel et al., 2011). The intrinsic low affinity between HOPS and Ypt7 (Lürick et al., 2017) prevented reconstitution of the complex for structural studies, therefore, we instead relied on AlphaFold predictions. Additionally, we solved the structure of the β-propeller of Chaetomium thermophilum Vps39 by X-ray crystallography, which largely confirmed the predicted model (Figure 3—figure supplement 1A). Surprisingly, in the AlphaFold model, Ypt7 binding occurs at the α-solenoid of Vps39 where it does not directly interact with the β-propeller (Figure 3—figure supplement 1C), as originally expected. Furthermore, the binding site on Vps39 is placed approximately 5–6 nm above the membrane if Ypt7-anchored HOPS is in an upright position on supported lipid bilayers (Füllbrunn et al., 2021). Membrane-bound Ypt7 can still reach this site due to its 10 nm long hypervariable domain (not shown in the prediction).

In the predicted complex of Vps41 (residues 1–919) with Ypt7 (residues 1–185), the GTPase binds directly to the Vps41 β-propeller, as anticipated. However, it interacts on the opposite side from the membrane-interacting amphipathic lipid-packing sensor (ALPS) motif (Cabrera et al., 2010; Figure 3—figure supplement 1B), suggesting that the hypervariable region of Ypt7 is required for binding, in analogy to Vps39. Curiously, in the predicted model, the ALPS motif faces away from the membrane which would hamper membrane binding. We noted, however, that this distal region of Vps41 (Figure 3—figure supplement 1B) displays substantial flexibility. The β-propeller of Vps41 might, therefore, be oriented differently in the structure than predicted by AlphaFold, which may bring the ALPS motif in contact with the bilayer if Vps41 is bound to Ypt7. Nevertheless, future experimental data will need to confirm the predicted AlphaFold model of Vps41 and Vps39 interaction with Ypt7.

Vps39 binds Ypt7 far stronger than Vps41 (Auffarth et al., 2014; Lürick et al., 2017), and may be assisted by Vps18 to sandwich Ypt7, whereas Vps41 binds to Ypt7 apparently alone (Figure 3—figure supplement 1D). Such a dual interaction could explain both tighter binding and a preferred orientation of HOPS on membranes (Füllbrunn et al., 2021). To test the functional importance, we generated HOPS complexes lacking the β-propeller of Vps11, Vps18, or Vps41 (Figure 3A). All complexes were purified in equimolar stoichiometry (Figure 1—figure supplement 4A), and interacted with Ypt7, but not the Golgi Rab Ypt1 in GST pull-down assay, suggesting that at least one Rab-binding site is maintained in all truncated complexes (Bröcker et al., 2012; Lürick et al., 2017; Ostrowicz et al., 2010; Zick and Wickner, 2016; Zick and Wickner, 2012; Figure 1—figure supplement 4B).

To determine the activity of HOPS mutants, we compared tethering and fusion. For tethering, we incubated liposomes bearing Ypt7-GTP with each complex and quantified clustering after centrifugation (Füllbrunn et al., 2021; Lürick et al., 2017). HOPS lacking the Vps41 β-propeller was inactive as shown (Lürick et al., 2017), whereas HOPS with truncated Vps11 or Vps18 was fully functional and as efficient as wild-type HOPS (Figure 3B). In contrast, when added to SNARE and Ypt7-GTP bearing liposomes, only wild-type HOPS, but none of the mutant complexes promoted fusion (Figure 3C). This was particularly puzzling for HOPS lacking either the Vps18 or Vps11 β-propeller as they had full tethering activity (Figure 3B). Therefore, we compared the structural features of HOPS mutants with the wild type using negative stain EM. Deletions of the β-propellers of Vps41, Vps11, or Vps18 indeed resulted in a loss of protein density at the expected positions, while preserving the densities of all other subunits (Figure 3D). Interestingly, HOPS complexes lacking β-propellers in Vps11 or Vps18 showed an alteration in the relative orientation of Vps39 within the complex in some 2D class averages. This observed structural variation might be a result of the increased flexibility of mutant HOPS due to the lack of structural support by β-propellers of Vps18 and Vps11. We conclude that the β-propellers of Vps18 and Vps11 contribute to the overall structure of the HOPS complex or may play a stabilizing role during the fusion process, which would explain, why they are essential for the full activity of HOPS.

Discussion

Our data suggest a working model of how HOPS catalyzes fusion at lysosomes and vacuoles (Figure 4). According to our structure, the three major ligand binding sites of HOPS are arranged in a triangular fashion. While the two Ypt7-binding sites show significant conformational variability, the SNARE-interacting unit is firmly connected to the stable backbone formed by Vps11 and Vps18. For tethering, HOPS Vps39 and Vps41 bind Ypt7 on target membranes. During this process, HOPS remains upright on membranes (Füllbrunn et al., 2021). Then, the SM protein Vps33 and possibly other sites on HOPS (Krämer and Ungermann, 2011; Baker et al., 2015; Lürick et al., 2017; Lürick et al., 2015; Song et al., 2020) bind SNAREs from the opposing membranes, and zipper them up toward their membrane anchor (Figure 4). Note, that this process can be blocked by Orf3a in the COVID-19 SARS-CoV-2 virus (Miao et al., 2021). In our model, the backbone of HOPS dampens the movement of the vesicles and acts as a lever arm holding on to SNAREs during zippering (D’Agostino et al., 2017). This three-point arrangement would cause membrane stress and could explain how HOPS catalyzes membrane fusion (Figure 4). The physiological function of HOPS can be bypassed if large complexes are redirected to SNAREs at the fusion site (D’Agostino et al., 2017; Song et al., 2017), which can even promote fusion of deficient SNARE complexes (Orr et al., 2022; Song et al., 2021). Zippered SNAREs may then dissociate from HOPS and allow access for α-SNAP and NSF to recycle SNAREs (Zhang and Hughson, 2021).

Figure 4 with 1 supplement see all
Working model for HOPS-mediated membrane tethering and fusion.

The HOPS complex binds to Ypt7 on the vacuole and vesicles via Vps39 (dark green) and Vps41 (light green). SNARE proteins are recruited to HOPS by their N-terminal domains and the SNARE-binding module (dark and light brown). The stable central core of HOPS keeps the membranes in place, while Vps41 and Vps39 may function as dampers due to their limited flexibility. Consequentially, zippering of SNAREs, which is initiated by binding to Vps33 (dark brown), begins. As the N-terminal domains of SNAREs bind to HOPS, further SNARE zippering may occur with the HOPS backbone acting as a lever (not shown here). This may cause membrane stress and thus catalyzes fusion. HOPS may let go of Ypt7 and SNAREs thereafter. For details see text.

How tethering complexes contribute to fusion poses a long-standing question in the field. The process necessitates the binding of two opposing membranes and exact coordination of the zippering procedure of membrane-bound SNAREs. Previous analyses suggested strong flexibility along the entire HOPS structure, which was interpreted as a hallmark of tethering complexes and an essential prerequisite to their function (Bröcker et al., 2012; Chou et al., 2016; Füllbrunn et al., 2021; Ha et al., 2016). Structural models of a largely open HOPS, based on negative stain EM. Chou et al., 2016, supported the importance of HOPS structural flexibility. Flexibility was also observed in our previous negative stain EM structure (Bröcker et al., 2012). Instead, our cryo-EM data, collected on a highly pure and homogeneous sample that was not modified by any crosslinker or other fixative agents like negative stain, show that HOPS flexibility is limited. Taking the obtained resolution as a measure of sample flexibility (Rawson et al., 2016), we conclude that the backbone and the SNARE-binding module are the least flexible parts of HOPS and appear to be stably associated with each other. In contrast, the membrane interacting units of Vps41 and Vps39 show some flexibility with 10–20° movements between particles, which substantially reduced their resolution (Figure 2—figure supplement 3). We relate this flexibility to their function within HOPS, where it may dampen the motion of HOPS between membranes and stabilizes the SNARE interaction, which are necessary for fusion (Laage and Ungermann, 2001; Krämer and Ungermann, 2011; Lürick et al., 2015; Baker et al., 2015; Song et al., 2020). Mammalian HOPS contains the same six subunits as yeast HOPS, which are generally highly conserved and will therefore function similarly (van der Beek et al., 2019; van der Kant et al., 2015). Our structure can hence be used to map disease prone mutations (Figure 4—figure supplement 1) and may thus explain the consequences on HOPS function.

Due to low binding affinities, no structures of tethering complexes bound to Rabs have been solved. Here, we used AlphaFold to predict the interfaces of HOPS with its bound Rab7-like Ypt7 protein to understand the positioning of HOPS during tethering (Figure 3—figure supplement 1). We suspect that this low binding affinity helps tethering complexes to also let go of the Rab when fusion progresses (Figure 4). Even though long coiled-coil tethers such as EEA1 can promote SNARE-mediated membrane fusion (Murray et al., 2016), their cooperation with Rab GTPases in fusion is likely quite different from the mechanism proposed here (Ungermann and Kümmel, 2019).

Some tethering complexes such as CORVET, CHEVI, or FERARI have attached SM proteins (van der Beek et al., 2019; Solinger et al., 2020), others such as COG, GARP, Exocyst, and Dsl cooperate with SM proteins (Ungermann and Kümmel, 2019), which may catalyze fusion similar to HOPS. For each of these complexes, in vivo models for their function exist, yet proteoliposome fusion assays in the presence of the required small GTPases are either not available or not yet completely developed (Balderhaar et al., 2013; Ha et al., 2016; Lee et al., 2022; Maib and Murray, 2022; Ren et al., 2009; Rossi et al., 2020; Solinger et al., 2020). We expect that similar approaches as established for HOPS will further support the key role of tethering complexes and reveal their cooperation with SM proteins in SNARE-mediated membrane fusion.

