Endoplasmic Reticulum Membrane Contact Sites, Lipid Transport, and Neurodegeneration

Abstract

The Endoplasmic Reticulum (ER) is an endomembrane system that plays a multiplicity of roles in cell physiology and populates even the most distal cell compartments, including dendritic tips and axon terminals of neurons. Some of its functions are achieved by a cross talk with other intracellular membranous organelles and with the plasma membrane at membrane contacts sites (MCSs). As the ER synthesizes most membrane lipids, lipid exchanges mediated by lipid transfer proteins at MCSs are a particularly important aspect of this cross talk, which synergizes with the cross talk mediated by vesicular transport. Several mutations of genes that encode proteins localized at ER MCSs result in familial neurodegenerative diseases, emphasizing the importance of the normal lipid traffic within cells for a healthy brain. Here, we provide an overview of such diseases, with a specific focus on proteins that directly or indirectly impact lipid transport.

The endoplasmic reticulum (ER) plays a multiplicity of roles in cell physiology. Not only is it the site where secretory and most membrane proteins, as well as lipids, are synthesized, but it also has a variety of other functions ranging from metabolic functions to Ca²⁺ storage and signaling. Some of these functions need to be performed at a local subcellular level and thus the ER network is broadly distributed within the cytoplasm reaching out to even the most remote cell compartments, including dendritic spines and axons of neurons (Fig. 1; Wu et al. 2017). Moreover, the ER also establishes close direct contacts with all other membranous subcellular organelles to allow many forms of communication between them (Wu et al. 2018). Due to the central importance of the ER for cellular life, a large variety of diseases result from the impairment of its functions. Additionally, given the long life of most neuronal cells, it is not surprising that subtle perturbations of its properties result in age-dependent neurodegenerative diseases. Even the unique shape of neurons may contribute to their liability to ER deficiencies, as mechanisms that may counteract the partial impairment of the ER in most cells and in cell bodies of neurons, may not be as effective at distant locations, such as nerve terminals, which may operate as functionally segregated compartments. So, mutations of proteins that control ER tubule formation, elongation and repair, such as reticulons and reticulon-related proteins, spastin and atlastin, are specially damaging for neurons. Proving this point is the abundance of genes encoding ER-shaping proteins among those responsible for hereditary spastic paraplegia (HSP), a condition due to degenerative changes in long axons and thus resulting primarily in defects in lower limbs motility. The role of mutations of ER-shaping proteins in neurodegeneration have been discussed in recent reviews (Blackstone 2018). The present work will focus on another aspect of ER dysfunction that has been implicated in age-dependent neurodegenerative conditions: its cross talk with other organelles at membrane contact sites (MCSs), with a primary emphasis on the lipid exchanges that occur at these sites.

Figure 1.

The neuronal endoplasmic reticulum (ER). The ER extends throughout the entire cytoplasm, including distal cell compartments such as dendritic spines and nerve terminals of neurons, and establishes membrane contact sites with all other membranous organelles.

MEMBRANE CONTACT SITES

Cells are dynamic environments so that membranous organelles have frequent opportunities to contact each other as a result of random encounters. In many cases such encounters, which can be facilitated by active transport within cells, are the premise to membrane fusion. However, the MCSs that will be discussed here are close membrane appositions mediated by protein tethers that do not lead to fusion of the two participating membranes, but mediate communication between the two compartments delimited by such membranes (Wu et al. 2018; Prinz et al. 2020). They can be visualized in electron microscopy (EM) micrographs by extended areas of close apposition of the two membranes, and in live dynamic fluorescence imaging by the “comobility” of the two adjacent organelles. MCSs can be stable or very transient. Moreover, a given MCS may be populated by different tethers at different times as MCS proteins present in the two participating membranes coalesce or disperse, depending on the functional state of the cell.

