Metazoan cell biology of the HOPS tethering complex

Abstract

Membrane fusion with vacuoles, the lysosome equivalent of the yeast Saccharomyces cerevisiae, is among the best undestood membrane fusion events. Our precise understanding of this fusion machinery stems from powerful genetics and elegant in vitro reconstitution assays. Central to vacuolar membrane fusion is the multi-subunit tether the HOmotypic fusion and Protein Sorting (HOPS) complex, a complex of proteins that organizes other necessary components of the fusion machinery. We lack a similarly detailed molecular understanding of membrane fusion with lysosomes or lysosome-related organelles in metazoans. However, it is likely that fundamental principles of how rabs, SNAREs and HOPS tethers work to fuse membranes with lysosomes and related organelles are conserved between Saccharomyces cerevisiae and metazoans. Here, we discuss emerging differences in the coat-dependent mechanisms that govern HOPS complex subcellular distribution between Saccharomyces cerevisiae and metazoans. These differences reside upstream of the membrane fusion event. We propose that the differences in how coats segregate class C Vps/HOPS tethers to organelles and domains of metazoan cells are adaptations to complex architectures that characterize metazoan cells such as those of neuronal and epithelial tissues.

Fundamental Insights from the Yeast Saccharomyces cerevisiae

The HOPS complex was originally discovered in the yeast Saccharomyces cerevisiae. Genetic screens in S. cerevisiae for mutant strains deficient in the localization of lysosomal enzymes revealed phenotypes in vacuole morphology and vacuolar protein secretion.1–4 The gene products of these strains came to be referred to as the vacuole protein sorting (Vps) proteins.3 Vps mutant strains were classified as class A–F based on vacuolar characteristics and morphology.1,3 Among those are the class B and C Vps genes, which encode subunits of the HOPS complex. Class B Vps strains were characterized by fragmentation of vacuoles3 and class C Vps strains were characterized by the absence of any visible vacuolar compartment.3 The class B proteins Vps39 and Vps41 and the class C proteins Vps11, Vps16, Vps18, and Vps33 assemble in a hexameric complex referred as the HOmotypic fusion and Protein Sorting (HOPS) complex (Fig. 1).7–9 The class C proteins Vps11, Vps16, Vps18, and Vps33 assemble in an additional hexameric complex with the Vps3 and Vps8 subunits referred as the class C core vacuole/endosome tethering complex (CORVET, Fig. 1).10,11 In the HOPS complex the class B and C subunits play specific roles in the tethering, docking, and fusion stages of membrane fusion.3,5,8,9,11–15 A similar function is performed by the class C Vps proteins present in the CORVET complex.10,11 For outstanding reviews please refer to Nickerson et al.,12 Wickner,16 Brocker et al.,15 and Epp et al.17

Elegant studies by Ostrowicz and Plemel lead to a model for HOPS complex organization.5,6 In this model, the class C subunits reside as an extended core of the complex with Vps33 interacting with membrane SNAREs while the class B subunits Vps39 and Vps41 interact with the Rab7-GTPase at the apposed membrane.5,6 This organization of HOPS complex subunits fits with the known functions of individual subunits. The core subunit Vps33 is a member of the Sec1/Munc18 protein family.18–20 Interactions of Vps33 with t-SNAREs through the Sec1/MUNC18 domain may have a role in the recognition of target-specific SNAREs, thereby aiding in trans-SNARE pairing or, may inhibit non-specific trans-SNARE pairs.21 The class B HOPS subunits Vps39 and Vps41 have specificity for interaction with the Rab7 S. cerevisiae homolog Ypt7. In S. cerevisiae, the HOPS subunit Vps39 has been shown to interact with Ypt7. Though Vps39 was originally proposed to serve as the GEF for this vacuolar Rab Ypt7, recent studies in vivo suggest Vps39 is not the Rab7 GEF.9,22,23 Instead, the dimeric Mon1-Ccz1 complex is the Rab7/Ypt7 GEF.24 The subunit Vps41 has also been shown to directly bind Ypt7 and act as the effector for Ypt7 in vacuolar tethering events.12,25 In addition to interactions with the Rab7 homolog, Vps41 has also been shown to interact with the vesicle adaptor AP-3 in S. cerevisiae.23,26–28