The overarching principle suggested here for HOPS is not limited to lysosomal fusion but may extend to synaptic transmission, where the tether Munc13 and the SM protein Munc18 cooperatively catalyze the N- to C-terminal zippering of SNAREs during fusion (Lai et al., 2017; Rizo, 2022; Stepien et al., 2022; Stepien and Rizo, 2021). Both Munc18 and Vps33 interact similarly with the N-terminal part of the SNARE domains of the R- and Qa-SNARE and may promote assembly until the central zero-layer of the SNARE domain (Baker et al., 2015; Stepien et al., 2022).

During HOPS-mediated fusion, SNARE zippering beyond the zero-layer could then proceed, while the Vps33 lets go of the forming four-helix bundles (Stepien et al., 2022). However, HOPS binds both the N-terminal extensions of the Qa-SNARE Vam3 and other SNAREs, possibly via different binding sites along the HOPS complex (Laage and Ungermann, 2001; Lürick et al., 2015; Song et al., 2020). This association may thus maintain the force on membranes to catalyze full fusion, even if the SM protein lets go of the assembling SNARE complex (Figure 4). In agreement, HOPS complexes with deficient Vps33 can catalyze fusion of proteoliposomes only if the SNARE density is high (Baker et al., 2015). In turn, vacuoles expressing a Qa-SNARE Vam3 variant lacking the N-terminal extension, which is needed to bind HOPS, show diminished fusion (Laage and Ungermann, 2001). This suggests that HOPS supports SNAREs by templating the association of R- and Qa-SNAREs and by binding the N-terminal regions of SNAREs.

We believe that the deletion of the β-propeller in Vps11 or Vps18 result in a similarly deficient HOPS due to a less stable backbone (Figure 3). In either case, coupling between a stable backbone of HOPS and SNARE binding may be impaired and could result in less specific activity as fusion catalysts. Future experiments are required to determine the precise reason for their fusion deficiency.

Overall, our insights provide a novel blueprint to understand HOPS function, dynamics, and regulation both in fusion and in other functions of its subunits (van der Beek et al., 2019; Elbaz-Alon et al., 2014; Hönscher et al., 2014; González Montoro et al., 2021; González Montoro et al., 2018; Wong et al., 2020), and imply a general role of tethering complexes as a catalytic part of the fusion machinery.

Materials and methods

Yeast strains

Request a detailed protocol

Yeast strains used in this study are listed in Supplementary file 1. In general, HOPS subunits were expressed under the control of the GAL1 promoter according to the standard protocol (Janke et al., 2004). For HOPS subunit truncation (Vps41, Vps11, and Vps18) of the N-terminal part, the GAL1 promotor was inserted into the genome at the respective position. The 3x-FLAG Tag was attached to the HOPS subunit Vps41, except for the 18 ΔN mutant.

Protein expression and purification from Escherichia coli

Request a detailed protocol

Rab GTPases for pulldown or tethering and fusion assays were expressed in Escherichia coli BL21 (DE3) Rosetta cells. Cells were grown in Luria broth (LB) medium complemented with 35 µg/ml kanamycin and 30 µg/ml chloramphenicol. Cultures were induced at OD600=0.6 with 0.5 mM isopropyl-β-d-thiogalactoside (IPTG) overnight at 16°C before harvesting by centrifugation (4800×g, 10 min, 4°C). Cells resuspended in buffer (150 mM NaCl, 50 mM HEPES/NaOH, pH 7.4, 10 % glycerol, 1 mM PMSF, and 0.5-fold protease inhibitor mixture [PIC]) were lysed by a Microfluidizer (Microfluidics Inc) and centrifugated at 25,000×g, 30 min, 4°C. Supernatants were incubated with glutathione Sepharose (GSH) fast flow beads (GE Healthcare) for GST-tagged proteins or nickel–nitriloacetic acid (Ni-NTA) agarose (Qiagen) for His-tagged proteins for 2 hr at 4°C. The proteins were eluted with buffer (150 mM NaCl, 50 mM HEPES/NaOH, pH 7.4, 10 % glycerol) containing 25 mM glutathione or 300 mM imidazole. Buffer was exchanged via a PD10 column (GE Healthcare). For tag cleavage, TEV protease was added after washing and incubated overnight. All proteins were stored at −80°C.

Purification of the 3xFLAG-tagged HOPS complex variants from yeast

Request a detailed protocol

Two liters of yeast peptone (YP) medium containing 2 % galactose (v/v) were inoculated with 6 ml of an overnight culture. Cells were grown for 24 hr and harvested by centrifugation (4800×g, 10 min, 4°C). Pellets were washed with cold HOPS purification buffer (HPB, 1 M NaCl, 20 mM HEPES/NaOH, pH 7.4, 1.5 mM MgCl2, and 5% (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 (AppliChem GmbH) and afterward dropwise frozen in liquid nitrogen before being lysed in a freezer mill cooled with liquid nitrogen (SPEX SamplePrep LLC). 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, followed by two centrifugation steps at 5000 and 15,000×g at 4°C for 10 and 20 min, respectively. After centrifugation, the supernatant was added to 2 ml of anti-FLAG M2 affinity gel (Sigma-Aldrich) and gently agitated for 45 min at 4°C on a nutator. Beads were briefly centrifuged (500×g, 1 min, 4°C) and the supernatant was removed. Beads were transferred to a 2.5 ml MoBiCol column (MoBiTec) and washed with 25 ml of HOPS washing buffer (HWB, 1 M NaCl, 20 mM HEPES/NaOH, pH 7.4, 1.5 mM MgCl2, and 20% (v/v) glycerol). FLAG-peptide was added and incubated on a turning wheel for 40 min at 4°C. The eluate was collected by centrifugation (150×g, 30 s, 4°C) and concentrated in a Vivaspin 100 kDa MWCO concentrator (Sartorius), which was previously incubated for 45 min with HWB containing 1% TX-100. The concentrated eluate was applied to a Superose 6 Increase 15/150 column (Cytiva) for size exclusion chromatography (SEC) and eluted in 50 μl fractions using ÄKTA go purification system (Cytiva). Peak fractions were used for further analysis.

Mass photometry

Request a detailed protocol

Mass photometry experiments were done using a Refeyn TwoMP (Refeyn Ltd). Data were obtained using AcquireMP software and analyzed using DiscoverMP (both Refeyn Ltd). Glass coverslips were used for sample analysis. Perforated silicone gaskets were placed on the coverslips to form wells for every sample to be measured. Samples were evaluated at a final concentration of 10 nM in a total volume of 20 μl in the buffer used for SEC.

Cryo-EM sample preparation and data acquisition

Request a detailed protocol

Prior to cryo-EM analysis, HOPS samples were tested by negative-stain EM, using 2% (w/v) uranyl formate solution as previously described (Januliene and Moeller, 2021). Negative-stain micrographs were recorded manually on a JEM-2100Plus transmission electron microscope (JEOL), operating 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). For cryo-EM, 3 μl of 0.6–0.9 mg/ml of freshly purified wt HOPS complex were applied onto glow-discharged CF grids (R1.2/1.3) (EMS) and immediately plunge-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific) with the environmental chamber set to 100% humidity and 4°C. Micrographs were recorded automatically with EPU (Thermo Fisher Scientific), using a Glacios microscope (Thermo Fisher Scientific) operated at 200 kV and equipped with a Selectris energy filter and a Falcon 4 detector (both Thermo Fisher Scientific). Images were recorded in Electron Event Representation (EER) mode at a nominal magnification of 130,000 (0.924 Å per pixel) in the defocus range from −0.8 to −2.8 µm with an exposure time of 8.30 s resulting in a total electron dose of approximately 50 e Å−2.

Cryo-EM image processing

Request a detailed protocol

All cryo-EM data processing (Figure 1—figure supplement 1 ) was performed in cryoSPARC v3.3.1 (Punjani et al., 2017). For all collected movies, patch motion correction (EER upsampling factor 1, EER number of fractions 40) and patch contrast transfer function estimation were performed using cryoSPARC implementations. To solve the structure of the core part of HOPS (Figure 1—figure supplement 1E,F and Figure 1—figure supplement 2) reference-free blob particle picking on 2186 pre-processed movies of the first data set and particle extraction using a box size of 672 pixels (px, 0.924 Å per pixel) binned to 128 px was performed. Extracted particles were subjected to 2D classification to eliminate bad picks. Selected good 2D classes (representative good 2D classes are shown in Figure 1—figure supplement 3A) were used for template particle picking on 2186 movies from the first data set combined with additional 6580 movies from the second data set, preprocessed alike. After removal of duplicates, picked particles were extracted with the same box size and subjected to rounds of 2D classification, ab-initio reconstruction with multiple classes and 3D heterogeneous refinement to remove false positive particle picks. From the best class, particles were extracted using a box size of 672 px (0.924 Å per pixel) binned to 320 px and subjected to 2D classification. Particles from this 2D classification were also used for flexibility analysis (see below). 407,996 selected particles from good 2D classes were used for ab-initio reconstruction with six classes and followed by 3D heterogeneous refinement. This heterogeneous refinement resulted in two best classes containing 130,009 and 116,151 particles, respectively, which were further refined individually. For each of both classes, a Non-Uniform (NU)-refinement was performed, followed by particle extraction with a box size of 672 px (0.924 Å per pixel, without binning) with homogeneous and NU-refinement afterward. Resulting consensus maps were used for local refinements of different parts of the structure (Figure 1—figure supplement 1A, Figure 1—figure supplement 2, and Figure 1—figure supplement 3B-E). First consensus NU-refinement that reached global resolution of 4.2 Å (Fourier shell correlation [FSC]=0.143) was used for local refinements of the upper ‘core’ part of the complex (4 Å) and the SNARE-binding module (3.6 Å), the second consensus NU-refinement, resolved to 4.4 Å (FSC=0.143), was used for local refinements resolving bottom parts of the complex (4.4 and 5 Å).