The first characterized MCSs were the ER–plasma membrane contacts involved in depolarization contraction coupling in muscle (Porter and Palade 1957). Many other MCSs have been subsequently described and molecularly characterized not only between the ER and other organelles but also between other organelles with each other (Prinz et al. 2020). This area of cell biology has advanced very rapidly over the last few years. MCSs have a multiplicity of functions, including the following: (1) They relay the state of an organelle to another membrane or membrane-bound compartment—for example, the STIM1-Orai mediated contacts between the ER and the plasma membrane signal Ca²⁺ depletion in the ER to trigger Ca²⁺ entry from the extracellular medium to replenish ER Ca²⁺ stores (Prakriya and Lewis 2015). (2) They mediate the exchange of ions between organelles with minimal ion leakage into the cytosol—for example, Ca²⁺ fluxes between the ER and mitochondria (Rizzuto et al. 1998). (3) They can affect organelle dynamics, as in the tethering of ER tubules around the mitochondria and endosomes to help mediate their fission (Friedman et al. 2011; Rowland et al. 2014). (4) Importantly, a function of a large number of these contacts is to mediate the transport of lipids between organelles via lipid transfer proteins (LTPs) (Saheki and De Camilli 2016; Wong et al. 2019; Prinz et al. 2020). This transport represents an important complement to trafficking of lipids as part of the membranes of vesicle carriers. It also plays an essential role in the exchange of lipids between the ER and both mitochondria and peroxisomes that are not connected to the endomembrane system of the secretory and endocytic pathways (Vance 1990; Petrungaro and Kornmann 2019; Schrader et al. 2020). The occurrence of LTPs had been known for several decades (Wirtz 1991), but only in recent years it has become clear that lipid transport occurs mostly at MCSs, where bilayer proximity facilitates transport speed and specificity.

Typically, LTPs that function at MCSs are thought to operate by a shuttle mechanism. In this mode of transport, the lipid harboring module is linked by flexible amino acid sequences to the membrane tethering domains of such proteins and shuttle back and forth between the two closely apposed membranes to extract and deliver lipids (Wong et al. 2019). In at least some cases, these proteins function in countertransport reactions and do not mediate net transfer of lipids (Mesmin et al. 2013; Kim et al. 2015; Dong et al. 2016). More recent studies revealed a mechanism of lipid transfer at MCSs not previously known in eukaryotic cells. Such mechanisms, which had been known to occur in the transport of lipids between the inner and outer membrane of Gram-negative bacteria (Bishop 2019; Isom et al. 2020; Ruiz et al. 2021; Douglass et al. 2022), involves proteins that function as bridges between adjacent membranes and contain a hydrophobic groove along which lipids are thought to slide from the cytosolic leaflet of one bilayer to the cytosolic leaflet of the other, thus achieving bulk delivery of lipids. As this process results in bilayer asymmetry, bridge-like proteins are expected to work in concert with lipid scramblases to allow equilibration between leaflets and there is evidence for such a partnership (Leonzino et al. 2021; Reinisch and Prinz 2021).

MCS proteins whose dysfunction results in neurodegeneration include proteins that transfer lipids by either type of mechanisms, underscoring the importance of intracellular membrane lipid homeostasis for the health of brain cells. MCSs not involved in lipid transport have also been implicated in neurodegeneration. For example, close appositions of mitochondria to the ER, so called “mitochondria-associated ER membranes” (MAMs) have been reported to be sites of accumulation of presenilin and γ-secretase whose function is altered in Alzheimer disease (Schon and Area-Gomez 2013). This topic, however, will not be discussed here. Likewise, we will not discuss non-neurodegenerative diseases resulting from abnormal MCS function. We will restrict our overview to some of the best characterized examples of monogenic neurodegenerative conditions that result from mutations in proteins that directly or indirectly affect lipid transport at contacts of the ER with other organelles.

VAP AND AMYOTROPHIC LATERAL SCLEROSIS (ALS)

ALS is a condition characterized by progressive degeneration of motor neurons, leading to muscle paralysis. In 2004, the missense mutation P56S in the protein VAP-B was identified as the cause of an autosomal-dominant familial form of this disease (ALS type 8) (Nishimura et al. 2004). Subsequently, other VAP-B missense mutations have been identified in ALS patients (Chen et al. 2010; Blitterswijk et al. 2012; Kabashi et al. 2013; Sun et al. 2017), but the P56S mutation remains the one most solidly linked to the condition and the best studied. VAP, a small ER protein encoded by two genes (VAP-A and VAP-B), comprises a cytosolic Ig-like MSP (major sperm protein) domain, a coiled-coil region through which it undergoes homo- and heterodimerization (Kim et al. 2010) and a carboxy-terminal transmembrane domain (Fig. 2A). The MSP domain, which is also found in the MOSPD family of human proteins, binds a short peptide called FFAT motif (two phenylalanines [FFs] in an acidic amino acid track [AT]) (Loewen et al. 2003; Loewen and Levine 2005; Kaiser et al. 2005). This binding consensus has been subsequently further refined and it has now been established that MSP domains bind many proteins (the so-called VAPome) via canonical or variant (phospho-FFATs, FFNTs) forms of this motif (Fig. 2B–D; Murphy and Levine 2016; Slee and Levine 2019; Cabukusta et al. 2020; Di Mattia et al. 2020). Such proteins include not only cytosolic proteins, but also proteins tethered to other organelles by noncovalent interactions or transmembrane regions, thus leading to MCS formation.

Figure 2.

The major sperm protein (MSP) domain of VAP mediates its interaction with FFAT motif–containing proteins. (A) Domain organization of human VAP-B indicating the most studied mutation implicated in amyotrophic lateral sclerosis (ALS) (P56S). (B) (Left panel) Surface representation of VAP-A bound to the FFAT motif of ORP1 (Kaiser et al. 2005). (Right panel) Ribbon representation of the predicted structure of the MSP domain of VAP-B showing residue P56. (C) Examples of FFAT- and FFAT-like motif sequences. pFFAT (or phospho-FFAT) motifs are suggested to depend on phosphorylation for VAP binding, and FFNT motifs bind preferentially MOSPD1 and 3 and do not require the acidic amino acid track of conventional FFAT motifs. (D) VAP and MOSPD proteins can regulate the binding of many proteins to the ER.

Relevant to this article, many of the VAP interactors that function at MCSs are LTPs (Fig. 3). At MCSs of the ER with either Golgi elements or endolysosomes, these proteins include OSBP and several OSBP-related proteins (ORPs). OSBP and ORPs bind PI4P on membranes opposite to the ER via a PH domain and transfer lipids through an ORD module (Arora et al. 2022). ORDs harbor in a mutually exclusive fashion PI4P and another lipid (e.g., phosphatidylserine [PtdSer] [Chung et al. 2015; von Filseck et al. 2015] or cholesterol [Im et al. 2005; de Saint-Jean et al. 2011] and function in countertransport reactions [Mesmin et al. 2013; Dong et al. 2016; Kawasaki et al. 2021]). Other VAP-binding LTPs at these sites include FAPP2 and STARD11/CERT, which transfer glucosylceramide and ceramide, respectively, to Golgi membranes (Hanada et al. 2003; Kawano et al. 2006; D'Angelo et al. 2007; Mikitova and Levine 2012) and STARD3 and ORP1L, which transfer cholesterol between the ER and endolysosomes (Rocha et al. 2009; Alpy et al. 2013). Another VAP-binding protein at ER-endolysosome contacts is the ER-anchored protein protrudin, which binds Rab7 on endolysosomes (Saita et al. 2009; Raiborg et al. 2015). Protrudin, in turn, interacts with the LTP PDZD8, another ER integral membrane protein and Rab7 interactor (Guillén-Samander et al. 2019), so that together VAP, protrudin, and PDZD8 form a lipid transport complex at the ER–endolysosome interface (Elbaz-Alon et al. 2020; Shirane et al. 2020; Gao et al. 2021). Finally, VAP and the LTP VPS13C tether the ER to endolysosomal structures (discussed below).

Figure 3.

VAP is an endoplasmic reticulum (ER)-adaptor for many lipid transfer proteins. (A) Overview of VAP-mediated ER contacts with other organelles. Proteins indicated in red are shown in the schematic. (B) Selected examples of interactions involving VAP at membrane contacts between the ER and endolysosomes.

At ER–plasma membrane contacts, VAP binds the phosphatidyl inositol transfer proteins (PITPs) Nir2 and Nir3 to regulate countertransport of phosphatidylinositol and its downstream metabolite phosphatidic acid (from the ER to the plasma membrane and vice versa, respectively) (Chang and Liou 2015; Kim et al. 2015; Yadav et al. 2018). Recently, the bulk lipid transporter VPS13A was also reported to bind VAP at these MCSs (discussed below). At contacts between the ER and mitochondria, VAP binds VPS13A and VPS13D (discussed below) as well as PTPIP51 (a putative LTP anchored to the outer mitochondrial membrane also called RMDN3 [De Vos et al. 2012; Yeo et al. 2021]), while at contacts between the ER and peroxisomes, VAP binds VPS13D (discussed below) and the LTPs ACBD4/ACBD5 in the peroxisomal membrane (Costello et al. 2017; Kors et al. 2022).

Moreover, even VAP interactions with proteins that are not LTPs may impact lipid transport indirectly by creating intermembrane tethers that may facilitate the recruitment of LTPs at these sites. Examples include the interaction of VAP with the potassium channel Kv2.1 (Johnson et al. 2018; Kirmiz et al. 2018) and with Nir1 (Quintanilla et al. 2022) at ER to plasma membrane contacts, and with the retromer component SNX2 (Dong et al. 2016) at ER to endosomes contacts.

Consistent with a key role of VAP in cell physiology, HeLa cells harboring genetic disruption of both VAP genes displayed dramatic phenotypic perturbations with a strong accumulation of PI4P in the Golgi complex and on endosomes, with accompanying major membrane traffic and cytoskeleton defects (Mesmin et al. 2013; Dong et al. 2016). Moreover, the KO of VAP-A in mice is sufficient to result in embryonic lethality (McCune et al. 2017). In contrast, VAP-B KO mice are viable with only mild neurological defects, suggesting that most functions of VAP-B can be compensated by VAP-A, which is generally expressed at much higher levels than VAP-B (Kabashi et al. 2013).

The P56S mutation of VAP-B, which affects a conserved proline in its MSP domain close to the FFAT motif binding site, results in a partial misfolding of the protein and impairs its FFAT motif-dependent interactions (Teuling et al. 2007; Furuita et al. 2010; Kim et al. 2010; Shi et al. 2010). Thus, the dominant nature of this mutation in ALS could be due to toxic effects of the aggregation-prone mutant protein or to haploinsufficiency. Overexpression of VAP-B^P56S results in the formation of ER aggregates both in cultured cells and in transgenic animals (Nishimura et al. 2004; Teuling et al. 2007; Fasana et al. 2010; Papiani et al. 2012; Kuijpers et al. 2013), but no such aggregates were observed in human or mice cells where mutant VAP-B is expressed from the endogenous locus (Mitne-Neto et al. 2011; Larroquette et al. 2015).

At present, the precise mechanisms underlying VAP-B^P56S-dependent disease remain unclear. If pathology was simply explained by the presence of a misfolded protein, mechanisms of disease could be independent of the physiological function of VAP and the selective impairment of motor neurons could be explained by the preferential expression of VAP-B in these neurons (Larroquette et al. 2015). If pathology is due to haploinsufficiency, the multiplicity of VAP functions makes the dissection of the pathogenetic mechanisms challenging. However, given the evidence for a major role of VAP in lipid trafficking within cells, one may expect that lipid dyshomeostasis would likely be important for pathogenesis. For further discussion of these possibilities we refer to the review by Borgese et al. (2021).

In the context of neurodegeneration, a functional interrelationship of VAP with NPC1 and NPC2 should also be considered. Mutations in these two proteins, which are responsible for the egress of cholesterol from lysosomes, cause Niemann–Pick type C disease, a neurodegenerative disease due to abnormal accumulation of cholesterol in lysosomes (Platt 2014). As VAP controls cholesterol traffic between the ER and the lysosome membrane (see above), a cross talk between the action of these genes is plausible and was, in fact, reported (Lim et al. 2019).

Finally, VAP-B itself has been identified as a Parkinson's disease (PD) candidate risk gene by GWAS studies (Kun-Rodrigues et al. 2015). Additionally, several VAP interactors have also been implicated in other neurodegenerative diseases, with the outstanding example of VPS13 family proteins, as discussed below.

VPS13 FAMILY PROTEINS AND NEURODEGENERATION

Vps13, a very large protein, was first identified in the yeast Saccharomyces cerevisiae in screens designed to find proteins that control the sorting of vacuolar carboxypeptidase Y from the Golgi complex to the vacuole, hence the name VPS (vacuolar protein sorting) (Bankaitis et al. 1986; Rothman and Stevens 1986; Brickner and Fuller 1997). Yeast has one VPS13 gene while mammals have four. The first mammalian Vps13 protein, VPS13A, was identified as the gene responsible for chorea acanthocytosis (Rampoldi et al. 2001; Ueno et al. 2001). Subsequently, the other three paralogs (VPS13B, C, and D) were implicated in neurological conditions (Kolehmainen et al. 2003; Velayos-Baeza et al. 2004; Lesage et al. 2016; Gauthier et al. 2018; Schormair et al. 2018; Seong et al. 2018; Ugur et al. 2020). Yeast Vps13 was hypothesized to be an LTP after the discovery that it localized at MCSs and that dominant mutations in this protein suppress defects caused by loss-of-function of ERMES (Lang et al. 2015; Park et al. 2016), a protein complex that mediates lipid transfer between the ER and mitochondria in yeast and other fungi but is absent in metazoa (Kornmann et al. 2009, 2011). It was proposed that suppression reflected a role of VPS13 in a bypass lipid transport route to mitochondria involving the vacuole (Thorsness and Fox 1993; Lang et al. 2015; Park et al. 2016; John Peter et al. 2017, 2022).

Subsequently, biochemical, crystallographic, and cryo-EM studies (De et al. 2017; Kumar et al. 2018; Li et al. 2020), and more recently predictions by AlphaFold (Jumper et al. 2021; Cai et al. 2022; Guillén-Samander 2022), showed that the core of VPS13 is a rod-like structure with a hydrophobic groove running along its entire length, where lipids, primarily glycerolipids identified by mass spectrometry studies (Kumar et al. 2018), are harbored. It was therefore proposed that Vps13 is an LTP that functions as a bridge at MCSs allowing bulk flow of lipids from one membrane (most likely the ER) to another (Li et al. 2020; Dziurdzik and Conibear 2021; Leonzino et al. 2021). The rod starts at its amino terminus with a highly conserved region (chorein motif) that may help in the extraction of lipids from the ER membrane (Kumar et al. 2018; Nakamura et al. 2021), and is flanked at its carboxy-terminal side by domains involved in protein–protein and protein–lipid interactions needed for binding to specific organelles. These domains include an arc-like module of six repeats that are composed of antiparallel β-sheets (called VAB domain [Bean et al. 2018; Adlakha et al. 2022], formerly referred to as a WD40-like domain [Kumar et al. 2018]), a bundle of four helices and a carboxy-terminal PH domain (Fig. 4A,B; Fidler et al. 2016).

Figure 4.

The VPS13 family of bridge-like lipid transfer proteins. (A) Domain organization of the human VPS13 family and the other chorein domain–containing proteins ATG2 and SHIP164. Domains and boundaries were based on AlphaFold structure predictions (Jumper et al. 2021) and HHpred homology searches (Gabler et al. 2020). The chorein and Apt1 domains mark the two ends of the lipid transfer bridge. (B) (Left) Ribbon representation of the predicted structure of human VPS13A, where the various domains are colored as in field A. (Right) Surface representation of the protein where the oxygen and nitrogen atoms of the twisting rod-like β-sheet that forms the core of the bridge (orange in field A) are shown in red and blue, respectively. Note the lack of these colors on the floor of the hydrophobic groove that runs along the entire length of the rod. Other domains are colored as in A. (C) Localization of proteins of the VPS13 superfamily to different membrane contact sites. (D) Cartoons depicting neurodegeneration-relevant localizations of VPS13 paralogs at different contact sites. The localization of VPS13A at ER–mitochondria contacts is not shown.

As mentioned above, genetic studies implicated yeast Vps13 in Golgi to endosome traffic and in a partially redundant function with ERMES at ER–mitochondria contacts. Other reported functions in yeast include growth of the sporulation membrane (Park and Neiman 2012) and ER-phagy (Chen et al. 2020). It is expected that bulk lipid transport may underlie all these different physiological functions with specific functions being dependent upon the specific adaptor engaged on target organelles. While mutations in VPS13B are responsible for neurodevelopmental disorders, mutations in the other three paralogs are responsible for neurodegenerative diseases and will be discussed here.

VPS13A AND CHOREA-ACANTHOCYTOSIS

VPS13A, also named chorein, is the causal gene in chorea-acanthocytosis (ChAc), a rare neurodegenerative Hungtington-like disease characterized by involuntary, irregular, and unpredictable muscle movements (chorea), neuropsychiatric deficits and by the presence of abnormally shaped erythrocytes (acanthocytes or spur cells) (Rampoldi et al. 2001; Ueno et al. 2001). VPS13A was shown to bind VAP in the ER via an FFAT motif located in its amino-terminal region, and to bridge the ER to mitochondria, lipid droplets, and the plasma membrane via its carboxy-terminal region (Kumar et al. 2018; Yeshaw et al. 2019; Guillén-Samander et al. 2022). ChAc is due to autosomal recessive mutations, most of which result in the absence of VPS13A. However, in a few cases, missense mutations in various portions of the protein or carboxy-terminal deletions have also been reported (Dobson-Stone et al. 2002). VPS13A KO HeLa cells do not show obvious mitochondrial abnormalities (Yeshaw et al. 2019), possibly because of a redundancy with the function of VPS13D (see below).

ChAc shares strikingly similar clinical manifestations to another neuroacanthocytosis syndrome known as McLeod syndrome (MLS), an X-linked condition caused by mutations in the gene XK (Ho et al. 1994). It is therefore of interest that VPS13A and XK interact with each other (Urata et al. 2019; Park and Neiman 2020; Guillén-Samander et al. 2022; Park et al. 2022; Ryoden et al. 2022) and mediate ER–plasma membrane contact site formation (Guillén-Samander et al. 2022). Importantly, XK is part of a family of multipass plasma membrane scramblases (Suzuki et al. 2014), with the best characterized member of the family being XKr8, a protein implicated in apoptosis via its property to externalize PtdSer, a lipid normally restricted to the inner leaflet of the plasma membrane (Suzuki et al. 2013; Sakuragi et al. 2021). An attractive scenario is that VPS13A may supply lipids to the cytosolic leaflet of the plasma membrane and that XK may mediate lipid scrambling to equilibrate lipid contents in the two plasma membrane leaflets. Accordingly, both proteins were identified as a complex required for PtdSer exposure and lysis of mice T cells (Ryoden et al. 2022). Additionally, a localization of the Drosophila VPS13A/VPS13C ortholog that could reflect ER–plasma membrane contacts was observed in nurse cells undergoing cell death in the ovary, and deletion of the protein prevented proper corpse clearance (Faber et al. 2020), possibly as a result of a defect in PtdSer exposure. Future studies should explore how a defect in plasma membrane scrambling is related to the ChAc and MLS degeneration of striatal neurons, where VPS13A and XK are highly expressed (Guillén-Samander et al. 2022). Local externalization of PtdSer was shown to be required for the engulfment of synaptic processes by microglia and impairment of this process may result in neurodegeneration (Damisah et al. 2020; Scott-Hewitt et al. 2020). Exploring the validity of this hypothesis will be critical to understand the pathogenesis of neuroacanthocytosis syndromes.

The other two diseases classified as neuroacanthocytosis syndromes are Huntington disease–like 2 (HDL2), caused by mutations in the gene that encodes junctophilin 3 (JPH3) (Holmes et al. 2001; Krause et al. 2015), and pantothenate–kinase associated neurodegeneration (PKAN), caused by mutations in PANK2 (Hayflick 2003; Seo et al. 2009). Interestingly, junctophilin 3 is an ER–plasma membrane tethering protein enriched in the brain (Nishi et al. 2003), which may potentially act synergistically with VPS13A-XK at these MCSs. However, the impact of the JPH3 patient mutations, a triple expansion repeat in an alternatively spliced exon, remains to be further elucidated (Rudnicki et al. 2007; Seixas et al. 2012). PANK2 is an enzyme that catalyzes the rate-limiting step of CoA synthesis inside mitochondria (Kotzbauer et al. 2005). Whether pathogenetic mechanisms due to PANK2 disruption may converge with those due to VPS13 mutations is not known. However, it is of interest that at least three of the genes responsible for neuroacanthocytosis have lipid-related functions, implicating lipid dyshomeostasis in these conditions.

VPS13C AND PARKINSON'S DISEASE

VPS13C mutations that result in complete loss of function are responsible for autosomal-recessive early-onset Parkinson's disease (PD); hence, the gene is also referred to as PARK23 (Lesage et al. 2016; Darvish et al. 2018; Schormair et al. 2018; Gu et al. 2020). Additionally, mutations in VPS13C were shown by GWAS studies to confer risk for PD (Nalls et al. 2019). VPS13C and VPS13A share the highest similarity within the human VPS13 family and likely appeared in chordates from the duplication of a single gene (Velayos-Baeza et al. 2004). VPS13C, similarly to VPS13A, binds the ER via an FFAT-dependent interaction, but bridges the ER to endolysosomes via an interaction of its carboxy-terminal region with Rab7, rather than to mitochondria (Kumar et al. 2018; Hancock-Cerutti et al. 2022). Consistent with this localization, VPS13C KO HeLa cells display perturbations of lysosomes, including an increase in their number accompanied by a lower phosphorylation state of the transcription factor TFEB, and a robust accumulation of di-22:6-BMP (Hancock-Cerutti et al. 2022), a lipid specifically found in the lysosomal lumen (Miranda et al. 2018; Gruenberg 2020). However, how the putative lipid transfer functions of VPS13C results in these lysosomal phenotypes remains unclear. Like VPS13A, VPS13C also tethers the ER to lipid droplets (Kumar et al. 2018).

The reported mitochondria defects in VPS13C KO cells may reflect the many connections that have emerged between lysosomal and mitochondrial homeostasis (Deus et al. 2020). Indeed, VPS13C KO HeLa cells display a basal activation of the cGAS-STING DNA-sensing pathway of cellular immunity that was attributed to defects in both organelles: leakage of mitochondrial DNA into the cytosol leading to cGAS activation and defective lysosomal degradation of activated STING (Hancock-Cerutti et al. 2022). Active STING induces the activation of interferon genes and an inflammatory response (Motwani et al. 2019). Activation of this pathway, which remains to be validated in models other than HeLa cells, is of special relevance to PD as it has also been implicated in mouse models defective in PINK1 and Parkin (Sliter et al. 2018). Such dysregulation of innate immunity would have to involve nonneuronal cells of the brain, as components of the cGAS-STING pathway are primarily expressed in microglia and other supporting cells of the nervous system (Saunders et al. 2018).