Metazoan Mutations in HOPS Subunits Reveal Diverse Functions of the HOPS Complex

S. cerevisiae studies have been instrumental in understanding the role HOPS subunits have in regulating tethering, SNARE pair formation, and fusion of organelles with the vacuolar compartment.1,3–5,7–9,12–15,26–42 Metazoan genetic deficiencies are consistent with the role of the HOPS complex in the delivery of vesicle contents to lysosomes and lysosome-related organelles. However, the existence of HOPS subunit isoforms, metazoan-specific HOPS subunits, such as SPE-39, as well as partially overlapping distribution or function of HOPS subunits along the endocytic route suggest a more complex picture in metazoans than in yeast.43–47

Vps33 exists as two isoforms (a and b) encoded by different genes in C. elegans, Drosophila melanogaster and Homo sapiens. Vps33A and Vps33B are not redundant in Drosophila suggesting distinct tissue and possibly organellar functions.48 Moreover, the Hermansky-Pudlak syndrome (HPS) and Arthrogryposis, Renal Dysfunction and Cholestasis (ARC), affecting Vps33a and Vps33b respectively, further support this idea in mammals. HPS is characterized by occulocutanoeous pigment dilution, prolonged bleeding, and pulmonary fibrosis.49–51 Symptoms of HPS arise from defects in sorting to lysosomes and lysosome-related organelles such as the melanosome and platelet dense granules.49,50,52 In humans, a subset of HPS patients displays mutations to the gene encoding the β subunit of AP-3.53,54 Models for HPS have been discovered in mouse and Drosophila.53,54 In mice, mutations to the genes that encode the clathrin adaptor AP-3 subunits AP-3 β1 and AP-3 δ1 as well as the class C HOPS subunit Vps33a result in decreased coat color, prolonged bleeding, and defects in lysosomal protein targeting.49,52,55,56 Drosophila homologs of class B and C HOPS subunits Vps33a, Vps41, Vps18, and adaptor protein AP-3 subunits also result in pigmentation defects.48,52,57–61 Importantly, mammalian phenotypes in AP-3 and Vps33a deficiencies are distinct from those of patients carrying mutations in Vps33b, a defect that leads to the ARC syndrome. ARC results from mutations to the gene responsible for encoding the class C HOPS subunit Vps33b or the Vps33b-interacting protein SPE-39, which is absent in yeast.46,62,63 Spe39 is also referred as VIPAR in Homo sapiens, a nomenclature that we have disputed.62,64 In Drosophila SPE-39 is known as Vps16b65 despite evidence indicating that SPE-39 possesses unique domains not present in Vps16.45,46 SPE-39 robustly immunoprecipitates Vps16,46 thus suggesting that rather than a Vps16 isoform, SPE-39 is an additional component of class C Vps complexes such as HOPS and a putative mammalian CORVET complex (Fig. 1). Patients with ARC syndrome display a variety of phenotypes including severe contracture of joints referred to as arthrogryposis, defects in renal function, cholestasis, bleeding disorders, dry, thickened, scaly or flaky skin referred to as ichthyosis, defects in metabolic absorption, absence or severe size decrease of the corpus callosum and defective organization of the anterior horn of the spinal cord.62,63,66–84 Importantly, severe joint contracture in patients with ARC syndrome results from a neurological defect as opposed to muscular defects.63,75,80 Analysis of ichthyosis in ARC patients suggests defects in lamellar body secretion as shown by increased amounts of lamellar granules in patients with ARC syndrome by electron microscopy.68,69,72 Kidney epithelial cells and hepatocytes show a loss of apically targeted proteins.71 All of these phenotypes are consistent with a role for Vps33b and SPE-39 in regulation of lysosome-related organelles and secretion in polarized mammalian cell types. Thus, differences between HPS and ARC phenotypes affecting isoforms of class C Vps proteins may reflect differential tissue distribution of Vps33 isoforms. Alternatively, Vps33 isoforms expressed in the same cell may mediate either distinct tethering events between organelles and target membranes whose identities are specified by Vps33a, Vps33b, and SPE-39 and/or differences in cargoes trafficked by SPE-39-, Vps33a-, or Vps33b-dependent mechanisms.