To better resolve distal parts of the complex, flexible Vps39 and Vps41 N-terminal fragments, the following approach was used (Figure 1—figure supplement 1B). Here, micrographs from the first two data sets were combined with micrographs from two additional data sets of 2841 and 8338 movies. After template picking, about 3896 million particles were extracted using a box size of 882 pixels (0.924 Å per pixel) and used for rounds of heterogeneous, homogeneous, and NU-refinements to obtain 3D reconstructions, which best resolve Vps39 and Vps41. At these steps, binning was applied for particle extractions. Then, particles belonging to one of such best classes were subjected to a round of NU-refinement followed by 3D variability analyses using masks covering either Vps39 or Vps41. The further 3D variability display procedures were used to better sort particles. Finally, the best particles were extracted with the same box size (882 pixels, 0.924 Å per pixel) without binning and used for local refinements of Vps39 or Vps41 (Figure 1—figure supplement 1B, Figure 1—figure supplement 3F,G).

For all local refinements, masks were generated in UCSF Chimera (Pettersen et al., 2004). During processing, no symmetry was applied. The quality of final maps is demonstrated in Extended Data Figure 3. All FSC curves were generated in cryoSPARC. Local resolutions of locally-refined maps (Figure S3B-G) were estimated in cryoSPARC and analyzed in UCSF ChimeraX (Pettersen et al., 2021). Data set statistics can be found in Supplementary file 2.

To analyze the flexibility of the complex, a different processing scheme was applied (Figure 1—figure supplement 1A, dashed arrows). For this, a set of 383,881 good particles was selected after 2D classification of 320 px-binned particles (initial box size of 672 px with 0.924 Å per pixel). These particles were subjected to either ab-initio reconstruction and heterogeneous refinement with ten classes (Figure 2—figure supplement 3A,B) or several refinement rounds followed by 3D variability analysis (Figure 2—figure supplement 3D). Prior to 3D variability analysis, ab-initio reconstruction with one class, homogeneous and NU-refinement were performed.

Model building and refinement

Request a detailed protocol

Models of HOPS subunits were initially generated using AlphaFold (Jumper et al., 2021; Varadi et al., 2022) and docked into locally refined maps using ‘Fit in Map’ tool in UCSF ChimeraX (Pettersen et al., 2021). The N-terminal parts of Vps41 (residues 1–863) and Vps39 (residues 1–700) and the C-terminal part of Vps16 (from residue 739) with no well-resolved densities assigned, were removed. The AlphaFold model of Vps11 was initially split into two parts (‘Vps11top’, residues 1–760, and ‘Vps11bottom’, residues 784–1025), which were first refined separately; the region predicted by AlphaFold as unfolded (residues 761–783) was deleted. Fitting of the C-terminal parts of Vps11 (residues 784–1025) and Vps39 (residues 701–1045) was improved using Namdinator (Kidmose et al., 2019). Afterward, models of single proteins were manually adjusted and refined in COOT (Emsley et al., 2010), followed by iterative rounds of refinements against corresponding locally refined and their composite maps in Phenix (Liebschner et al., 2019) and COOT. In Phenix Graphical User Interface, real space refine tool (Afonine et al., 2018) with or without option ‘rigid body’ was used. After several refinement rounds, two separate models were created by joining models of the upper (Vps33, Vps16, Vps41, Vps18, and Vps11 top) and lower (Vps11 bottom, Vps39) parts of the complex. These partly combined models were subjected to further several iterations of refinements in Phenix and COOT. Afterward, the two refined models were fused into a single model, which was again refined in Phenix and COOT. Sequences of the model in the bottom part of the complex were changed to polyalanines (residues 784–1025 in Vps11, residues 1–493 in Vps18, and residues 701–1045 in Vps39), since no assignment of side chains was possible at the resolution obtained there. In other parts of the complex, where blurred densities did not allow unambiguous model building, respective short fragments of the model were also replaced by polyalanine chains or deleted. Afterward, the initially deleted part of Vps11 (residues 761–783) was built de novo according to the cryo-EM density (residues 769–783 were replaced by alanines). Finally, the complete model was subjected to another round of refinement in Phenix followed by manual refining in COOT. Model validation was performed using MolProbity (Williams et al., 2018). Figures were prepared using UCSF ChimeraX. Model refinement and validation statistics can be found in Supplementary file 2.

ALFA pulldowns for mass spectrometry

Request a detailed protocol

One liter of YP medium containing 2% glucose (v/v) was inoculated with an overnight preculture. Cells were grown to OD600 1 at 26°C, followed by 1 hr incubation at 38°C. Cultures were harvested by centrifugation at 4800×g for 10 min at 4°C. Pellets were washed with cold Pulldown buffer (PB), 150 mM KAc, 20 mM HEPES/NaOH, pH 7.4, 5% (v/v) glycerol, and 25 mM CHAPS. The pellet was resuspended in a 1:1 ratio (w/v) in PB. supplemented with Complete Protease Inhibitor Cocktail (Roche) and afterward dropwise frozen in liquid nitrogen before lysed in 6875D LARGE FREEZER/MILL (SPEX SamplePrep LLC). Powder was thawed on ice and resuspended in PB by using a glass pipette, followed by two centrifugation steps at 5000 and 15,000×g at 4°C for 10 and 20 min. The supernatant was added to 12.5 µl prewashed ALFA Selector ST beads (2500×g, 2 min, 4°C) (NanoTag Biotechnologies) and incubated for 15 min at 4°C while rotating on a turning wheel. After incubation, beads were washed two times in PB and four times in PB without CHAPS. Samples were digested using PreOmics sample kit (iST Kit, preomics) and analyzed in Q ExativePlus mass spectrometer (Thermo Fisher Scientific).

GST pulldowns

Request a detailed protocol

Nucleotide-specific interaction of the Rab-GTPase Ypt7 with purified HOPS variants was analyzed in GST pulldowns using GST-Ypt7 or GST-Ypt1 as a negative control. About 125 µg purified Rab-GTPases was preloaded with 1 mM GDP or GTP in the presence of 20 mM EDTA and wash buffer (150 mM NaCl, 50 mM HEPES-NaOH, pH 7.4, 2 mM MgCl2, 0.1 % Triton X-100) in a water bath for 30 min at 30°C. For nucleotide stabilization, 25 mM MgCl2 was added. Prewashed GSH Sepharose 4B (GE Healthcare) was added to loaded Rab-GTPases and incubated for 1 hr at 4°C on a turning wheel. Beads were centrifuged for 1 min at 300×g before adding 25 µg of respective HOPS variants, followed by 1.5 hr incubation at 4°C on a turning wheel. Beads were washed three times with wash buffer, followed by two elution steps with 600 µl wash buffer containing 20 mM EDTA. After an incubation at room temperature for 20 min while rotating, the supernatant was TCA precipitated. About 10% of the final sample was loaded on a 7.5% SDS gel next to 5% of protein input. Samples were analyzed via western blot using antibodies against the FLAG-Tag. Beads were boiled in 50 µl Laemmli buffer. About 2% of the sample was loaded on an 11% SDS gel for Coomassie staining as Rab-GTPase loading control. Bands were quantified relative to the Rab-GTPase content.

Tethering assay

Request a detailed protocol

HOPS-mediated tethering assays were performed as described (Füllbrunn et al., 2021). For this, ATTO488 labeled liposomes were prepared and loaded with prenylated Ypt7 (Langemeyer et al., 2018). About 50 nmole liposomes was incubated with 50 pmole pYpt7:GDI complex together with GTP for 30 min at 27°C. For reactions, 0.17 mM Ypt7-loaded liposomes were incubated with different concentrations of HOPS complex or buffer (300 mM NaCl, 20 mM HEPES-NaOH, pH 7.4, 1.5 mM MgCl2, 10 % (v/v) glycerol) for 10 min at 27°C, followed by sedimentation for 5 min at 1000×g. The overall tethered liposomes in the pellet fraction were determined in a SpectraMax M3 fluorescence plate reader (Molecular Devices) comparing the ATTO488 fluorescent signal of the supernatant before and after sedimentation.

Fusion assays

Request a detailed protocol

Fusion assays and the purification of all proteins were performed as described (Langemeyer et al., 2018) with a protein to lipid ratio of 1:8000. Reconstituted proteoliposomes (RPLs) were composed of the vacuole mimicking lipid (VML) mix (Zick and Wickner, 2014). One population of RPLs carried the SNARE Nyv1 and the other set contained Vti1 and Vam3. RPLs were preloaded with prenylated Ypt7 with the help of 100 mM Mon1-Ccz1 and 0.5 mM GTP (Langemeyer et al., 2018). Then, 25 nM HOPS complex, 50 nM Sec18, 600 nM Sec17, and finally 100 nM Vam7 were added. Fusion of liposomes was followed by content mixing of the RPLs and subsequent increase in fluorescence which was monitored in a SpectraMax M3 fluorescence plate reader (Molecular Devices).

Cloning and protein purification of CtVPS391-500

Request a detailed protocol

Codon-optimized synthetic DNA (GenScript) encoding amino acids 1–500 of Vps39 from C. thermophilum (CtVp391-500; NCBI XP_006691033) was subcloned into a modified pET28a expression vector yielding an N-terminally His6-SUMO-tagged fusion protein (His6-SUMO-CtVps391-500). In short, His6-SUMO-CtVps39 was purified by Ni-NTA affinity chromatography followed by proteolytic cleavage with SUMO protease at 4°C overnight. SEC was performed to separate CtVps391-500 from the expression tag and SUMO protease and yielded >95% pure protein. To obtain phase information for structure determination, selenomethionine-substituted CtVps391-500 was prepared according to well-established methods (Doublié, 1997).

