Unveiling the cryo-EM structure of retromer

Abstract

Retromer (VPS26/VPS35/VPS29) is a highly conserved eukaryotic protein complex that localizes to endosomes to sort transmembrane protein cargoes into vesicles and elongated tubules. Retromer mediates retrieval pathways from endosomes to the trans-Golgi network in all eukaryotes and further facilitates recycling pathways to the plasma membrane in metazoans. In cells, retromer engages multiple partners to orchestrate the formation of tubulovesicular structures, including sorting nexin (SNX) proteins, cargo adaptors, GTPases, regulators, and actin remodeling proteins. Retromer-mediated pathways are especially important for sorting cargoes required for neuronal maintenance, which links retromer loss or mutations to multiple human brain diseases and disorders. Structural and biochemical studies have long contributed to the understanding of retromer biology, but recent advances in cryo-electron microscopy and cryo-electron tomography have further uncovered exciting new snapshots of reconstituted retromer structures. These new structures reveal retromer assembles into an arch-shaped scaffold and suggest the scaffold may be flexible and adaptable in cells. Interactions with cargo adaptors, particularly SNXs, likely orient the scaffold with respect to phosphatidylinositol-3-phosphate (PtdIns3P)-enriched membranes. Pharmacological small molecule chaperones have further been shown to stabilize retromer in cultured cell and mouse models, but mechanisms by which these molecules bind remain unknown. This review will emphasize recent structural and biophysical advances in understanding retromer structure as the field moves towards a molecular view of retromer assembly and regulation on membranes.

Introduction

Membrane trafficking is essential for human health and physiology. Trafficking pathways control and maintain the spatio-temporal organization of transmembrane proteins (sometimes called protein cargoes) within cells. Trafficking pathways enable diverse physiological processes, including nutrient uptake, synaptic transmission, signal transduction, and activation of the immune response. Many trafficking genes are essential for organism viability, while others are affected in a variety of acquired and genetic diseases, including neurodegenerative diseases. Endosomes are busy sorting hubs that direct protein cargoes for degradation or recycling [1,2]. Mutations in endosomal trafficking proteins, including retromer and sorting nexin (SNX) proteins, contribute to human brain disease. Retromer dysfunction is widely linked to neurological, neurodevelopmental and neurodegenerative conditions, including Alzheimer’s disease (AD), Parkinson’s disease (PD), and Down’s dyndrome (DS) [3–14]. In addition, multiple bacterial and viral pathogens [15,16] target retromer to hijack or modify the endosomal system during the cellular invasion.

Elucidating molecular mechanisms that govern membrane trafficking is therefore critical for understanding both behavior and etiology underlying these neurological disorders. A major outstanding question in cell biology is how mammalian retromer assembles with distinct SNXs to sort different cargoes to multiple destinations from a common origin, the endosome. Furthermore, we seek to understand how defects in the assembly of SNX/retromer contribute to the etiology and pathology of human disease. This review provides an update on the structural biology of retromer. Multiple advances in the past few years have fundamentally improved our understanding of SNX/retromer-mediated endosomal transport, with broad consequences for cell biology, organelle physiology, and human health and disease.

Retromer

Retromer (VPS26/VPS35/VPS29 subunits) is a cytoplasmic protein complex that plays a critical role in endosomal trafficking. Retromer was first identified and characterized in Saccharomyces cerevisiae [17], where it is required for retrograde trafficking of vacuolar cargoes including Vps10 to the trans-Golgi network (TGN). Retromer is now firmly established as an evolutionarily conserved heterotrimer that orchestrates the sorting of important receptor cargoes from the endosome to both the TGN and to the plasma membrane, thereby maintaining homeostasis of transmembrane cargoes at the plasma membrane and within the endosomal/lysosomal system [14,17–20].

Retromer (sometimes called cargo selective complex, or CSC, based on yeast studies) is composed of three vacuolar protein sorting (VPS) proteins (VPS26, VPS35, and VPS29 subunits) that form a stable, soluble heterotrimer [19–21]. VPS35 serves as the key structural component, because its helical solenoid structure provides a platform for binding VPS26 and VPS29. Mammals express two VPS26 orthologs, VPS26A and VPS26B [19,20,22–25]. VPS26 possesses an arrestin fold and binds the highly conserved N-terminal region of VPS35 [25,26]. VPS29 possesses a metallophosphoesterase fold and binds the VPS35 C-terminus [27–29]. Retromer heterotrimer is recruited to endosomes by binding short amino acid motifs or sorting signals found in integral membrane proteins. Retromer also directly binds multiple other membrane-associated cargo adaptors and accessory proteins, including SNXs, Rab GTPases, VARP, and the WASH complex [18,22,30–33]. SNX/retromer complexes associate with the cytosolic face of endosomal compartments to facilitate retrieval of transmembrane cargoes to the TGN and the plasma membrane [34–36]. Retromer was originally identified in yeast as a regulatory complex required to sort acid hydrolases to the endo-lysosomal network, and recent data suggest Vps10 and other receptors use a bipartite sorting motif to ensure recognition by retromer in yeast [37]. Work in mammalian systems later suggested retromer sorts cation-independent mannose 6-phosphate receptor (CI-MPR) [34]; this has recently been revisited in light of biochemical and structural work focused on SNX-BAR proteins (discussed below). More recent data has demonstrated retromer sorts a variety of cargoes, including the iron transporter (DMT1-II) [38], transferrin receptor [39], Wntless [40–43], glutamate receptors, and the amyloid precursor protein (APP) adaptor SorLA [44,45]. In humans, membrane recruitment of retromer is mediated by Rab7 (Ypt7 in yeast) on late endosomes through a direct interaction with N-terminal conserved regions in VPS35 and VPS26 [46]. TBC1D5, a putative Rab GTPase-activating protein (GAP) for Rab7, inhibits retromer recruitment in mammalian cells [33,47]. Genetic and structural analyses of different SNX/retromer complexes suggest retromer is a modular sorting device that associates with different SNXs (e.g. SNX-BARs, SNX3, SNX27) to establish cargo-specific sorting, membrane remodeling, and trafficking pathways [22,31,48–53].

Sorting nexins function as key cargo adaptor proteins

Retromer cannot bind lipids directly, so its recruitment to the endosome occurs through multiple protein–protein interactions. One important class of retromer-binding proteins are the SNXs, which associate with membranes through direct binding to phosphatidylinositol-3-phosphate (PtdIns3P) headgroups by PX domains. Humans have 49 SNX proteins, but only three classes are known to interact with retromer: the membrane tubulating SNX-BAR proteins [19,52,53]; the PX domain only SNX3 [19,49]; and the PX-FERM domain family member, SNX27 [19,50,51]. In the presence of different SNXs, retromer orchestrates tubulovesicular-based cargo sorting through three different endosomal pathways (Figure 1) that have been firmly established in the literature: SNX-BAR/retromer pathway [17]; SNX3–retromer pathway [49,54]; and SNX27–retromer pathway [36,50,51].

Figure 1. Overview of metazoan retromer-mediated trafficking pathways.

Metazoan retromer mediates three distinct endosomal pathways. The canonical SNX-BAR/retromer pathway also occurs in yeast, and it comprises the retromer heterotrimer together with SNX-BAR heterodimers (Vps5/17 in yeast; SNX1/2 with SNX5/6 in metazoans). This pathway retrieves cargoes like CI-MPR from endosomes to the TGN. The SNX3/retromer pathway also occurs in both yeast and metazoans; here, the retromer heterotrimer and SNX3 are implicated in sorting receptors in both yeast (e.g. Fet3/Ftr1 iron transporters) and metazoans (e.g. Wntless, WLS) from endosomes to the TGN. Finally, the metazoan-specific SNX27/retromer pathway mediates cargo recycling directly to the plasma membrane. In this pathway, cargoes including β2 adrenergic and glutamate receptors contain PDZ binding motifs recognized by SNX27.

SNX-BAR/retromer

Following recruitment to endosomal sub-domains, retromer and associated cargoes are concentrated in nascent membrane tubules generated by dimers of SNX proteins containing a bin/amphiphysin/rvs (BAR) domain [30,55]. In the yeast SNX-BAR/retromer pathway, the BAR domains of SNXs form heterodimers composed of Vps5 and Vps17 [17]. The Vps5/Vps17 heterodimer then associates with retromer. In metazoans, SNX-BAR family members include SNX1, SNX2, SNX5, and SNX6. All SNX-BAR proteins include two membrane-binding domains, the phox homology (PX) and BAR domains. The PX domain senses and stabilizes membrane curvature by binding to membrane phospholipids, which drives endosomal budding in yeast and mammalian cells [55–60]. The precise recruitment of SNX-BAR dimers to the endosomal membrane requires PX binding to the canonical early endosome component, PtdIns3P [41,61,62]. Additionally, SNX1 and SNX2 have also been reported to bind phosphatidylinositol bis-phosphate (PtdInsP₂) on late or maturing endosomes [60,63–65]. This difference in lipid binding ability gives rise to the idea of selective, spatio-temporal sorting of cargoes by SNX-BAR/retromer in conjunction with endosome maturation. Finally, retromer cargoes are thought to become concentrated at the ‘ends’ of membrane tubules that form retromer-positive vesicular compartments by the process of tubule scission. These carriers are then transported toward the TGN via the microtubule network [32].

ESCPE-1

Retromer’s role in the retrograde sorting of the CI-MPR to the TGN, remains controversial. Multiple studies in mammalian cells are consistent with retromer in regulating CI-MPR transport [34,35,66–70]. However, recent structural, biochemical, and functional evidence instead suggest how a SNX-BAR dimer (SNX5/SNX6) itself constitutes coat complex, named ‘Endosomal SNX-BAR sorting complex for promoting exit 1’ (ESCPE-1). The authors propose ESCPE-1 mediates retromer independent transport of transmembrane proteins, including CI-MPR, from endosomes to TGN through direct recognition of a bipartite sorting motif (Φ×Ω×Φ(x)_nΦ) in cytosolic tails (Φ, hydrophobic; Ω, aromatic; x, any amino acid with variable linker region) [53,70,71].

Using time-resolved analysis of cargo trafficking, Simonetti et al. showed acute retromer inactivation leads to robust defects in endosomal recycling of GLUT1 but no perturbation of CI-MPR distribution. In contrast, acute depletion of ESCPE-1 drives aberrant CI-MPR trafficking. These data suggest a limited role for retromer in ESCPE-1 dependent CI-MPR retrograde sorting [53,70,71]. Overall, increasing evidence using biochemical, structural, and functional methods suggest how retromer and SNX-BAR proteins (including ESCPE-1) have evolved into two functionally distinct sorting complexes.

SNX3/retromer

SNX3 contains only a PX domain that binds PtdIns3P, and it is known to play an important role in retromer recruitment and activity on endosomal membranes [19,72]. The SNX3–retromer pathway is implicated in sorting the divalent metal ion transporter Dmt1-II [38], transferrin receptor [39], Wntless receptors [41–43], and CI-MPR [69] to distinct carriers. Molecular details of SNX3-mediated cargo sorting were revealed by an X-ray crystal structure of a tripartite complex containing SNX3, VPS26, and N-VPS35 bound to a short peptide motif from Dmt1-II [73]. The structure revealed how SNX3 primarily binds VPS26, with a minor secondary binding site on the N-terminal region of the VPS35 α-helical solenoid. SNX3 and VPS26 interact to form a groove, which provides the binding surface for the cargo peptide. Cargo binding induces VPS26 to undergo a conformational change to engage SNX3, resulting in the formation of a combined binding surface for dual recognition of the cargo peptide. Based on these structural data, the authors propose retromer associates with membranes using two copies of SNX3 located at each end, thereby forming a stable membrane-associated SNX3/retromer complex. However, SNX3 lacks a BAR domain, so it remains unclear how SNX3/retromer mediates membrane remodeling to form endosomal transport carriers.

SNX27/retromer

The SNX27–retromer pathway is implicated in the recycling of cargo proteins from the endosome directly back to the cell surface [36,50,51]. SNX27 is unique to metazoans; it is a multi-domain scaffolding protein with an N-terminal psd95/dlg/zo-1 (PDZ) domain; central PX domain; and C-terminal 4.1/ezrin/radixin/moesin (FERM) domain. The PDZ domain acts as a cargo binding module by recognizing transmembrane proteins (ion channels, solute carriers, and GPCRs) via a highly specific class I PDZ-binding motif (PDZbm) present at the C- terminus of the cargo proteins. Additionally, the PDZ domain directly interacts with the retromer VPS26 subunit, which allosterically enhances cargo binding affinity for SNX27 [51]. The central PX domain binds PtdIns3P, which recruits SNX27 to early endosomes [74,75]. There is also a secondary phosphoinositide-binding site within the C-terminal of the FERM domain, which may also modulate its membrane-binding dynamics. The FERM domain also binds cargo proteins having the consensus sequence motif FxNPxY. Structural studies have demonstrated how retromer is able to interact simultaneously with cargo molecules and adaptor proteins [50]. The crystal structure of the SNX27 PDZ domain in complex with VPS26 revealed an exposed SNX27 β-hairpin responsible for engaging a conserved groove in VPS26. Interestingly, the association of SNX27 PDZ with retromer increases the affinity for PDZ binding motifs, suggesting cargo sorting is allosterically coupled to the formation of the SNX27/retromer assembly [50]. The role of retromer in SNX27-mediated cargo sorting and recycling is well documented in the literature, but how SNX27/retromer assembles with cargo to generate tubulovesicular carriers remains elusive.

Retromer models from structural and biophysical studies

In recent years, both cryo-electron (cryo-EM) microscopy and X-ray crystallography have provided molecular insights into assembly and structures of SNX/retromer-coated tubules [19,76]. The earliest crystal structures of different retromer subunits and sub-complexes revealed how individual subunits fold in three-dimensional space and interact with each other [25–28, 73, 77]. More recently, the first direct view of intact yeast retromer heterotrimer was determined using electron microscopy [78]. A major breakthrough came from a cryo-electron tomography (cryo-ET) reconstruction of thermophilic yeast retromer, which revealed the overall architecture of reconstituted retromer with a Vps5 homodimer [79]. Subsequently, the first structure of the mammalian retromer was determined using single-particle cryo-EM [80]. Finally, a recent study on retromer oligomerization on supported lipid bilayers (SLB) suggests mammalian retromer exists as low-order oligomers [48]. Together, these studies have improved our understanding of how retromer functions across eukaryotes. In discussions below, we use established conventions when referring to yeast (e.g. Vps35) versus mammalian (e.g. VPS35) proteins.

Yeast Vps5/retromer cryo-ET reconstructions

Kovtun and colleagues described the structure of retromer from the thermophilic yeast Chaetomium thermophilum assembled on membrane tubules with a Vps5 SNX-BAR homodimer (Figure 2A) [79]. The cryo-ET reconstruction revealed how the retromer heterotrimer adopts two different dimeric conformations when assembled in vitro on membranes in the presence of Vps5 BAR homodimers. The C-terminus of Vps35 subunits mediates the formation of repeating V-shaped arch-like structures by joining two copies of Vps35; Vps29 subunits reside on each side of the interface at the apex. Vps26 forms a symmetrical dimer and engages Vps5 via a complementary interaction with its BAR domain. The Vps26 homodimer acts as a ‘foot’ between each Vps35 α-solenoid ‘leg’ and the Vps5 inner layer, and thus Vps26 connects adjacent arches. These adjoining connections provide the platform for extended polymers of retromer arches to form a discrete outer layer and enable Vps5/retromer coats to adopt to differing membrane curvatures. Retromer membrane recruitment depends on the inner Vps5 layer to yield stable membrane coated tubules. The Vps5 BAR domain forms antiparallel homo- and heterodimers to decorate the inner layer of the coat: single layer interactions occur between tips of BAR domain dimers, and lateral interactions occur through PX domains. The architecture of the thermophilic yeast retromer is consistent with the crystal structure of human SNX3/VPS26/N-VPS35 described above (SNX3/retromer section), where retromer was proposed to associate with the bilayer via two copies of SNX3 positioned at each end [73]. Both the heterotrimer and SNX proteins are conserved across eukaryotes, so this seminal work provides a compelling model for SNX-BAR/retromer architecture on endosomal membranes.

Figure 2. Retromer structural models.

(A) Model of SNX-BAR/retromer centered on the VPS35/VPS35 dimer arch and based on cryo-ET reconstructions (PDB: 6H7W) generated from reconstituted thermophilic yeast retromer on membranes. VPS35 is shown in red ribbons; VPS29 in green ribbons; VPS26 in blue ribbons, and the SNX-BAR (yeast Vps5 homodimer) in yellow ribbons. Yeast retromer is considered to be a stable pentamer composed of Vps5/Vps17/Vps26/Vps35/Vps29 proteins (schematic); only Vps5 BAR as a homodimer was included in cryo-ET reconstructions. Both retromer and SNX-BAR proteins are highly conserved from yeast to humans, suggesting a compelling model for how SNX-BAR/retromer assembles on curved tubular membranes across eukaryotes. Binding sites for key protein partners are marked including Rab7; the RabGAP, TBC1D5; VARP; and FAM21. The VPS29 binding site for RidL, a Legionella effector protein, is also shown. (B) Retromer model centered on flat VPS35/VPS35 dimer, based on single-particle cryo-EM reconstructions of murine retromer chains (PDB: 6VAB). The membrane is shown here for clarity but was not included in structural studies. In the absence of SNX-BAR proteins, retromer alone forms elongated chains with a much flatter arch centered on the VPS35 dimer interface. Known VPS26 binding locations for SNX3 PX and SNX27 PDZ domains are marked. Increasing evidence suggests Retromer (VPS26/VPS35/VPS29) and SNX-BAR proteins (SNX1/SNX2 and SNX5/SNX6) have functionally diverged in mammals.

However, several questions remain about Retromer assembly, since this study used a Vps5 BAR homodimer for technical reasons. Yeast require both Vps5 and Vps17 for function; this structure contains only the Vps5 BAR domain and lacks Vps17 altogether. Both Vps5 and Vps17 contain elongated N-termini predicted to be unstructured, followed by a PX domain to mediate phospholipid binding. Yeast may require both proteins for regulatory purposes. Recent structural work from the Ford laboratory [81] reveals the structure of another yeast SNX-BAR protein, Mvp1. This protein uses its flexible N-terminus to auto-inhibit BAR dimer formation. It is tempting to speculate other SNXs may use their N-termini for regulation.

Mammalian retromer single-particle cryo-EM structures

A major outstanding question in the field has been whether metazoan retromer assembles in the same way as yeast retromer. Our group recently provided the first structural snapshots of multiple murine retromer oligomers using single-particle cryo-EM methods (Figure 2B) [80]. Two-dimensional class averages in ice reveal murine retromer forms multiple oligomers: the retromer heterotrimer; dimers of trimers; a tetramer of trimers; and elongated flat chains. These different oligomers suggest retromer may function as a flexible scaffold that is further ordered in the presence of SNXs and cargo embedded in membranes. Together, mammalian cryo-EM structures and yeast cryo-ET reconstructions reveal emerging principles of retromer assembly. Like yeast retromer, mammalian retromer forms dimers in solution and in ice, and both mammalian VPS26 and VPS35 subunits form homodimers. VPS26-mediated dimers are clearly observed in 2D classes of chains, but this sub-structure is not well-resolved. The murine VPS35-mediated dimer interface looks similar to the yeast Vps35 dimer observed at the top of the V-shaped arches, but the curvature of the mammalian VPS35 dimer interfaces observed in dimers and chains is different and flatter (Figure 2B). This flatness could bring arches closer to the membrane, or alternatively, these flat chains may suggest how retromer could bind SNX3 or SNX27 on flat portions of endosomal membranes. This raises an interesting question about whether retromer then would undergo a conformational change as it engages the SNX-BAR heterodimers that promote curved membranes. In either case, SNX27 or SNX3 binding may stabilize either curved arch-like retromer dimers or flatter chains either by restricting the conformation of VPS26 on a membrane or by positioning either SNX27 and SNX3 close to the membrane via the PDZ and PX domains, respectively. The SNX27 PX-FERM module would further orient the SNX27–retromer interface, while the PtdIns3P-binding pocket in SNX3 would likely place retromer in proximity to the membrane. Therefore, VPS26 subunits may form interfaces that extend retromer into repeating units with VPS29 subunits positioned at arch apexes. Further structural studies will be required to determine whether flat chains are observed in the presence of SNX proteins that lack BAR domains.

Retromer oligomers on supported lipid bilayers

Retromer has been proposed to induce cargo clustering prior to packaging into a nascent transport carrier when retromer associates with regulatory proteins. Recently, Deatherage and colleagues developed an elegant reconstituted retromer system on SLB and used single-particle fluorescence to demonstrate how mammalian retromer assembles alone and in the presence of key cellular binding partners [48]. These studies suggest mammalian retromer exists as monomers and low-order oligomers (dimers, trimers, tetramers) on SLB membranes. These data are consistent with single-particle cryo-EM data in which dimers and tetramers were the prevalent retromer species in vitrified ice. Frequency distribution of retromer complexes on the SLB further suggests that membrane association does not substantially influence retromer oligomeric state. Moreover, the addition of integral transmembrane cargoes, SNX3, Rab7, and/or the WASH complex component FAM21, either alone or in combination, do not further drive retromer oligomerization on SLB membrane [82]. Altogether, these findings suggest neither cargo nor accessory factors are sufficient to promote retromer oligomerization on a SLB. This raises important questions about how and whether the association of membrane or accessory factors can drive retromer coat oligomerization.

Implications and future directions

Retromer and human brain disease

Disruption of the endosomal system and mutations in machinery that control endosome function are now widely linked to neurological and neurodegenerative conditions. These diseases collectively have an enormous impact on society, affecting large numbers of people and placing huge financial and logistical burdens on health care systems. Neurodegenerative diseases are progressive, irreversible brain disorders characterized by a decline in motor and cognitive functions; histologically, these pathologies are often accompanied by protein aggregation and selective loss of neurons. Over the past decade, compounding evidence supports a direct link between retromer-mediated endosomal trafficking and onset of neurodegenerative diseases, including AD and PD [3–6,8,10,13,14]. AD is caused by abnormal accumulation of neurotoxic amyloid-beta (Aβ) peptides in the brain produced by the cleavage of APP. Following internalization, APP can be recycled to the cell surface, transported to the TGN, or sorted to the lysosome for degradation. Several lines of evidence support the involvement of retromer in AD pathology: reduced VPS25 and VPS26 protein levels in the brains of AD patients [83]; the role of retromer in recognition and trafficking of APP receptors and regulators; and the interaction of retromer with BACE1, an APP processing enzyme [84,85]. PD is the second most prevalent neurodegenerative disease, and it mainly affects people in the 60–65 year age group [86,87]. PD is pathologically defined by the loss of dopaminergic (DA) neurons and the accumulation of α-synuclein-enriched Lewy bodies [88]. Genome-wide association studies have provided extensive evidence that retromer mutations (e.g. VPS35 D620N) cause late-onset PD [89], where endosomal and lysosomal perturbations are believed to inhibit the clearance of α-synuclein [90]. The loss-of-function VPS35 D620N mutation also impairs neurotransmitter receptors and dopaminergic signaling in PD [91,92]. Overall, retromer regulates the trafficking of important receptors involved in neurodegeneration and hence can be considered as a potential therapeutic target.

Retromer stabilization by small molecules

Data from Petsko and colleagues suggested pharmacological small molecule chaperones can both stabilize retromer and enhance its function, and thus retromer is now believed to be a potential therapeutic target for neurodegenerative diseases [93]. The small molecule, R55, is chemically known as thiophene-2,5-diylbis(methylene) dicarbamimidothioate dihydrochloride and is predicted to bind at the VPS35/VPS29 interface. Binding studies established a low micromolar K_d (~5 μM) [93]. Recent work from Muzio and colleagues showed how R55 modification (using bis-guanylhydrazones connected by a 1,3-phenyl ring linker) further enhance retromer stability in an ALS mouse model [94]. Phenyl-1,3-bis-guanylhydrazone 2a (also called compound 2a) is one version of modified R55 that acts as a potent interactor at the VPS35/VPS29 interface [94]. This compound was shown to increase VPS35 levels in inducible pluripotent stem cell (iPSC)-derived motor neurons (MNs). It also appears to attenuate locomotion impairment in G93A mice and increases the number of surviving MNs [94]. This compound may be a promising therapeutic starting point to delay degenerative processes associated with ALS. However, substantial work remains to dissect the underlying structural basis for how potential small molecules can stabilize retromer. Structural studies would allow the field to better understand retromer assembly through the identification of small molecule binding site(s), which in turn will help us to understand the suitability of retromer as a drug target. This would open many exciting avenues in translational research and further suggests how an iterative process of small molecule drug screening coupled to structural biology is a worthwhile avenue to pursue.

Retromer and friends

Current data suggest retromer acts as a structural scaffold that is likely oriented by binding SNX-BAR proteins on endosomal membranes. However, retromer interacts with multiple additional important regulatory proteins, including Rab GTPases (Rab7); the RabGAP, TBC1D5; actin-remodelling proteins like the WASH complex; and Vps9-ankyrin repeat protein (VARP). We lack structural data for many of these interactions, and thus many important molecular and functional questions remain about retromer coat assembly and regulation in endosomal trafficking.

• How do SNX3/retromer and SNX27/retromer coats assemble on membranes?

• Do SNX3 and SNX27 interact with SNX-BAR/retromer? Or do they form distinct coats?

• What is the molecular basis of the retromer–WASH interaction?

• How does VARP associate with retromer?

Ongoing structural studies research will undoubtedly uncover the distinct functions of retromer-linked complexes in constitutive endosomal trafficking pathways and will pave the way to allow us to understand how and why retromer mutations are linked to human brain disease (Table 1).

Table 1.

Retromer mutations linked to neurodegenerative diseases

Retromer subunit mutation	Disease	References
VPS35-D620N	Parkinson’s disease (PD)	[12,13,95–102]
VPS35-P316S	Parkinson’s disease (PD)	[12,13,102,103]
VPS35-R524W	Parkinson’s disease (PD)	[13,95,102,103]
VPS35-L774M	Parkinson’s disease (PD)	[13,95,97,102,103]
VPS35-R32S	Parkinson’s disease (PD)	[102–104]
VPS35-I560T	Parkinson’s disease (PD)	[13,102,103,105]
VPS35-H599R	Parkinson’s disease (PD)	[13,102,103,105]
VPS35-M607V	Parkinson’s disease (PD)	[13,102,103,105]
VPS35-G51S	Parkinson’s disease (PD)	[13,97,102,103,106]
VPS35-E787K	Parkinson’s disease (PD)	[13,102,103]
VPS35-L625P	Alzheimer’s disease (AD)	[13,102,103,107,108]
VPS26A-K93E	Parkinson’s disease (PD)	[100,106,109–111]
VPS26A-M112V	Parkinson’s disease (PD)	[106,110,111]
VPS26A-K297X	Parkinson’s disease (PD)	[106,110,111]
VPS29-N72H	Parkinson’s disease (PD)	[110]

Perspective.

• The endosomal retromer complex plays a fundamental role in maintaining critical protein cargoes required for cellular maintenance.

• Structural and biophysical studies on yeast and metazoan retromer suggest the heterotrimer functions as an adaptable scaffold to engage SNX proteins, transmembrane cargoes, and regulatory partners on PtdIns3P-enriched membranes.

• Ongoing structural and mechanistic studies will cement our understanding of retromer coat assembly and regulation, and may provide promising avenues to understand how to develop retromer as an effective therapeutic target for neurodegenerative diseases.

Abbreviations

AD

Alzheimer’s disease

APP

amyloid precursor protein

CI-MPR

cation-independent mannose 6-phosphate receptor

cryo-EM

cryo-electron microscopy

cryo-ET

cryo-electron tomography

GAP

GTPase-activating protein

MNs

motor neurons

PD

Parkinson’s disease

PDZ

psd95/dlg/zo-1

PX

phox homology

SLB

supported lipid bilayers

SNX

sorting nexin

TGN

trans-Golgi network

VPS

vacuolar protein sorting