A role for Vps13 and Chorein-N motif proteins as inter-organelle lipid transfer channels?

Abstract

Membrane contact sites, where two organelles are in close proximity, are critical regulators of cellular membrane homeostasis, with roles in signaling, lipid metabolism, and ion dynamics. A growing catalog of specialized lipid transfer proteins carry out lipid exchange at these sites. Currently characterized eukaryotic lipid transport proteins are shuttles that typically extract a single lipid from the membrane of the donor organelle, solubilize it during transport through the cytosol, and deposit it in the acceptor organelle membrane. Here we highlight the recently identified chorein_N family of lipid transporters, including the Vps13 proteins and the autophagy protein Atg2. These are elongated proteins that, distinct from previously characterized transport proteins, bind tens of lipids at once. They feature an extended channel, most likely lined with hydrophobic residues. We discuss the possibility that they are not shuttles but instead are bridges between membranes, with lipids traversing the cytosol via the hydrophobic channel.

Introduction

In eukaryotic cells, most lipids are synthesized in the endoplasmic reticulum (ER), from which they are distributed to other organelle membranes, in part via vesicular transport. A second major lipid redistribution mechanism, and the only means by which organelles not in the secretory pathway acquire lipids, is via membrane contact sites, where organelles come into close apposition (10-30 nm). Here non-vesicular lipid exchange occurs in a controlled manner, helping to create and maintain distinct organellar lipid compositions. Lipid transfer proteins (LTPs) localizing to these sites, with some well-characterized examples including the extended synaptotagmins, StART-like proteins, and the Osh and Orp proteins, are the primary agents facilitating this exchange [1–7]. Most also act as tethers, directly or indirectly linking the participating membranes together [8–11]. While many questions remain regarding the mechanistic details of their functions, most known LTPs are thought to behave as “shuttles”, interacting with the source membrane to extract lipids, then shielding their hydrophobic moieties while carrying the lipids to the target membrane for delivery [12]. This stands in contrast to a so-called “tunnel” model, in which a lipid transfer protein forms a stable bridge between two membranes across which lipids can travel between the membranes.

A recently discovered class of high-molecular weight LTPs resident at contact sites may function by such a tunnel mechanism. This new family is defined by sequence homology at members’ N-termini in the so-called chorein_N motif, named for the ~120 N-terminal amino acids of human family member Vps13A or “chorein”, though the proteins have only little to no homology outside this region. To date, three members of the family have been identified: Vps13 [13], the autophagy-related protein Atg2 [14], and SHIP164 [15], of which little is known. Recent bioinformatic work has identified several possible bacterial homologues, suggesting the fold predates eukaryotes [16]. Despite recent progress regarding Vps13 and Atg2, much remains to be known about the roles of this family of proteins in the cell and their mechanism of action.

Vps13 is a lipid transfer protein found at multiple membrane contact sites

Vps13 was the first chorein_N protein to be characterized as a lipid transport protein [17]. With a single isoform in yeast and four in humans (A-D; [18]), Vps13 localizes to multiple inter-organellar contact sites. In yeast, Vps13 is found at mitochondria-vacuole contacts (vCLAMPs; [19]) and nuclear-vacuolar junctions (NVJ; [19]), as well as at more specialized sites such as the prospore membrane ([20,21]. In humans, Vps13 populates ER contacts with mitochondria (Vps13A), endosomes (Vps13C), or lipid droplets (Vps13A/C). Mutations affecting human Vps13 isoforms are associated with distinct neurological diseases, including chorea-acanthocytosis (Vps13A), Cohen syndrome (Vps13B), Parkinson’s disease (Vps13C), and spastic ataxia (Vps13D), suggesting paralogs are specialized for specific functions [22–25].

Vps13 localization to organelle contacts is guided by multiple determinants. In the human proteins an N-terminal FFAT motif [26] likely mediates ER contacts via interaction with VAP-A and -B proteins [17], while C-terminal putative WD40 and DH_L-PH domains recruit Vps13 to endosomes and lipid droplets, respectively [17]. Interestingly, several interaction partners of yeast Vps13, including Mdm10-complementing protein (Mcp1) [27,28], the sorting nexin Ypt35 [29], and Spo71 [30,31], share in common a proline-x-proline sequence motif required to target Vps13 to three different organelles (mitochondria, endosomes/vacuoles, and the prospore membrane, respectively) via its WD40 repeats [21], and other putative PxP motifs targeting Vps13 to other organelle membranes may yet remain undiscovered. It has been suggested that competition among these PxP motifs may determine the localization of Vps13 for different processes [21]. No similar motif has yet been identified in higher eukaryotes. Additionally, yeast Vps13 was shown to interact with membranes containing acidic phospholipids in vitro [32,33], but the significance of these interactions in vivo is unclear.

At low resolution Vps13 resembles a bubble wand, complete with a “stalk” and a “loop” [32](Fig. 1A). The “stalk” most likely comprises N-terminal portions of Vps13, which are rod-like at low resolution [17], whereas the putative WD40 domain at the C-terminus might form the “loop”. The crystal structure of a small N-terminal fragment of Vps13 showed that it folds to resemble a utility scoop, with the chorein_N motif capping its back [17] and β-strand sequences immediately C-terminal to this motif forming a cavity (Fig 1A). Strikingly, the concave surface of the scoop is lined entirely by hydrophobic residues, making it suitable for solubilizing lipid fatty acid moieties. Indeed, biochemical experiments showed that both this small fragment and a larger one comprising more of the Vps13 “stem” bind glycerolipids and can transport them between membranes in vitro [17]. Taken together with genetics studies in yeast [19] and the localization of Vps13 proteins to membrane contact sites [17,27,34], where protein-mediated lipid transport frequently takes place, these findings strongly supported a role for VPS13 as a LTP.

Figure 1. Structural and mechanistic features of the chorein_N motif proteins.

A. Left and middle panels: Crystal structure of Vps13 N-terminal domain (PDB ID: 6CBC) in ribbon and surface representations. The ribbon representation is colored from blue at the N-terminus to red at the C-terminus. Surface representation colored by atom (carbons are white, oxygens red, nitrogens blue, and sulfurs yellow) to highlight the hydrophobic interior of the pocket. Right panel: Structure of Vps13 determined by negative stain EM, adapted from [32]. A region proposed to correspond to the crystal structure is colored orange. B. Left and middle panels: Crystal structure of Atg2A N-terminal domain (PDB ID: 6A9J) represented as in A. Right panel: CryoEM reconstruction of Atg2A at 15 Å, shown in surface representation with the region corresponding to the crystal structure colored orange, and in mesh representation, with the interior tunnel/groove highlighted in blue. C. Reconstructed model of the bacterial lipopolysaccharide transporter assembled from crystal structures (PDB IDs: 6MIT, 5IV9, 2R1A), adapted from [55]. The direction of LPS transport along the hydrophobic groove comprising LptC/A/D is indicated by the arrows. D. Model depicting mechanisms by which accessory proteins could facilitate bridge-like lipid transport by chorein_N motif proteins.

The biochemical experiments also showed that larger fragments of Vps13 bound more glycerolipids than the smaller fragment characterized structurally [17], suggesting a more extensive lipid binding cavity that extends beyond the crystal structure along the Vps13 “stem”. Bioinformatics analysis indicates that Vps13 consists primarily of β-strands, like those forming the walls of the scoop. The Vps13 “stem” could then comprise a series of lipid binding modules or, alternatively, a single continuous channel. In light of the low resolution structure of ATG2 described below, which features an elongated cavity [35], as well as our still unpublished data, we favor the idea that the “stem” portion of Vps13 harbors a continuous channel for lipid transport. Lipids would be bound with their fatty acid moieties within the channel and their headgroups exposed to solvent.

Lipid transport by another chorein_N protein Atg2 is required for autophagophore biogenesis.

Atg2 is an important component of the complex machinery required for growth of the phagophore membrane during autophagy [14]. Although Atg2 was initially characterized as a tether [36,37], the crystal structure of its N terminal fragment, including the chorein_N motif, showed the same fold as Vps13, including the hydrophobic pocket, raising the possibility that it is a LTP [38](Fig. 1B). Moreover, ATG2 can support glycerolipid transfer in vitro [35,38,39], and this lipid transfer activity is required for autophagophore growth in vivo [35]. A reconstruction of full-length Atg2A by negative-stain EM [36] revealed an elongated club-like shape, and a low-resolution model by cryo-EM showed an internal tunnel-like cavity, as proposed for Vps13 [35](Fig. 1B). It has thus been suggested that Atg2 mediates direct lipid transfer to the growing phagophore [35,36,38], most likely by a mechanism similar to Vps13.

Localization of Atg2 to MCSs is not yet fully understood, but its contacts involve the ER, mitochondrial, and phagophore membranes [14,40]. Evidence of direct Atg2 interaction with high-curvature membranes in vitro is consistent with its localization to the highly curved rims of the phagophore membrane [37,41], but protein-protein interactions appear to be key for Atg2 targeting. Atg18, a soluble WD40 repeat protein, and its human ortholog, WIPI4/WDR45, are proposed to recruit Atg2 to the PI3P-containing phagophore membrane via a Y/HF motif in the Atg2 C-terminus [37,42], while mitochondrial Tom40 and Tom70 target Atg2A to mitochondria-associated ER membranes [40]. The role of Atg2 interaction with Atg9, a multi-span transmembrane protein which cycles in small vesicles between the trans-Golgi network and the growing phagophore membrane [43–45], is unclear, but is required for growth of the autophagophore, suggesting a role in expansion of the phagophore membrane.

Might Chorein_N proteins transfer bulk lipid to support membrane expansion?

Many of the contact sites in which Vps13 (mitochondria, lipid droplets, phagosome, yeast prospore membrane) and Atg2 (phagophore) participate share in common the requirement for rapid redistribution of lipids to support membrane expansion. Indeed, growth of an autophagosome or prospore or carrying out mitochondrial replication requires the movement of millions of lipids in a relatively brief period [39]. One attractive possibility is that the chorein_N proteins may act as high-flux lipid transfer proteins, carrying bulk membrane lipids to support membrane biogenesis. Thus, they differ from known shuttle-like lipid transporters at MCSs, whose lipid binding modules typically transfer only one lipid at a time [12] to modulate local membrane composition. The chorein_N architecture not only enables much greater lipid-binding capacity, but may provide a continuous tunnel that bypasses the need for “shuttling” between membranes. That both Vps13 and Atg2 are recruited to organelle membranes via interactions with lipids or proteins residing on or in both the source and target membrane offers the possibility that they behave as stable bridges mediating lipid transfer.

By way of analogy, an instructive example is unexpectedly found in a prokaryotic protein complex. The extended hydrophobic groove of Vps13 bears a superficial resemblance to components (LptAC) of the Gram-negative bacterial lipopolysaccharide (LPS) transport system, which feature an extended hydrophobic groove through which LPS is transferred unidirectionally through the aqueous environment separating the inner and outer cell membranes (Fig. 1C). At each end of the hydrophobic tunnel, additional complex components are tasked with loading LPS into the tunnel (ABC transporter LptB₂FG) and then mediating insertion into the outer membrane and flipping to the outer leaflet (LptDE) [46–53]. While the hydrophobic groove observed in the Vps13 and Atg2 structures are wider than that of LptAC, suggesting some differences in mechanism, the idea that the proteins that tether Vps13 and Atg2 to different organelle membranes might have more complex biochemical roles in lipid transfer remains an open possibility.

The notion that chorein_N proteins act as lipid bridges to function in membrane expansion implies directional lipid transfer, raising the question of how directionality is achieved. We propose that, like the LPS transport system, chorein_N proteins cooperate with integral membrane proteins resident in the donor and acceptor organelles. In the donor organelle, we can imagine an interaction with lipid biosynthetic machinery, which could couple steps in lipid synthesis with the loading of lipids into the transport tunnel (Fig. 1D). Although neither Vps13 or Atg2 seem to preferentially bind a particular glycerolipid species [17,35], their partner protein(s) in the donor membrane(s) might select for certain lipids. Or alternatively, even if there is not direct coupling, the chorein_N proteins could be targeted to sites of active lipid synthesis [54] so that lateral pressure in the cytosolic leaflet of the membrane could drive lipid loading onto the chorein_N bridge. The receiving membrane could contain a scramblase or a flippase protein, which may or may not interact directly with the chorein_N protein, to equilibrate incoming lipids between leaflets to allow for membrane expansion. The presence of integral membrane proteins may also perturb lipid packing in the membranes, thus facilitating lipid extraction or insertion, the rate determining steps in protein-mediated lipid transport. While the last few years have yielded important insights regarding the function of Vps13 and Atg2 in lipid transport, there clearly is still much to be learned about their mechanism of action and we eagerly anticipate new findings in this field.