Signaling receptors are attractive cargo candidates for explanation of phenotypic differences in HPS and ARC patients. For example, epidermal growth factor and Notch receptor degradation are impaired in HOPS complex deficiencies.46,85 Notch signaling is critical for embryonic development of neural and epidermal tissues and in the adult Notch modulates synaptic plasticity.86,87 Notch regulation could contribute to the severe neuronal and neuromuscular symptoms in patients with ARC syndrome. In the canonical Notch signaling pathway, Notch binds extracellular ligands and undergoes cleavage at the extra- and intracellular domains.85,86 Following cleavage, the intracellular domain is transported to the nucleus where it regulates gene expression.85,86 In addition, there is an endosomal pathway for Notch activation where Notch receptor is internalized in the absence of extracellular ligand.85 Internalized Notch is targeted into intraluminal vesicles in multi-vesicular bodies where receptor degradation terminates signaling.85 However in the endosomal pathway, which is AP-3- and HOPS-dependent, Notch is sorted away from intralumenal vesicles and remains at the multi-vesicular body limiting membrane for delivery to lysosomes where the intracellular domain undergoes cleavage and translocates to the nucleus.85,86 The proposed role of HOPS subunits in this study was in regulating the fusion of late endosomes with the lysosome.85 Such a model is consistent with the presence of HOPS in late endosomes, as demonstrated in yeast88 and mammalian cells (see below). In such a model an AP-3 pool localized to incoming vesicles would encounter HOPS complexes present in late endosomes resulting in fusion of AP-3 vesicles with late endosomes (see Fig. 2A).28,88,89 An alternative yet not exclusive view, considers that since HOPS and AP-3 interact28,47,88–90 and AP-3 and HOPS are present in vesicle carriers47 (Fig. 2B), perhaps the HOPS complex subunits may regulate Notch receptor sorting away from multivesicular bodies, to AP-3-HOPS clathrin-coated vesicles for delivery to lysosomes.

Comparative Cell Biology of S. cerevisiae and Metazoan HOPS-Coat Interactions

The role of HOPS in the membrane fusion event itself is well understood in S. cerevisiae thanks to in vitro vacuole fusion assays.16 However, a comparable assay in a metazoan system does not yet exist, limiting our ability to draw parallels with S. cerevisiae. Thus, we limit our discussion to HOPS subcellular localization and HOPS-coat interactions where information allows a preliminary comparison between unicellular and multicellular eukaryotes. The S. cerevisiae's HOPS subunits are localized to the vacuolar compartment.28,31,36,40,41,91 HOPS complex subunits have not been identified at the Golgi or endosomal intermediates in S. cerevisiae.28 Thus, it is postulated that in S. cerevisiae, cytosolic HOPS complex cycles on and off the target membrane.23 In contrast with this unique localization of HOPS in yeast, metazoan cells yield a different localization of HOPS subunits.46,47,92,93 Vps class C proteins Vps11, 16 and 18 colocalize with early and late endosome compartment markers in mammalian cells.92,94 This observation is consistent with the HOPS complex interaction with mammalian late endosomal SNARE syntaxin-7 and Rab7.92,95 However, some studies indicate that the class C Vps proteins Vps11, 16, 18, and 33b as well as the HOPS subunits Vps39 and Vps41 mostly localize to early endosomes but minimally with late endosomal and lysosomal markers.47,93 These data suggest that additional mechanisms beyond what is present in yeast have evolved to localize HOPS complexes to diverse endosomal compartments in metazoans.

The different HOPS localization patterns observed in S. cerevisiae and mammals could be due to changes in the organization of AP-3 trafficking mechanisms that evolved concurrent with multicellularity. In S. cerevisiae, the adaptor AP-3 recruits cargoes at the Golgi compartment for inclusion into vesicles that will be targeted to the vacuole, whereas in mammals, AP-3 functions at the early endosome for vesicle formation.96–100 Unlike mammals, S. cerevisiae AP-3 does not bind clathrin and clathrin is not required for the proper localization of AP-3 cargoes.97,100,101 In contrast, clathrin does bind AP-3 and clathrincoated vesicles contain AP-3 subunits in mammals.90,102–105

Early studies of HOPS subunits suggested interaction of AP-3 with the HOPS subunit Vps41 in S. cerevisiae.26,27 Direct interaction of the ear domain of AP-3 with Vps41 was presumed to occur with the isolated Vps41 subunit and not with the entire HOPS complex.26,27 These findings suggested that Golgi-derived AP-3 vesicles would require Vps41 as a coat complex in lieu of clathrin in S. cerevisiae.26 However, elegant experimentation by Angers et al.28 indicate AP-3 interacts with the whole hexameric HOPS complex rather than with the isolated Vps41 subunit. Furthermore, they showed that HOPS subunits are not associated with the Golgi, where S. cerevisiae AP-3 vesicle formation occurs, or on AP-3 vesicular intermediates.28 Instead the results by Angers et al. as well as Cabrera et al. suggested a model where coated or partially coated AP-3 vesicles reached a vacuole decorated with HOPS complexes. Interestingly, the AP-3 cargo kinase Yck3 phosphorylates the HOPS subunit Vps41, present at the vacuole, making HOPS permissive to bind AP-3 in the vesicle. It is at the vacuole that an AP-3-HOPS interaction materializes as a step of the membrane tethering-fusion process (Fig. 2A).28,88