Limited proteolysis and protein crystallization

Request a detailed protocol

Since initial crystallization approaches did not yield protein crystals suitable for X-ray structure determination, flexible parts of CtVps391-500 were removed by limited proteolysis with α-Chymotrypsin (Merck) at 37°C for 10 min. Limited proteolysis was stopped by addition of protease inhibitors (Pierce Protease Inhibitor Tablets, EDTA-free, Thermo Fisher Scientific), and an additional SEC in buffer containing 20 mM BIS-TRIS pH 6.5, 200 mM NaCl, and protease inhibitors (1:1000) was performed as a final polishing step. Best crystals were obtained by seeding at 12°C and a protein concentration of 7.5 mg/ml in a crystallization condition containing 0.1 M MES pH 7.25 and 20% PEG 2000 MME. Selenium-derivative crystals were flash-cooled in liquid nitrogen in latter condition with 25% glycerol as cryoprotectant.

Crystal structure determination

Request a detailed protocol

Anomalous X-ray data were collected from a single crystal at 100K at beamline P13, EMBL Hamburg, Germany and diffraction data were processed using XDSAPP3 (Krug et al., 2012). Phase determination using single-wavelength anomalous dispersion at the selenium peak was not successful. A manually trimmed AlphaFold (Jumper et al., 2021) model of CtVps39 1-500 was used as a search template in molecular replacement and yielded a single solution with a TFZ score of 24.8 in phenix.phaser (Adams et al., 2010; Liebschner et al., 2019; McCoy et al., 2007), which was then used for MR-SAD phasing in phenix.autosol (Adams et al., 2010; Liebschner et al., 2019; Terwilliger et al., 2009) and subsequent density modification and automated model building using phenix.autobuild (Adams et al., 2010; Liebschner et al., 2019; Terwilliger et al., 2009). Iterative cycles of model building in COOT (Emsley et al., 2010) and refinement using phenix.refine (Afonine et al., 2012; Liebschner et al., 2019; Terwilliger et al., 2009) led to a final model of CtVps391-500 with Rfactors of Rwork 26.9% and Rfree 29.8%. The model contained no Ramachandran outliers with 96.74% residues within favored regions. Crystallographic statistics are summarized in Supplementary file 3. PYMOL and UCSF ChimeraX were used for visualization and graphical analysis.

Data availability

All diffraction data are deposited in the PDB as indicated in the manuscript. PDB files are mentioned there.

The following data sets were generated
    1. Shvarev D
    2. Schoppe J
    3. Koenig C
    4. Perz A
    5. Fuellbrunn N
    6. Kiontke S
    7. Langemeyer L
    8. Januliene D
    9. Schnelle K
    10. Kuemmel D
    11. Froehlich F
    12. Moeller A
    13. Ungermann C
    (2022) EMDataResource
    ID EMD-14965. HOPS tethering complex from yeast, consensus map covering the upper part of the complex.
    1. Shvarev D
    2. Schoppe J
    3. Koenig C
    4. Perz A
    5. Fuellbrunn N
    6. Kiontke S
    7. Langemeyer L
    8. Januliene D
    9. Schnelle K
    10. Kuemmel D
    11. Froehlich F
    12. Moeller A
    13. Ungermann C
    (2022) EMDataResource
    ID EMD-14966. HOPS tethering complex from yeast, consensus map covering the bottom part of the complex.
    1. Shvarev D
    2. Schoppe J
    3. Koenig C
    4. Perz A
    5. Fuellbrunn N
    6. Kiontke S
    7. Langemeyer L
    8. Januliene D
    9. Schnelle K
    10. Kuemmel D
    11. Froehlich F
    12. Moeller A
    13. Ungermann C
    (2022) EMDataResource
    ID EMD-14967. HOPS tethering complex from yeast, local refinement map of the SNARE-binding module.
    1. Kiontke S
    2. Ungermann C
    3. Kuemmel D
    (2022) RCSB Protein Data Bank
    ID 7ZTY. Structure of Vps39 N-terminal domain from Chaetomium thermophilum.
    1. Shvarev D
    2. Schoppe J
    3. Koenig C
    4. Perz A
    5. Fuellbrunn N
    6. Kiontke S
    7. Langemeyer L
    8. Januliene D
    9. Schnelle K
    10. Kuemmel D
    11. Froehlich F
    12. Moeller A
    13. Ungermann C
    (2022) EMDataResource
    ID EMD-14968. HOPS tethering complex from yeast, local refinement map of the backbone part of the complex.
    1. Shvarev D
    2. Schoppe J
    3. Koenig C
    4. Perz A
    5. Fuellbrunn N
    6. Kiontke S
    7. Langemeyer L
    8. Januliene D
    9. Schnelle K
    10. Kuemmel D
    11. Froehlich F
    12. Moeller A
    13. Ungermann C
    (2022) EMDataResource
    ID EMD-14969. HOPS tethering complex from yeast, local refinement map of the bottom part of the complex (Vps18).
    1. Shvarev D
    2. Koenig C
    3. Perz A
    4. Fuellbrunn N
    5. Kiontke S
    6. Langemeyer L
    7. Januliene D
    8. Schnelle K
    9. Kuemmel D
    10. Froehlich F
    11. Moeller A
    12. Ungermann C
    (2022) EMDataResource
    ID EMD-14970. HOPS tethering complex from yeast, local refinement map of the bottom part of the complex (Vps39).

References

    1. Emsley P
    2. Lohkamp B
    3. Scott WG
    4. Cowtan K
    (2010) Features and development of coot
    Acta Crystallographica. Section D, Biological Crystallography 66:486–501.
    https://doi.org/10.1107/S0907444910007493

Decision letter

  1. Benjamin S Glick
    Reviewing Editor; The University of Chicago, United States
  2. Vivek Malhotra
    Senior Editor; The Barcelona Institute of Science and Technology, Spain
  3. Benjamin S Glick
    Reviewer; The University of Chicago, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Structure of the lysosomal membrane fusion machinery" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Benjamin S Glick as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Vivek Malhotra as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

All of the reviewers are excited about the novel (and beautiful) structure of HOPS. This advance is a major step forward for the field, and it is especially important because HOPS has been studied so intensively. However, the reviewers are also unanimous in expressing concern about the functional interpretations. The conclusion that HOPS has a rigid core is not adequately justified. This uncertainty, combined with the wobbly membrane connections at the ends of the complex, cast doubt on the idea that HOPS acts as a stiff scaffold to promote fusion. In addition, Reviewer #2 points out that Vps33 is unlikely to remain engaged with the SNAREs as they zipper. Unless the authors can bolster their arguments far more effectively, the manuscript needs to be thoroughly rewritten to remove the unsupported claims.

Instead, the authors are encouraged to mine the structural data for further insights based on novel aspects of the structure, comparisons with other multi-subunit tethers, and fresh interpretations of the extensive literature on HOPS. This manuscript has the potential to be an elegant contribution if the structure is placed in a more appropriate context.

These suggestions are fleshed out in the detailed comments provided by the reviewers.

Reviewer #1 (Recommendations for the authors):

The data are well presented, and I have no detailed recommendations.

Reviewer #2 (Recommendations for the authors):

I suspect that the structure itself, with extensively intertwined subunits, is also novel. Overall, the paper is a must-accept for eLife, yet the following points also must be addressed, necessitating substantial re-writing but not requiring additional experiments.

Some questions about the paper and its presentation:

1. Repeated claims are made about which parts of the structure are flexible and which are rigid, yet there are no measurements of structural bending vs force applied ie, rigidity. If parts of the structure showed little variability among the many images analyzed, I presume that this was taken as a measure of rigidity; is there literature to support this, and if so it should be cited and discussed. How do the imputed bending energies compare to the binding energies for the things which bind to HOPS, such as Ypt7 and the SNAREs?

2. The model appears to claim that Vps33 would bind Nyv1 and Vam3 and then Vps33 will continue to be engaged with the SNAREs as the SNAREs zipper. What is the basis for this claim? Conventional thought is that each of these SNAREs is bound to a site on Vps33 with their R or Q residue and the apolar heptad repeat residues facing "in" to the binding site, in parallel and in the register as presented in 2015 by Baker et al., but that for zippering to proceed these side chains will have to leave these sites and face each other in the center of the 4-helical 4-SNARE coiled coils.

The structure of HOPS certainly is in accord with its well-established function as a tether and shows that Vps33 is off to the side between the two Ypt7 binding sites, Vps39 and Vps41, well-positioned to engage SNAREs to perform their alignment. It's unclear how the structure goes beyond that functionally. This is not to diminish the beauty and novelty of the structure, but to question their repeated claims that the structure now reveals substantial new insights about function.

3. Figures 1 and 2 are models of clarity and elegance.

4. With all respect for the effort that went into the Figure 3 experiment, it adds little to the paper or to mechanistic understanding. Unsurprisingly, the deletion in Vps41 blocks both tethering and fusion, while the deletions in Vps11 and 18 block fusion but not tethering. Figure 3 simply suggests that there's more to HOPS than a rigid structure and 3 active sites, one with Sec1/Munc18 function and two for tethering by means of binding Ypt7.

5. Readers will want to know whether Vps39 and 41 structures are similar to structures of other proteins that bind stoichiometrically to Rabs. Are there other proteins with a similar overall structure as the HOPS structure presented here, or is this extended structure with extensively interdigitated chains unique? Substantial additional discussion to place this structure in the context of others would build on the core of the paper, the determination of HOPS structure.

6. An earlier structure of crosslinked HOPS from this group proposed that each of the 6 subunits had an overall globular fold and that these subunits had binding surfaces adjacent to each other. In contrast, the current structure shows that there is extensive interdigitation of the subunits. Is this seen for many other proteins?

7. The claim that "HOPS complexes lacking β-propellers in Vps11 or Vps18 showed an alteration in the relative orientation of Vps39 within the complex, which indicates a more flexible backbone, indicating that backbone integrity is essential for full HOPS activity" (lines 207 to 209) is particularly puzzling. There are no data about whether these deletions are more or less flexible, and it's unclear how this can be inferred from an altered Vps39 orientation. There are no data showing that the loss of HOPS activity is caused by loss of rigidity.

The paper should be re-written to remove claims that it reveals new aspects of function or of rigidity but should emphasize that this is a novel structure.

Reviewer #3 (Recommendations for the authors):

With regard to the structural analyses, it would be useful to speculate on the possible cause(s) for the difference between this structure and previous lower resolution structures, perhaps illuminating some additional (perhaps biologically relevant?) differences? Is the rigid structure shown in this manuscript the only one possible for HOPS? Or are there additional conformations accessible in vivo that are not represented with this purified cryoEM sample? Does the fact that HOPS was massively overexpressed for purification indicate that certain post-translational modifications may not be present?

Are the resolution and map really suitable to accurately model the RING finger domain residues (Figure 2 and Figure 2-S1), or to map disease-associated mutations (Figure 4-S1)?

In Figure 3-S1 and the text, how good is the evidence that these Ypt7 binding sites predicted by Alphafold are accurate, especially as the location does not make a lot of biological sense with the location of the ALPS motif and membrane binding of Ypt7?

The authors' focus on the "rigid core" of HOPS, with respect to how this could help facilitate SNARE-mediated fusion, is a bit unclear, especially with the noted flexibility of both the Vps39 and Vps41 ends. Also, it is unclear how to interpret the angle differences of these flexible regions, in terms of overall length changes and possible changes during the tethering and fusion reactions.

With regard to this rigid core, the claim is that a number of helices "are tightly interlocked through coiled-coil motifs" (Line 127, Page 8), but the images in Figure 2 do not appear to be particularly interlocked or tight. Perhaps a calculation of buried surface might support this description?

Interpretation of the negative stain images in Figure 3D (especially with vps18-deltaN) is not entirely convincing, especially as the gel in Figure 1-S4 seems to indicate that Vps41 is missing from the vps18-deltaN pulldown? Were these mutants analyzed by mass spectrometry (similar to that for the RING finger mutants in Figure 2-S2?)? It would also be helpful to speculate more mechanistically about why these complexes can tether, but not drive fusion.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Structure of the HOPS tethering complex, a lysosomal membrane fusion machinery" for further consideration by eLife. Your revised article has been evaluated by Vivek Malhotra (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but the reviewers are not yet fully satisfied with the changes. They ask that claims of rigidity either be backed up with solid evidence or else presented as untested speculation, and that claims that this study tells us about how HOPS works be removed from earlier parts of the manuscript and moved to the Discussion as a model.

Below are the comments of the three reviewers. These issues need to be addressed for publication in eLife.Reviewer #1 (Recommendations for the authors):

– While the authors are free to propose their favored interpretations, you need to separate speculation from evidence-based interpretation.

– Rigidity has a well-defined meaning as resistance to deformation induced by stress. Nothing in this study measures rigidity. The central core of HOPS may be conformationally stable to thermal fluctuations as an isolated complex, but there is no way to predict how resistant it would be to deformation when stressed by a process such as SNARE zippering.

– The authors propose that Vps33 lets go of the SNAREs during zippering but that due to other interactions, HOPS remains tightly enough associated with the SNAREs to promote fusion. That idea might be correct but no experimental support is provided, so it is a weak link in the model.

– Another weak link is that the flexibility of the Ypt7-binding ends of HOPS are still challenging to reconcile with the idea that HOPS is a rigid scaffold.

– The bottom line is that as before, this study represents a major advance by describing a high-quality structure of HOPS. But the authors need to take seriously our request to separate speculation from interpretation, and to tone down the conclusion that this structure has revealed a new mechanism.

Reviewer #2 (Recommendations for the authors):

The authors clearly thought that his manuscript only needed a few minor tweaks, and didn't do a thorough re-write or come to grips with our central points.

Ungermann and his colleagues are still confusing protein stability with rigidity. Well-resolved parts of structures must be stable, but may or may not be rigid, the former term referring to the energy needed to unfold, either locally or globally, the latter to the energy needed to deform conformation. The data give no indication on rigidity, while the authors strongly leans on rigidity in speculations about HOPS function. Ungermann hasn't addressed this at all, and the two references he cites in his reviewers responses are neither cited nor discussed in the revised manuscript.

The Abstract remains sadly misleading, stating that "HOPS is surprisingly rigid" and "The SNARE binding module is rigidly attached to the core.." (see above). The last sentence, "Our results explain HOPS dual functionality and unravel why tethering complexes are …essential…fusion" is therefore very inaccurate.

There are no data on rigidity, as opposed to which regions of HOPS fold stably, but the manuscript persists in this vein. At the bottom of p6, they state "..it was unknown how tethering and fusion activities of HOPS are linked mechanistically. To clarify this…" but what follows addresses what a tethering structure of HOPS binding its Rab might look like through predictions and modeling. This doesn't address how "tethering and fusion activities of HOPS are linked mechanistically".

They show that removing the β-propeller of Vps41 alters HOPS structure, but confuse this with an altered rigidity.

The last paragraph before the Discussion is speculation and should be explicitly labeled as such and put in the Discussion. They present no evidence of HOPS continued association with the SNAREs as zippering proceeds. Their speculative model wasn't altered at all, and still shows Vps33 associating with the SNAREs as zippering progresses, though he acknowledges that this is very unlikely.

In short, the authors didn't take our reviews seriously. Re-reading the decision letter now, it was very clear that we wanted a thorough re-write, but we didn't get it.

I suggest that we return the paper to the authors, and that they take our reviews to heart and thoroughly re-write this paper.

Reviewer #3 (Recommendations for the authors):

Overall, I am still excited about this structure, and what it suggests for how tethering complexes can function. However, although some good changes have been made to the revised version, some of the conclusions are more speculative than the text would suggest.

Some of the changes in the revision still do not adequately address all the reviewers' concerns. Most importantly, the idea of the central core being "rigid" is clearly supported by lack of other conformations in the cryoEM data, but that conclusion is still speculative with regard to the conformation during the fusion reaction in vivo, and needs to be softened a bit in the text. The fact that other conformations are not observed in the structure indicates that the purified complex under these conditions does not seem flexible. However, the structure may just represent the lowest energy, stable state. The data does not indicate that this region is "rigid" in a way that suggests that it cannot bend under other conditions, e.g. in vivo. Certainly future experiments will help understand this, but the readers should not come away with this idea as definitive.

Although a little more information regarding other tethering complexes/membrane fusion machinery was added, it still is likely to be more relevant for a specialized readership, not a broad eLife audience.

This previous concern was not addressed: "With regard to the structural analyses, it would be useful to speculate on the possible cause(s) for the difference between this structure and previous lower resolution structures, perhaps illuminating some additional (perhaps biologically relevant?) differences?" It is unclear how improvements and acceleration in the sample preparation helped yield a "rigid" molecule vs. previous studies.

"In Figure 3-S1 and the text, how good is the evidence that these Ypt7 binding sites predicted by Alphafold are accurate, especially as the location does not make a lot of biological sense with the location of the ALPS motif and membrane binding of Ypt7?" For this concern, I respectfully disagree that all Alphafold predictions are that accurate, especially if the biological conclusion is unclear, and this speculation should be better explained or the conclusion softened.

https://doi.org/10.7554/eLife.80901.sa1

Author response

Reviewer #1 (Recommendations for the authors):

The data are well presented, and I have no detailed recommendations.

We thank the Reviewer for her/his kind comment.

Reviewer #2 (Recommendations for the authors):

I suspect that the structure itself, with extensively intertwined subunits, is also novel. Overall, the paper is a must-accept for eLife, yet the following points also must be addressed, necessitating substantial re-writing but not requiring additional experiments.

We thank the Reviewer for the kind words and helpful comments. Below and in our revised manuscript, we have addressed the concerns raised.

Some questions about the paper and its presentation:

1. Repeated claims are made about which parts of the structure are flexible and which are rigid, yet there are no measurements of structural bending vs force applied ie, rigidity. If parts of the structure showed little variability among the many images analyzed, I presume that this was taken as a measure of rigidity; is there literature to support this, and if so it should be cited and discussed. How do the imputed bending energies compare to the binding energies for the things which bind to HOPS, such as Ypt7 and the SNAREs?

Averaging of individual projection views is fundamental to cryo-EM. Therefore, during image processing, rigid particles (or their rigid parts) reach higher resolutions than flexible ones. According to our extensive 2D and 3D classification and variability analyses, we conclude that the core part of HOPS is the most rigid part of the complex, since it shows the highest resolution and better 2D-3D alignment (the general problem of internal particle flexibility in cryo-EM is also addressed in the literature, for example in Rawson et al. Methods 2016 and Punjani and Fleet J. Struct. Biol. 2021).

The affinities between HOPS and Ypt7 are relatively low (in the μM range); given the rigidity of the complex with its large hydrophobic interface (see reply to reviewer #1), the binding energies will therefore be minor.

2. The model appears to claim that Vps33 would bind Nyv1 and Vam3 and then Vps33 will continue to be engaged with the SNAREs as the SNAREs zipper. What is the basis for this claim? Conventional thought is that each of these SNAREs is bound to a site on Vps33 with their R or Q residue and the apolar heptad repeat residues facing "in" to the binding site, in parallel and in the register as presented in 2015 by Baker et al., but that for zippering to proceed these side chains will have to leave these sites and face each other in the center of the 4-helical 4-SNARE coiled coils.

The structure of HOPS certainly is in accord with its well-established function as a tether and shows that Vps33 is off to the side between the two Ypt7 binding sites, Vps39 and Vps41, well-positioned to engage SNAREs to perform their alignment. It's unclear how the structure goes beyond that functionally. This is not to diminish the beauty and novelty of the structure, but to question their repeated claims that the structure now reveals substantial new insights about function.

3. Figures 1 and 2 are models of clarity and elegance.

We thank the Reviewer for her/his nice words.

4. With all respect for the effort that went into the Figure 3 experiment, it adds little to the paper or to mechanistic understanding. Unsurprisingly, the deletion in Vps41 blocks both tethering and fusion, while the deletions in Vps11 and 18 block fusion but not tethering. Figure 3 simply suggests that there's more to HOPS than a rigid structure and 3 active sites, one with Sec1/Munc18 function and two for tethering by means of binding Ypt7.

We respectfully disagree with the reviewer, as Figure 3 adds important experimental perspectives. For example, we were surprised that the deletion of the b-propeller domains of Vps11 and Vps18 resulted in a complex with the same tethering ability like the wild-type HOPS, but with almost complete inactivity with respect to fusion. As Vps39 changes its relative orientation in the 2D classes of the mutants in comparison to wild-type HOPS, we conclude that the particle gains flexibility. In our view, the mutant HOPS has lower specific activity to promote fusion. We agree that there is more to this, but our observation is very similar to findings of Baker et al., 2015. Here, a Vps33 mutation resulted in a HOPS complex that could promote fusion of liposomes with a high SNARE density like wild-type. However, at low SNARE density, this complex was inactive, whereas the wild-type HOPS promoted fusion. We find it important to correlate structure to function and, therefore, consider Figure 3 an essential part of our study.

5. Readers will want to know whether Vps39 and 41 structures are similar to structures of other proteins that bind stoichiometrically to Rabs. Are there other proteins with a similar overall structure as the HOPS structure presented here, or is this extended structure with extensively interdigitated chains unique? Substantial additional discussion to place this structure in the context of others would build on the core of the paper, the determination of HOPS structure.

To our knowledge, such an extended structure that includes intertwined α solenoids, coiled coils and RING finger domains which together serve to locate functional modules (β-propellers) to the periphery is quite unique. For instance, the exocyst complex that mediates the tethering of post-Golgi secretory vesicles to the plasma membrane has a completely different structure with α-solenoids and long α-helices (Mei et al. Nat. Struct. Mol. Biol. 2018). EEA1 or similar long coiled-coil tethers function completely different. They do not bind SM proteins, nor do they bind SNAREs. Their binding to Rabs is thus not comparable. We followed the reviewer’s suggestion and discuss these aspects now in more detail in the manuscript.

6. An earlier structure of crosslinked HOPS from this group proposed that each of the 6 subunits had an overall globular fold and that these subunits had binding surfaces adjacent to each other. In contrast, the current structure shows that there is extensive interdigitation of the subunits. Is this seen for many other proteins?

The previous structure of HOPS was obtained using the method of negative-stain TEM, which can only provide low resolution enabling only domain-based interpretations which necessarily lead to uncertainties during the interpretation. Our current results are improved dramatically and demonstrate the most detailed structure of HOPS to date. Consequently, we were only now able to accurately access the molecular structure.

7. The claim that "HOPS complexes lacking β-propellers in Vps11 or Vps18 showed an alteration in the relative orientation of Vps39 within the complex, which indicates a more flexible backbone, indicating that backbone integrity is essential for full HOPS activity" (lines 207 to 209) is particularly puzzling. There are no data about whether these deletions are more or less flexible, and it's unclear how this can be inferred from an altered Vps39 orientation. There are no data showing that the loss of HOPS activity is caused by loss of rigidity.

The paper should be re-written to remove claims that it reveals new aspects of function or of rigidity but should emphasize that this is a novel structure.

We suggest that altering of Vps39 orientation in the mutants can be an indication of increased flexibility of the complex. We agree with the Reviewer that more data is needed in order to confirm the importance of backbone rigidity for HOPS functioning. We have revised the manuscript (lines 209-211) with words “which might be a result of increased flexibility of mutant HOPS due to the lack of structural support by β-propellers of Vps18 and Vps11, indicating that wild-type HOPS rigidity is essential for its full activity”. Currently, we are preparing additional mutants to probe the rigidity of the complex, which will be in the scope of another manuscript in the future.

Reviewer #3 (Recommendations for the authors):

With regard to the structural analyses, it would be useful to speculate on the possible cause(s) for the difference between this structure and previous lower resolution structures, perhaps illuminating some additional (perhaps biologically relevant?) differences? Is the rigid structure shown in this manuscript the only one possible for HOPS? Or are there additional conformations accessible in vivo that are not represented with this purified cryoEM sample? Does the fact that HOPS was massively overexpressed for purification indicate that certain post-translational modifications may not be present?

We cannot exclude that other conformations or modifications of HOPS exist in vivo. We also agree with the Reviewer that the solved structure might allow for other conformations in vivo, possibly due to binding partners such as Rabs. Our preliminary data did not reveal that addition of Ypt7 in very high concentrations altered HOPS structure. However, further studies involving additional methods like cryo-ET or analyzing HOPS in the membrane context will be necessary to get more insight into the structure and mechanism of HOPS in vivo.

Are the resolution and map really suitable to accurately model the RING finger domain residues (Figure 2 and Figure 2-S1), or to map disease-associated mutations (Figure 4-S1)?

We agree with the Reviewer that the resolution in the region of Vps11 RING finger domain is not high enough to assign the residues unambiguously, hence we refrained from showing any side chains in the figure. However, the resolution in the upper part of the complex is sufficient for accurate residue assignment. In Figure 2 and Figure 2-S1 we attempted to highlight the general architectural similarity between Vps18 and Vps11 modular elements composed of a long helix followed by a RING finger domain and did not to focus on specific residues of the RING finger domains. In Figure 2-S1, where we additionally compared the structures of these modular elements, we, due to limitations caused by the resolution of our map, assigned the residues not only based on our structure but also with the help of respective AlphaFold models which we used for model building.

In Figure 4-S1, we aimed to show that most of the mutations are generally located at the interfaces between the subunits or in the elements important for subunit interactions without going into detail about the specific roles of residues affected by the mutations. We therefore think that the resolution of our structure is sufficient for this purpose.

In Figure 3-S1 and the text, how good is the evidence that these Ypt7 binding sites predicted by Alphafold are accurate, especially as the location does not make a lot of biological sense with the location of the ALPS motif and membrane binding of Ypt7?

Since AlphaFold has already demonstrated high quality of structure prediction of protein complexes (Evans et al. bioRxiv 2022), we have little reasons to not trust the accuracy of AlphaFold prediction. However, we are currently conducting mutational analyses and are planning further structural studies of HOPS and its partners in order to confirm or correct AlphaFold predictions of Ypt7-interaction sites.

The authors' focus on the "rigid core" of HOPS, with respect to how this could help facilitate SNARE-mediated fusion, is a bit unclear, especially with the noted flexibility of both the Vps39 and Vps41 ends. Also, it is unclear how to interpret the angle differences of these flexible regions, in terms of overall length changes and possible changes during the tethering and fusion reactions.

In our model, the flexibility of HOPS is limited to the distal termini of the complex, which might assist Ypt7 locating binding by HOPS and membrane tethering. However, when HOPS is already connected to both membranes, we assume it to be a much more rigid particle due to fixation of Vps41 and Vps39 at the membranes and because of the stiffness of HOPS core. The flexibility of Vps41 and Vps39 is in the range of 10-20°, which is not much. Moreover, there is evidence that the binding site to Ypt7 faces away from the membrane, thus limiting also the contribution of their hypervariable domains. This limited flexibility will position the stiff core at the membrane interface, when bound to SNAREs. Consequently, such a positioning would provide stability to the SNARE-binding module and thereby facilitate fusion.

With regard to this rigid core, the claim is that a number of helices "are tightly interlocked through coiled-coil motifs" (Line 127, Page 8), but the images in Figure 2 do not appear to be particularly interlocked or tight. Perhaps a calculation of buried surface might support this description?

The rigidity of the core is provided by the large area of interface between Vps11 and Vps18 alphasolenoids, which is 1972 A2 according to the PDBePISA tool. We have also added additional Supplement Figure to Figure 2 (Figure 2-S4) describing the interaction between Vps11 and Vps18 and revised the manuscript text in line 111 with words “…large interface area of 1972 Å2 which provides a…”. The intertwined long helices are rather essential for tight coupling of functional modules (Vps41, Vps39, SNARE-binding module) to the core of HOPS but not for the interaction between the core subunits Vps11/18.

Interpretation of the negative stain images in Figure 3D (especially with vps18-deltaN) is not entirely convincing, especially as the gel in Figure 1-S4 seems to indicate that Vps41 is missing from the vps18-deltaN pulldown? Were these mutants analyzed by mass spectrometry (similar to that for the RING finger mutants in Figure 2-S2?)? It would also be helpful to speculate more mechanistically about why these complexes can tether, but not drive fusion.

We suggest that due to the reduced rigidity of the complex, the mutants would not be able to fix the SNARE-module in place for effective SNARE-mediated fusion. For tethering rigidity is not as essential, since it is mediated mostly by the already quite flexible Vps41 and Vps39 subunits.

We made no observations that the mutant complex (HOPS Vps18-deltaN) is not stable. We also performed Cryo EM analysis where we could clearly see a full assembled complex. It contains Vps33 in the same stoichiometry like wild-type HOPS. Regarding Figure 1 S4, it looks like that there is one subunit missing due to tagging of the complex. In the mutants (Vps11-deltaN and Vps41-deltaN), Vps41 was tagged with a 3x-FLAG tag. In Vps18-deltaN the tag was placed instead to the C-terminus of Vps18 for more stability during purification. Therefore, it looks like that Vps41 is missing on the gel, since the subunit runs a little bit lower without tag and at the same size as Vps39. However, we observed densities for all six subunits as mentioned above.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #1 (Recommendations for the authors):

– While the authors are free to propose their favored interpretations, you need to separate speculation from evidence-based interpretation.

We have now moved any discussion and speculations to the Discussion section, including the presentation of Figure 4.

– Rigidity has a well-defined meaning as resistance to deformation induced by stress. Nothing in this study measures rigidity. The central core of HOPS may be conformationally stable to thermal fluctuations as an isolated complex, but there is no way to predict how resistant it would be to deformation when stressed by a process such as SNARE zippering.

We agree and removed or replaced the term “rigidity” throughout the text by “stable” as suggested by the reviewer. Among the 2.5 Mio individual particles, the core is the portion with the highest resolution, suggesting stability, or as we initially named it, rigidity. However, as the reviewer points out future studies have to test whether this part is indeed rigid.

– The authors propose that Vps33 lets go of the SNAREs during zippering but that due to other interactions, HOPS remains tightly enough associated with the SNAREs to promote fusion. That idea might be correct but no experimental support is provided, so it is a weak link in the model.

We present a working model based on previous data, which is – as every model – an interpretation of the data. Multiple studies identified binding sites of HOPS for N-terminal domains for vacuolar SNAREs (e.g. Laage et al., MBoC 2001; Lürick et al., JBC 2015; Song et al., eLife 2020). In addition, HOPS binds the entire SNARE complex (Lürick et al., JBC 2015; Krämer et al., MBoC 2011), maybe also via the N-terminal parts.

For HOPS, we propose that SNAREs remain bound via their N-terminal domains, even if Vps33 lets go. However, even the last step is only suggested from intermediates of structures by the Rizo and Hughson lab (Stepien et al., Science Advances 2022; Baker et al., Science 2015). We thus feel that our current model is not an overinterpretation, but a working model that is based on the structure we obtained – which has to be put to the test. We are as curious as you and the reviewers which part is further confirmed by data and which part requires revision.

– Another weak link is that the flexibility of the Ypt7-binding ends of HOPS are still challenging to reconcile with the idea that HOPS is a rigid scaffold.

This is another part that we can only speculate on, and we do not claim that we know the entire answer. We did not determine structures of Ypt7 with Vps41 or Vps39 as they did not withstand cryo-conditions and therefore used AlphaFold2 to model the interactions. Time will tell, how reliable this is, but the binding sites fit to all previous work by us and others (e.g. Plemel et al., Traffic 2011; Ostrowicz et al., Traffic 2010; Lürick et al., MBoC 2017).

The Kd between small GTPases and effectors is in the µM range, but may be further supported by other membrane binding sites. We tested, however, binding of HOPS to lipidated Ypt7-GTP on supported lipid monolayers. Here, HOPS required Ypt7 to bind to these membranes and we showed that it was sitting here upright with the Vps39 site (which has the higher affinity) proximal to the membrane (Füllbrunn et al., eLife 2021). As we obtained very consistent data, we concluded that the association of HOPS with Ypt7 is anything but floppy, even though we observe some dynamics. However, one has to keep in mind that, in this experiment, we only follow the mobility of the GFP associated with HOPS, and not HOPS itself between membranes.

We speculate that the Vps41-Ypt7 interface is also stable as HOPS drives tethering (Füllbrunn et al., eLife 2021; this study). And it may work this way, if Ypt7 binds to the membrane distal site of Vps41. Here, the hypervariable domain may be critical to allow Ypt7 to reach the site – very similar to the way TRAPPII can activate only Ypt31 (Bagde et al., Science Advances 2022).

We discuss part of this in the text and would like to keep it to this. We made sure that we here separate observations and models from interpretation.

– The bottom line is that as before, this study represents a major advance by describing a high-quality structure of HOPS. But the authors need to take seriously our request to separate speculation from interpretation, and to tone down the conclusion that this structure has revealed a new mechanism.

Please have a look. We think that we did.

Reviewer #2 (Recommendations for the authors):

The authors clearly thought that his manuscript only needed a few minor tweaks, and didn't do a thorough re-write or come to grips with our central points.

We respectfully disagree. However, we have now further adapted our manuscript and think we made an even stronger effort in responding to all requests.

Ungermann and his colleagues are still confusing protein stability with rigidity. Well-resolved parts of structures must be stable, but may or may not be rigid, the former term referring to the energy needed to unfold, either locally or globally, the latter to the energy needed to deform conformation. The data give no indication on rigidity, while the authors strongly leans on rigidity in speculations about HOPS function. Ungermann hasn't addressed this at all, and the two references he cites in his reviewers responses are neither cited nor discussed in the revised manuscript.

Please see our response to reviewer #1.

The Abstract remains sadly misleading, stating that "HOPS is surprisingly rigid" and "The SNARE binding module is rigidly attached to the core.." (see above). The last sentence, "Our results explain HOPS dual functionality and unravel why tethering complexes are …essential…fusion" is therefore very inaccurate.

We adjusted the abstract accordingly.

There are no data on rigidity, as opposed to which regions of HOPS fold stably, but the manuscript persists in this vein. At the bottom of p6, they state "..it was unknown how tethering and fusion activities of HOPS are linked mechanistically. To clarify this…" but what follows addresses what a tethering structure of HOPS binding its Rab might look like through predictions and modeling. This doesn't address how "tethering and fusion activities of HOPS are linked mechanistically".

We agree that we did not test this, but rigidity was a term used to describe the part with the highest resolution among the 2.5 Mio particles analyzed by cryo-EM. We now went through the text and removed or replaced this term throughout.

They show that removing the β-propeller of Vps41 alters HOPS structure, but confuse this with an altered rigidity.

We agree. We observe an altered structure, which allows for tethering. The complex still has Vps33 bound. We interpret this as a less stable structure, and now rewrote the text accordingly.

The last paragraph before the Discussion is speculation and should be explicitly labeled as such and put in the Discussion. They present no evidence of HOPS continued association with the SNAREs as zippering proceeds. Their speculative model wasn't altered at all, and still shows Vps33 associating with the SNAREs as zippering progresses, though he acknowledges that this is very unlikely.

The reviewer is right, we moved this part to the discussion. As we wrote above, this is a working model and we do not go into the details here, how Vps33 may act. We feel that any further alterations of the model would not be more correct as it remains a working model, which is based on structure of intermediates (Munc18 with SNAREs; Vps33 with R and Q-SNAREs) and interactions between the N-terminal domains and HOPS – and our interpretation in light of our structure.

In short, the authors didn't take our reviews seriously. Re-reading the decision letter now, it was very clear that we wanted a thorough re-write, but we didn't get it.

I suggest that we return the paper to the authors, and that they take our reviews to heart and thoroughly re-write this paper.

We hope that this reviewer acknowledges that we made an effort to include his/her suggestions throughout the text.

Reviewer #3 (Recommendations for the authors):

Overall, I am still excited about this structure, and what it suggests for how tethering complexes can function. However, although some good changes have been made to the revised version, some of the conclusions are more speculative than the text would suggest.

Some of the changes in the revision still do not adequately address all the reviewers' concerns. Most importantly, the idea of the central core being "rigid" is clearly supported by lack of other conformations in the cryoEM data, but that conclusion is still speculative with regard to the conformation during the fusion reaction in vivo, and needs to be softened a bit in the text. The fact that other conformations are not observed in the structure indicates that the purified complex under these conditions does not seem flexible. However, the structure may just represent the lowest energy, stable state. The data does not indicate that this region is "rigid" in a way that suggests that it cannot bend under other conditions, e.g. in vivo. Certainly future experiments will help understand this, but the readers should not come away with this idea as definitive.

We thank the reviewer for pointing this issue out to us. Our data revealed that the core of the HOPS complex, the backbone of Vps11 and Vps18 and the associated Vps16 and 33 subunits, had the highest resolution. In addition, the Vps11 and Vps18 subunits form a long hydrophobic interface. We agree that this does not indicate rigidity against distortion and thus removed this wording and the implications from the Results section. In the discussion, we speculate on the implications of the structure for HOPS function. We therefore moved also Figure 4 to the discussion as requested by the other reviewers as well and went through the entire text again to soften our interpretation and separate conclusions from speculations.

Although a little more information regarding other tethering complexes/membrane fusion machinery was added, it still is likely to be more relevant for a specialized readership, not a broad eLife audience.

We now extended this part further to describe the crosstalk between tethering complexes and membrane fusion. One main problem is, however, that functional data on tethering complexes is largely available from work on HOPS, whereas full reconstitution is mostly lacking for others. We acknowledge this and also mention thus the limitations in the discussion.

This previous concern was not addressed: "With regard to the structural analyses, it would be useful to speculate on the possible cause(s) for the difference between this structure and previous lower resolution structures, perhaps illuminating some additional (perhaps biologically relevant?) differences?" It is unclear how improvements and acceleration in the sample preparation helped yield a "rigid" molecule vs. previous studies.