While it remains unclear how VPS13C deficiency results in PD, it is of interest that several other PD genes encode proteins that impact endolysosomal function (Vidyadhara et al. 2019; Malpartida et al. 2021). This is a field of very active investigation. One challenge toward a better understanding of how loss of VPS13C function leads to disease is the lack of obvious phenotypes of VPS13C KO mice (Hancock-Cerutti et al. 2022), a problem encountered with other PD genes as well.

VPS13D, ATAXIAS, AND OTHER NEUROLOGICAL CONDITIONS

VPS13D is the only VPS13 protein essential for the survival of mammalian cells and organisms (Blomen et al. 2015; Wang et al. 2015; Seong et al. 2018). Loss of VPS13D is embryonically lethal in mice and results in the arrest of fly development at the larval stage (Seong et al. 2018). Patients with compound heterozygous VPS13D mutations and probably with some residual VPS13D function, develop movement disorders of variable severity, whose onset ranges from birth to adulthood (Gauthier et al. 2018; Seong et al. 2018; Koh et al. 2019; Lee et al. 2020). Most patients develop spinocerebellar ataxia, hence the alias SCAR4 for VPS13D (spinocerebellar ataxia, autosomal recessive 4). Some studies describe VPS13D-associated disease phenotypes as similar to those of Leigh syndrome, which is a childhood neurological disorder primarily associated with mitochondrial dysfunction (Lee et al. 2020). Patients who develop disease at an older age present with a form of hereditary spastic paraplegia (Koh et al. 2019).

VPS13D bridges the ER to mitochondria and peroxisomes (Guillén-Samander et al. 2021). Like VPS13A and C, VPS13D binds the ER via an interaction with VAP, but via an unconventional FFAT motif. Binding to mitochondria and peroxisomes is mediated by the interaction of its carboxy-terminal region with the GTPase Miro (Miro1/Miro2), which resides on the outer mitochondrial membrane, and with a peroxisomal splice variant of Miro1 (Guillén-Samander et al. 2021). Several point mutations implicated in disease have been identified, with the most studied being N3521S, which lies in a highly conserved residue of the VAB domain of VPS13D (i.e., the domain that is important for the binding of yeast Vps13 to multiple adaptor proteins) (Dziurdzik et al. 2020).

Impairment of VPS13D results in major defects of mitochondria, which acquire a spherical shape and accumulate in the perinuclear region (Anding et al. 2018; Seong et al. 2018; Baldwin et al. 2021; Du et al. 2021; Shen et al. 2021b). Studies in flies have also revealed a role for VPS13D in mitochondrial fission and mitophagy (Anding et al. 2018; Insolera et al. 2021; Shen et al. 2021a). Such mitophagy, which requires the UBA domain unique to VPS13D, is dependent on Pink1 and the lipid scramblase VMP1, but independent of Parkin (Insolera et al. 2021; Shen et al. 2021a,b). How the lipid transport function of VPS13D and its role in mitophagy are interconnected remain to be elucidated.

VPS13D also promotes the biogenesis of peroxisomes (Baldwin et al. 2021). A similar function was reported in fungal cells lacking the single Vps13 (Yuan et al. 2022). These observations are consistent with a role of VPS13D in delivering lipids for the growth of peroxisomal membrane, although this phenotype is only observed in a subpopulation of cells, suggesting the occurrence of bypass mechanisms (Baldwin et al. 2021; Yuan et al. 2022).

Collectively, these findings suggest that neurodegeneration resulting from the partial loss of VPS13D function results from dramatic cell perturbations that start with a primary defect of mitochondria and peroxisomes.

OTHER PROTEINS IMPLICATED IN NEURODEGENERATION WITH SIMILARITY TO VPS13

Two other proteins that share structural (a rod with a hydrophobic groove) and biochemical (lipid harboring and transfer properties) similarities with VPS13 family proteins may be linked to neurodegeneration. They are ATG2 and SHIP164, each encoded by two distinct genes (ATG2A and ATG2B; SHIP164/UHRF1BP1L and UHRF1BP1) (Osawa et al. 2019; Valverde et al. 2019; Maeda et al. 2020; Hanna et al. 2022).

ATG2 plays a critical role in the growth and expansion of the isolation membrane during autophagosome formation (Velikkakath et al. 2012), a function that can be attributed to bulk lipid transport from the ER. Accordingly, ATG2 was reported to localize at the interface between the ER and growing autophagosomes (Kotani et al. 2018; Valverde et al. 2019) and to function together with scramblases, ATG9 on the isolation membrane (Guardia et al. 2020; Maeda et al. 2020; Matoba et al. 2020; Orii et al. 2021), and VMP1 and TMEM41B on the ER surface (Ghanbarpour et al. 2021; Mailler et al. 2021). A defect in autophagy results in major perturbation of cell homeostasis and is expected to be particularly damaging in the control of protein and organelle turnover in neuronal processes. Accordingly, a role of impaired autophagy in neurodegeneration has been detailed in several excellent reviews (Vijayan and Verstreken 2017). No mutations in ATG2 have been reported so far in neurodegenerative diseases, but its binding partner WDR45/WIPI4 (ATG18 in yeast), a WD-repeat protein belonging to the WIPI family (Lu et al. 2011; Velikkakath et al. 2012; Chowdhury et al. 2018), was implicated in a neurodegenerative condition termed β-propeller-protein-associated neurodegeneration (BPAN) (Haack et al. 2012; Hayflick et al. 2013; Saitsu et al. 2013). WIPI4 binds PI(3)P on the phagophore membrane helping to localize and possibly orient/regulate ATG2 (Chowdhury et al. 2018; Bueno-Arribas et al. 2021).

The precise function of SHIP164 (UHRF1BP1L) and its paralog UHRF1BP1 remains elusive, but we mention it here as SHIP164 was identified by a GWAS as a Parkinson's disease susceptibility candidate gene (Jansen et al. 2017). SHIP164 is localized on clusters of endocytic vesicles that carry several receptors and its function is required for the proper endocytic traffic of these receptors (Hanna et al. 2022). Thus, it remains unclear whether SHIP164 functions as a MCS protein.

SNX14/SCAR20 and Cerebellar Ataxia

SNX14 is the causal gene for distinctive autosomal-recessive cerebellar ataxia and intellectual disability syndrome, spinocerebellar ataxia type 20. Disease is due to SNX14 loss of function, leading to degeneration of cerebellar Purkinje cells with the presence of enlarged and abnormal autophagosomes/lysosomes (Akizu et al. 2015). Snx14 belongs to the SNX-RGS branch (SNX13, SNX14, SNX19, and SNX25) of the sortin nexin family. These proteins are localized at MCSs and have a similar domain organization: an amino-terminal transmembrane hairpin embedded in the ER and cytosolically exposed domains (PXA, RGS, PX, and PXC), which include binding sites for lipid droplets and/or endolysosomes (Henne et al. 2015; Datta et al. 2019; Saric et al. 2021). Based on structural predictions, the PXA and PXC domains coassemble in a single module with a hydrophobic cavity suggesting a lipid transfer function for this protein family (Hariri and Henne 2022; Paul et al. 2022). Genetic and cell biological studies in different species have strongly implicated SNX14 in autophagic flux and metabolism of lipids, primarily neutral lipids (Akizu et al. 2015; Bryant et al. 2018, 2020). Recently, SNX13 was also implicated in the control of cholesterol export form lysosomes (Lu et al. 2021).

CONCLUDING REMARKS

We have provided here an overview of major monogenic neurodegenerative conditions resulting from mutations in genes that control lipid dynamics at contacts of the ER with other membranous organelles. While MCSs with a putative role in lipid transport are not limited to those involving the ER, the ones involving the ER are particularly important, as the ER is at the core of the synthesis and metabolism of most membrane lipids. The involvement of defects of MCS proteins in neurodegeneration points to the importance of lipid homeostasis, membrane lipid homeostasis in particular, for a healthy brain. Lipid dyshomeostasis may result directly in neurodegeneration by cell-autonomous effects, or indirectly, for example, by triggering neuroinflammation due to loss of organelle integrity and activation of innate immunity. The importance of nonneuronal cells both in starting, and then in sustaining, neurodegeneration is a particularly important avenue of current research that needs to be further developed. Most “neurodegeneration genes” encode housekeeping proteins, yet their mutations generally primarily affect specific neuronal populations. Thus, elucidating the mechanisms underlying the differential sensitivity of neuronal subsets is another critical question.

Neurodegenerative diseases are by definition age dependent, as they are due to a progressive demise of the nervous system after it has developed, and in many cases it has functioned normally for many years. So, clearly, the genetic defects underlying these conditions are well compensated during development and early life. In the case of lipid transport, this may be due to the coexistence of many LTPs with partially overlapping functions, and to the occurrence of robust bypass lipid transport pathways when the function of a specific LTP is impaired. Chronic defects, however, coupled to modifications due to aging, eventually lead to the failure of such compensation.

We are still very far from having a good understanding of the mechanisms of neurodegenerative diseases. This understanding is critical to develop preventive and/or therapeutic strategies. This area of research is rapidly advancing with fast progress in the identification of genes whose mutations underlie these conditions. It will be particularly useful to identify pathways shared by multiple genes as this will make it possible to identify therapeutic targets of broad importance in the field.