Like in S. cerevisiae, mammalian AP-3 and HOPS complexes interact.47,90 However, the temporal and spatial organization of mammalian AP-3 and HOPS complex interactions differ from yeast in the following ways:

1. HOPS subunits interact and colocalize on early endosomes with the coats, AP-3 and clathrin;47

2. HOPS subunits co-fractionate with clathrin-coated vesicles;47

3. HOPS subunits localize to endosomal compartments in a clathrindependent manner as revealed by chemical-genetic perturbation of clathrin;47

4. HOPS subunits are distributed in a polarized manner by clathrin-dependant mechanism.47

The basic tenet of HOPS complex localization to the yeast vacuole is that it occurs through recruitment from the cytoplasm directly to the vacuolar membrane (Fig. 2A). This model supports class B and C Vps HOPS subunits localization exclusively to the vacuole in S. cerevisiae. However, this model is not sufficient to explain the clathrin-dependent subcellular localization of HOPS complexes in mammalian cells.47 Thus, we propose a novel and complementary model of class B and C Vps HOPS subunit localization in mammalian cells (Fig. 2B). In this model, HOPS localizes to early endosomes at sites of vesicle formation. HOPS complex subunits are included in clathrin-coated membranes as cargoes for traffic to the late endosome/lysosome or polarized domain of cells, such as those found in epithelial and neuronal cells (Fig. 2B).

The prevailing notion is that coats recruit target-specific cargoes for inclusion into vesicles destined for its target.106 However, how does a vesicle “know” its target location?107 Coats could confer information to vesicles for specific fusion with target organelles by inclusion of tethers and SNAREs at the vesicle formation stage (Fig. 2B). This view is not represented in canonical models of vesicle-mediated membrane traffic that have conceptually segregated vesicle formation from vesicle fusion machineries for analytical purposes.106 Our work demonstrated that tethers, such as the HOPS complex, are located on coated vesicles. These findings provide a mechanism for target recognition by coupling the vesicle formation and fusion machineries (Fig. 2B). In addition, coat-tether associations would provide a coat-dependent vesicular mechanism governing organelle-specific tether localization and delivery. These principles may be universal as suggested by the interaction between the coat COPII and the tether TRAPPI in coated vesicles targeted from the ER to the Golgi complex.108

Figures and Tables

Figure 1 Organization of the class C Vps proteins into the multisubunit tethers HOPS and CORVET. Diagrams depict the proposed architecture of the HOPS and CORVET complexes based on studies in S. cerevisiae5,6 and a putative organization of HOPS complexes in metazoans. Class C Vps subunits are represented by gray spheres and class B Vps subunits are diagrammed as green and blue spheres, CORVET-specific subunits are depicted by purple spheres. At least four HOPS complexes could exist defined by Vps33a and Vps33b isoforms and the binding of SPE-39 to Vps33 isoforms. The existence of the mammalian CORVET complex has not been documented yet. It may be possible that Vps33a and Vps33b isoforms and the binding of SPE-39 to Vps33 isoforms may define similarly diverse CORVET complexes.

Figure 2 Models of HOPS complex localization to early and late endosome/lysosome compartments. (A) HOPS complex localization to late endosome/lysosomes is specified by the properties of the late endosome/lysosomal membrane. HOPS complexes cycle on and off of the late endosome/lysosomal membrane. Interactions between coat and HOPS occur at the fusion stage where an incoming coated vesicle encounters the acceptor membrane that already contains the HOPS tether. (B) In a second model of HOPS complex localization, the HOPS complex is incorporated into newly forming vesicles at the donor endosome by interactions with, but not restricted to, the coat. A coated vesicle delivers the HOPS complex to a late endosome/lysosomal compartment. Coats confer information to vesicles for specific fusion with target organelles by inclusion of tethers and at the vesicle formation stage in this second model of HOPS localization.

Acknowledgments

This work was supported by grants from the National Institutes of Health to V.F. (NS42599 and GM077569) and S.W.L. (GM082932). S.A.Z. was supported by T32 GM008367, National Institutes of Health, Training Program in Biochemistry, Cell, and Molecular Biology. We are indebted to the Faundez lab members for their comments.

Addendum to: