The bacterium Chlamydia trachomatis is an intracellular pathogen, causing non-congenital blindness, and is the most common sexually transmitted infection worldwide. The bacterium proliferates inside the cell in a membranous compartment, called an inclusion. The integral inclusion membrane protein IncV has been found to directly interact with VAPA/B via two FFAT motifs upon C. trachomatis infection (₂₈₀DSSSSS EYMDALE TV; ₂₅₆ESSSSS SFHTPPN SD; Figure 2A) (Stanhope et al., 2017). Overexpression of IncV in C. trachomatis-infected cells enhanced the recruitment of VAPA and the ER to the inclusion membrane, suggesting that IncV promotes the formation of inclusion-ER MCS. However, depletion of IncV had only a moderate impact on VAPA enrichment at the inclusion membrane, suggesting that other proteins contribute to the stability of inclusion-ER MCS (e.g. IncD-CERT-VAPA/B complex, see below). The two FFAT motifs of IncV both have an acidic tract consisting of multiple serine residues, suggesting that, like ACBD5, its interaction with VAPA/B could be regulated by phosphorylation of these residues, to mimic the negative charge of the conventional acidic residues (Stanhope et al., 2017). The exact contribution of the IncV-VAP interaction in Chlamydia pathogenesis remains to be determined.

In mammalian cells, ceramide is transported from the ER to the Golgi complex at MCS by transport protein CERT for the synthesis of the membrane lipid sphingomyelin (Table 1). Another C. trachomatis integral inclusion membrane protein, IncD, has been found to interact with host CERT, recruiting CERT and thus its binding partner VAPA/B to the inclusion membrane (Figure 2A) (Derré et al., 2011; Agaisse and Derré, 2014; Kumagai et al., 2018). In this way the IncD-CERT-VAPA/B complex may facilitate ceramide transfer from the ER to the inclusion membrane, where it is converted to sphingomyelin with the use of host and/or bacterial sphingomyelin synthases (Elwell et al., 2011; Tachida et al., 2020). This CERT-dependent sphingomyelin pathway is critical for C. trachomatis replication.

Another example of how C. Trachomatis hijacks components usually present at host ER-organelle MCS is STIM1. This ER-resident Ca²⁺ sensor protein colocalized with VAPB at inclusion-ER MCS (Agaisse and Derré, 2015). However, the plasma membrane Ca²⁺ channel ORAI1, the interaction partner of STIM1 at PM-ER MCS, did not associate with the inclusion membrane. Instead, STIM1 may work with another Ca²⁺ channel: IP3R, an ER protein present at mitochondria-ER MCS. IP3R has been found to bind both STIM1 (Santoso et al., 2011) and the inclusion membrane protein MrcA, presumably forming a Ca²⁺ signalling complex at the inclusion-ER MCS (Nguyen et al., 2018). Both STIM1 and ITPR3 are required for chlamydial release via extrusion of the inclusion. The regulation of local Ca²⁺ levels may influence the myosin motor complex, which promotes the extrusion.

Overall, the interaction of Chlamydia membrane protein IncV with VAPA/B promotes the formation of inclusion-ER MCS within cells. At these sites, C. trachomatis redirects several host proteins for sphingomyelin synthesis (e.g. CERT via IncD; important for bacterial replication) and Ca²⁺ signalling (e.g. STIM1 via MrcA; bacterial extrusion) to assist its pathogenicity.

Noroviruses are non-enveloped RNA viruses and the primary cause of gastroenteritis. The murine and human (GI) norovirus protein NS1/2 has been reported to interact with VAPA/B (Figure 2B) (Ettayebi and Hardy, 2003; McCune et al., 2017). Structural analysis revealed that the murine NS1/2-VAP interaction is mediated by a FFAT-motif mimic located in the N-terminal NS1 domain of NS1/2 (₄₀ESEDEV NYMTPPE QE) (McCune et al., 2017). The FFAT-motif is conserved across murine norovirus strains, although the inherently disordered NS1 domain itself is not well conserved in contrast to the NS2 domain (Baker et al., 2012). Interestingly, NS1/2 has been found to form dimers, a property of many FFAT motif-containing proteins, which could stabilize the interaction with VAP-dimers. It would be interesting to determine whether the human NS1/2^GI-VAP interaction is also mediated via a FFAT motif.

Strikingly, mutagenesis of the NS1/2 FFAT residues critical for VAP binding eliminated virus replication (McCune et al., 2017). Additionally, VAPA depletion in cells showed that VAPA was important in the early stage of norovirus replication. However, it is not clear how the NS1/2-VAP interaction contributes to the viral replication cycle. Localisation of NS1/2 to the ER might contribute to the formation of the membranous viral replication compartment, possibly by bridging the ER and replication membrane via its interaction with VAP and putative transmembrane domain (Baker et al., 2012).