We included now further explanations in the text regarding the sample preparation. There are several important improvements that helped us to solve the structure. One key factor was a protocol, where cells were cryo-lysed, affinity purification occurred via the FLAG tag on HOPS subunits, and the complex was eluted fast via the FLAG peptide from beads, whereas previously a TAP-tagged HOPS was used (where elution occurred over night). This is mentioned at the beginning of the results part and in detail described in the methods part. In brief, we accelerated all intermediate steps, improved concentration conditions and gel filtration and in the end had just a lot more intact particles to analyze. All these issues are mentioned, though we did not do a methods discussion. Of course, these improvements do not result in a different structure of the complex but more intact particles that we were able to analyze. When we previously analyzed HOPS, we had to use GRAFIX (an overnight glycerol gradient) to stabilize the complex, and the preparation took almost two day (Bröcker et al., 2012). Moreover, previously we and others used negative stain EM for analysis which has numerous well-known disadvantages for structure interpretation, for example flattening and strong inaccuracies for the euler angle assignment which easily can lead to wrong models and interpretations. Using cryoEM we can examine the complex in solution at very low temperatures, which diminishes drawbacks in negative stain.

"In Figure 3-S1 and the text, how good is the evidence that these Ypt7 binding sites predicted by Alphafold are accurate, especially as the location does not make a lot of biological sense with the location of the ALPS motif and membrane binding of Ypt7?" For this concern, I respectfully disagree that all Alphafold predictions are that accurate, especially if the biological conclusion is unclear, and this speculation should be better explained or the conclusion softened.

The reviewer is right, we can only interpret the AlphaFold model and need mutagenesis of the respective sites for confirmation of the Ypt7 binding sites. We noted however that these predictions match previous analyses (e.g. Plemel et al., Traffic 2011; Ostrowicz et al., Traffic 2010; Lürick et al., MBoC 2017), such as the deletion of the b-propeller of Vps41, which abolished tethering (Figure 3). We extended this part in the text and acknowledge the limitations of our interpretation.

https://doi.org/10.7554/eLife.80901.sa2

Article and author information

Author details

  1. Dmitry Shvarev

    Department of Biology/Chemistry, Structural Biology section, Osnabrück University, Osnabrück, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Jannis Schoppe and Caroline König
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9776-268X
  2. Jannis Schoppe

    Department of Biology/Chemistry, Biochemistry section, Osnabrück University, Osnabrück, Germany
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - review and editing
    Contributed equally with
    Dmitry Shvarev and Caroline König
    Competing interests
    No competing interests declared
  3. Caroline König

    Department of Biology/Chemistry, Biochemistry section, Osnabrück University, Osnabrück, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - review and editing
    Contributed equally with
    Dmitry Shvarev and Jannis Schoppe
    Competing interests
    No competing interests declared
  4. Angela Perz

    Department of Biology/Chemistry, Biochemistry section, Osnabrück University, Osnabrück, Germany
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Nadia Füllbrunn

    Department of Biology/Chemistry, Biochemistry section, Osnabrück University, Osnabrück, Germany
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Stephan Kiontke

    Department of Plant Physiology and Photo Biology, University of Marburg, Marburg, Germany
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5822-913X
  7. Lars Langemeyer

    Department of Biology/Chemistry, Biochemistry section, Osnabrück University, Osnabrück, Germany
    Contribution
    Data curation, Supervision, Validation, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4309-0910
  8. Dovile Januliene

    Department of Biology/Chemistry, Structural Biology section, Osnabrück University, Osnabrück, Germany
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3279-7590
  9. Kilian Schnelle

    Department of Biology/Chemistry, Structural Biology section, Osnabrück University, Osnabrück, Germany
    Contribution
    Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8808-594X
  10. Daniel Kümmel

    Department of Chemistry and Pharmacy, Institute of Biochemistry, University of Münster, Münster, Germany
    Contribution
    Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Visualization, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3950-5914
  11. Florian Fröhlich

    Department of Biology/Chemistry, Molecular Membrane Biology group, Osnabrück University, Osnabrück, Germany
    Contribution
    Data curation, Formal analysis, Funding acquisition, Validation, Visualization, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8307-2189
  12. Arne Moeller

    1. Department of Biology/Chemistry, Structural Biology section, Osnabrück University, Osnabrück, Germany
    2. Center of Cellular Nanoanalytic Osnabrück (CellNanOs), Osnabrück University, Osnabrück, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Visualization, Methodology, Writing - original draft, Project administration
    For correspondence
    arne.moeller@uni-osnabrueck.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1101-5366
  13. Christian Ungermann

    1. Department of Biology/Chemistry, Biochemistry section, Osnabrück University, Osnabrück, Germany
    2. Center of Cellular Nanoanalytic Osnabrück (CellNanOs), Osnabrück University, Osnabrück, Germany
    Contribution
    Conceptualization, Supervision, Funding acquisition, Validation, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    cu@uos.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4331-8695

Funding

Deutsche Forschungsgemeinschaft (UN111/5-6)

  • Christian Ungermann

Deutsche Forschungsgemeinschaft (INST190/196-1 FUGG)

  • Arne Moeller

Bundesministerium für Bildung und Forschung (BMBF/DLR 01ED2010)

  • Arne Moeller

Deutsche Forschungsgemeinschaft (SFB 944 P11)

  • Christian Ungermann

Deutsche Forschungsgemeinschaft (SFB 944 P27)

  • Arne Moeller

Deutsche Forschungsgemeinschaft (SFB 944 P20)

  • Florian Fröhlich

Deutsche Forschungsgemeinschaft (SFB 944 P16)

  • Daniel Kümmel

Deutsche Forschungsgemeinschaft (MO 2752/3-6)

  • Arne Moeller

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors thank Kristian Parey for help with model building, and all members of the Ungermann and Moeller lab for feedback. This work was supported by the SFB 944 (P11 to CU; P16 to DK, P20 to FF, P26 to AM), the DFG (UN111/5-6 to CU and MO 2752/3-6 to AM), the DFG INST190/196-1 FUGG (AM) and the BMBF/DLR 01ED2010 (AM).

Senior Editor

  1. Vivek Malhotra, The Barcelona Institute of Science and Technology, Spain

Reviewing Editor

  1. Benjamin S Glick, The University of Chicago, United States

Reviewer

  1. Benjamin S Glick, The University of Chicago, United States

Publication history

  1. Preprint posted: May 5, 2022 (view preprint)
  2. Received: June 8, 2022
  3. Accepted: September 12, 2022
  4. Accepted Manuscript published: September 13, 2022 (version 1)
  5. Version of Record published: October 21, 2022 (version 2)

Copyright

© 2022, Shvarev, Schoppe, König et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,467
    Page views
  • 906
    Downloads
  • 4
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Dmitry Shvarev
  2. Jannis Schoppe
  3. Caroline König
  4. Angela Perz
  5. Nadia Füllbrunn
  6. Stephan Kiontke
  7. Lars Langemeyer
  8. Dovile Januliene
  9. Kilian Schnelle
  10. Daniel Kümmel
  11. Florian Fröhlich
  12. Arne Moeller
  13. Christian Ungermann
(2022)
Structure of the HOPS tethering complex, a lysosomal membrane fusion machinery
eLife 11:e80901.
https://doi.org/10.7554/eLife.80901

Further reading

    1. Cell Biology
    2. Immunology and Inflammation
    Ana J Caetano, Yushi Redhead ... Paul T Sharpe
    Research Article Updated

    The interplay among different cells in a tissue is essential for maintaining homeostasis. Although disease states have been traditionally attributed to individual cell types, increasing evidence and new therapeutic options have demonstrated the primary role of multicellular functions to understand health and disease, opening new avenues to understand pathogenesis and develop new treatment strategies. We recently described the cellular composition and dynamics of the human oral mucosa; however, the spatial arrangement of cells is needed to better understand a morphologically complex tissue. Here, we link single-cell RNA sequencing, spatial transcriptomics, and high-resolution multiplex fluorescence in situ hybridisation to characterise human oral mucosa in health and oral chronic inflammatory disease. We deconvolved expression for resolution enhancement of spatial transcriptomic data and defined highly specialised epithelial and stromal compartments describing location-specific immune programs. Furthermore, we spatially mapped a rare pathogenic fibroblast population localised in a highly immunogenic region, responsible for lymphocyte recruitment through CXCL8 and CXCL10 and with a possible role in pathological angiogenesis through ALOX5AP. Collectively, our study provides a comprehensive reference for the study of oral chronic disease pathogenesis.

    1. Cell Biology
    2. Developmental Biology
    Eunjin Cho, Xiangguo Che ... Tae-Hoon Lee
    Research Article

    Peroxiredoxin 5 (Prdx5) is involved in pathophysiological regulation via the stress-induced cellular response. However, its function in the bone remains largely unknown. Here, we show that Prdx5 is involved in osteoclast and osteoblast differentiation, resulting in osteoporotic phenotypes in Prdx5 knockout (Prdx5Ko) male mice. To investigate the function of Prdx5 in the bone, osteoblasts were analyzed through immunoprecipitation (IP) and liquid chromatography combined with tandem mass spectrometry (LC–MS/MS) methods, while osteoclasts were analyzed through RNA-sequencing. Heterogeneous nuclear ribonucleoprotein K (hnRNPK) was identified as a potential binding partner of Prdx5 during osteoblast differentiation in vitro. Prdx5 acts as a negative regulator of hnRNPK-mediated osteocalcin (Bglap) expression. In addition, transcriptomic analysis revealed that in vitro differentiated osteoclasts from the bone marrow-derived macrophages of Prdx5Ko mice showed enhanced expression of several osteoclast-related genes. These findings indicate that Prdx5 might contribute to the maintenance of bone homeostasis by regulating osteoblast differentiation. This study proposes a new function of Prdx5 in bone remodeling that may be used in developing therapeutic strategies for bone diseases.