Hepatitis C virus (HCV) is an enveloped RNA virus that predominantly infects liver cells, and can cause liver cirrhosis and cancer. Upon HCV infection, a so-called membranous web, consisting primarily of double-membrane vesicles (DMVs), is formed, that is thought to be the site of viral RNA replication. Three HCV proteins have been reported to associate with VAP. While a direct interaction for the viral NS3/4A protease was not examined (Ramage et al., 2015), structural studies have looked into the binding domains of HCV proteins NS5A and NS5B. The viral RNA-dependent RNA polymerase NS5B interacts via its C-terminal auto-regulatory motif with the MSP domain of VAPA/B (Figure 2C) (Gupta and Song, 2016). This C-terminal motif seems to define whether NS5B is in a folded, auto-inhibitory state, or in a disordered, active state that binds to VAP and initiates RNA synthesis. Additionally, several studies report an interaction between HCV protein NS5A and VAPB, although they attribute the interaction to different domains. One study reveals that NS5A forms a dynamic complex with VAP-MSP by interacting via its disordered C-terminal D3 domain (Gupta et al., 2012). However, other studies report that the coiled-coil domain and transmembrane domain of VAPA/B and other residues of NS5A are essential for NS5A-VAP binding (Tu et al., 1999; Hamamoto et al., 2005; Goonawardane et al., 2017; Wang and Tai, 2019). Phosphorylation of NS5A has been reported to regulate the interaction with VAP (Evans et al., 2004; Goonawardane et al., 2017).

Overexpression and knockdown studies show that the VAP proteins play an important, but yet undefined role in the formation of the HCV replication complex and in RNA replication (Gao et al., 2004; Hamamoto et al., 2005). Although the function of NS5A/B-VAP binding in HCV infection is not fully understood, recent studies are starting to decipher how VAPA/B supports the viral replication. It has been suggested that VAP, NIR2 and OSBP operate in a phosphoinositide cycle between the ER and HCV DMV membrane (Figure 2C). Both VAP and the VAP-interactor NIR2 are required to upregulate phosphatidylinositol-4-phosphate (PI4P) levels during HCV infection (Wang and Tai, 2019), indicating that the phosphatidylinositol (PI) transfer protein NIR2 transfers PI from the ER membrane to the DMV membrane, which is then used to generate PI4P by phosphatidylinositol 4-kinase III α (PI4KA) (Berger et al., 2011). Interestingly, NS5A was shown to associate with and stimulate PI4KA activity. The PI4P is then exchanged for cholesterol from the ER by the VAP-interactor OSBP (Wang et al., 2014). The PI4P enrichment of the DMV membrane can also recruit other PI4P-interacting proteins to the DMV-ER MCS such as the VAP-interactor CERT, which transports ceramide from the ER to the DMV, where it can be converted to sphingomyelin, important for the biosynthesis of DMVs (Gewaid et al., 2020). NIR2, OSBP and CERT normally function at the Golgi-ER MCS (Table 1).

To inhibit the replication of HCV, the cell has mechanisms to disrupt the NS5A/B-VAPA/B binding. The ER-associated virus inhibitory protein Viperin, which binds to both NS5A and the C-terminal region of VAPA (Table 1), promotes the degradation of NS5A, an effect that is enhanced by VAPA (Wang et al., 2012; Ghosh et al., 2020). VAPC, an unstructured VAPB splice variant, acts as an endogenous inhibitor by binding to NS5B, interrupting the interaction of NS5B with VAPA/B (Kukihara et al., 2009; Goyal et al., 2012). The ability of VAPC to negatively regulate HCV replication has been of interest in anti-HCV drug development (Wen et al., 2011). Another potential anti-HCV drug also acts via disrupting the viral-host protein interaction; bicyclol restricts HCV replication by upregulating FFAT-motif containing protein GLTP (Table 1), which interrupted the interaction between VAPA and NS5A (Huang et al., 2019).

Overall, it seems that the VAP proteins anchor the viral RNA replication machinery to the ER membrane via viral NS5A/B interaction, and recruit host VAP interactors (e.g. NIR2, OSBP, CERT) for the synthesis of cholesterol and sphingomyelin, important for HCV replication. Targeting VAP in this way allows pathogens to use a single degenerate and potentially regulatable FFAT motif to interact with a range of useful host proteins.

In addition to HCV, several other viruses hijack cholesterol trafficking within the cell. The Aichi virus (AiV) proteins 2B, 2BC, 2C, 3A, and 3AB are found in a complex with VAPA/B, OSBP and other components of the cholesterol transport machinery at Golgi-ER MCS such as the PI4P-phosphatase SAC1 and ACBD3 (which recruits PI4KB) (Sasaki et al., 2012; Ishikawa-Sasaki et al., 2018). The proteins are recruited to AiV genome replication sites at the replication organelle (RO)-ER MCS, where cholesterol accumulates in the RO membrane. Knockdown of each component resulted in inhibition of AiV RNA replication. Other viruses that utilise the OSBP-cholesterol transport to facilitate RNA synthesis at RO membranes include poliovirus (Arita, 2014), rhinovirus (Roulin et al., 2014) and encephalomyocarditis virus (EMCV) (Dorobantu et al., 2015). Although no virus-VAP (interactor) complexes have been reported for these viruses, poliovirus and EMCV proteins bind to PI4KB and PI4KA respectively, which stimulates PI4P production and leads to recruitment of OSBP.

The Adenovirus-2 (Ad2) adopts a different mechanism to employ the host cholesterol transport pathway. The Ad2 membrane protein RIDα directly interacts and recruits sterol-binding protein ORP1L to maturing early endosomes to form endosome-ER MCS via ORP1L-VAP binding (Cianciola et al., 2017) (Figure 2D). RIDα stabilizes the interaction between ORP1L and VAP, which supports the transport of cholesterol from maturing endosomes to the ER under high cholesterol conditions. The RIDα-ORPL1-VAP interaction induces the conversion of cholesterol into cholesteryl esters, which are stored in lipid droplets. This change in cholesterol trafficking attenuates proinflammatory TLR4 signalling involved in the innate immune response. ORP1L is also hijacked by the intracellular bacterium Coxiella burnetii, which forms a lysosome-like parasitophorous vacuole (PV) in the host cell for its replication (Justis et al., 2017). ORP1L is recruited to the PV by an unknown PV membrane protein, while also associating with ER-VAP. Although the function of ORP1L at PV-ER MCS in C. burnetii pathogenicity is unclear, ORP1L is important for PV expansion.

Herpes simplex virus type-1 (HSV-1) replicates its DNA and assembles its capsids in the host cell nucleus. The virion then crosses the nuclear envelope for further maturation in the cytoplasm. VAPB contributes to this nuclear egress as knockdown led to nuclear virion accumulation, however its exact role in this is still unclear (Saiz-Ros et al., 2019). VAP also plays a role in the replication of another DNA virus, the human papillomavirus 16 (HPV-16). However, instead of a role in nuclear egress, VAP is important in the nuclear entry pathway of HPV-16 (Siddiqa et al., 2018). Virus particles enter the cell via endocytic uptake, disassemble into protein complexes that traffic to the trans-Golgi-network (TGN) and then access the nucleus during mitosis when the nuclear envelope breaks down. VAP is required for the endosome-to-Golgi viral protein delivery, as it is essential for the formation of endosomal tubules induced upon HPV-16 infection. Whether these viruses exploit VAP directly via viral protein interactions or via other mechanism needs to be further elucidated.

The genetic disease cystic fibrosis (CF) is caused by mutations in the CF transmembrane conductance regulator (CFTR) protein. Patients have an increased susceptibility to bacterial infections such as Pseudomonas aeruginosa infection, which aggravates CF. P. aeruginosa exploits VAPB’s mitochondrial tethering function for infection (Rimessi et al., 2020). The bacteria induced increased VAPB and PTPIP51 expression in CF bronchial cells, but not in non-CF cells. The consequent increase in mitochondria-ER contacts caused impairment of autophagy, inducing inflammation and disease progression.

Overall, a variety of different pathogens utilise VAP interaction and modulation to allow them increased access to host resources. This likely reflects the multifunctionality of VAP as a versatile access point (Murphy and Levine, 2016) to the ER membrane and also the diversity of its interaction partners, which have roles in many different cellular functions. Therapeutic strategies which attempt to prevent pathogen access to VAP could perhaps be feasible but would need to be carefully targeted as inhibition of VAP function itself has a dramatic effect on cellular function and is linked to numerous neuronal disorders, as addressed in the following section.

Several studies have reported that overexpression of VAPB P56S induces the formation of insoluble cytosolic aggregates in neuronal and non-neuronal cells (Nishimura et al., 2004b; Kanekura et al., 2006; Teuling et al., 2007), in culture as well as in transgenic mice and Drosophila ALS models (Chai et al., 2008; Ratnaparkhi et al., 2008; Qiu et al., 2013). The aggregation-prone VAPB P56S recruits wild-type VAPB and, to a lesser extent, VAPA to the aggregates, having a dominant-negative effect on normal VAP function (Kanekura et al., 2006; Teuling et al., 2007; Chai et al., 2008; Ratnaparkhi et al., 2008; Suzuki et al., 2009). The VAPB mutant has also been shown to sequesters ER-Golgi recycling protein YIF1A (via its TMD) to the aggregates, depleting the protein from these organelles (Kuijpers et al., 2013b). Nevertheless, VAPB P56S does not seem to induce “classical protein aggregates,” formed of insoluble fibrils, a hallmark of other neurodegenerative disorders like Huntington’s disease (huntingtin), PD (α-synuclein) and ALS (SOD1). For example, VAPB P56S forms aggregates rapidly after expression (<2 h), while the formation of SOD1 aggregates takes hours to days (Matsumoto et al., 2005; Fasana et al., 2010). Additionally, ultrastructural studies showed that overexpression of the mutant VAPB protein caused accumulation of large membranous aggregates, consisting of ribbons of stacked ER cisternae (Teuling et al., 2007; Fasana et al., 2010; Papiani et al., 2012). Live cell photobleaching experiments, using ER membrane-targeted GFP, revealed that VAPB P56S-ER subdomain inclusions are continuous with the rest of the ER (Fasana et al., 2010). However, there is some discrepancy between overexpression studies about the presence of proteins from the secretory pathway in the VAPB aggregates; for instance, ER luminal proteins calreticulin and PDI, and ER membrane protein calnexin associate with mutant VAPB aggregates is some studies, whilst others observed exclusion of these proteins (Kanekura et al., 2006; Teuling et al., 2007; Prosser et al., 2008; Fasana et al., 2010; Kuijpers et al., 2013a). This may be attributed to differences in cell lines, VAPB expression levels, and the exclusion of some (rough) ER membrane proteins from the aggregates (Fasana et al., 2010).

To understand how VAPB P56S induces aggregate formation, we will first discuss how the mutation affects the protein structure. VAPB proline 56 is conserved in VAPA, but mutating this residue does not seem to have such a significant effect, with some studies suggesting that no aggregation was observed whilst others observe minor levels of aberrant aggregation for VAPA P56S, notably in HeLa cells (Teuling et al., 2007; Prosser et al., 2008; Suzuki et al., 2009). VAPA P56S’s resistance to aggregation seems to rely on two other proline residues present in this region, whereas VAPB P56S has only one remaining proline residue (Nakamichi et al., 2011). Substituting one of the prolines in VAPA P56S to the equivalent in VAPB P56S (VAPA P56S/P63A), resulted in the formation of membranous aggregates indistinguishable from those observed with VAPB P56S. The three proline residues of VAPA are conserved in the yeast VAP protein Scs2p, which is also resistant to the ALS8-causing mutation, showing that the proline distribution is an important feature in the pathophysiology of ALS8 (Nakamichi et al., 2011).

P56 is located in the MSP domain of VAPB and is critical for the correct folding of the seven β-strands of the MSP domain (Figure 3A) (Shi et al., 2010). P56 stabilizes the cis-peptide bond within the S-shaped loop that connects strands D1 and D2 (Kaiser et al., 2005; Teuling et al., 2007; Shi et al., 2010). The P56S mutation induces a conformational change within the recombinant MSP domain, resulting in the exposure of hydrophobic patches, which may enhance oligomerization of the mutant VAPB protein under physiological conditions (Figure 3B) (Kim et al., 2010). However, studies with recombinant MSP P56S domains show differences in structural stability and solubility (Kim et al., 2010; Shi et al., 2010). P56S eliminates the native β-sheet structure in water, and the exposed hydrophobic patches seem to drive aggregation of recombinant MSP P56S, making the structure highly insoluble in various buffers (Shi et al., 2010; Qin et al., 2013a). This makes it difficult to understand exactly how the VAPB P56S structure behaves under physiological conditions. Nevertheless, it has been shown that MSP P56S retains its ability to bind to FFAT motif-containing proteins in HeLa cells, but the FFAT-binding of full-length VAPB P56S is perturbed (Kim et al., 2010). The aberrant oligomerization of full-length VAPB P56S may interfere with the binding of FFAT motifs to the MSP domain. In line with this, no FFAT-motif containing proteins were observed in pull-down assays using biotinylation-tagged VAPB P56S (Teuling et al., 2007). However, overexpression of a FFAT motif peptide rescued the aggregation phenotype of the mutant, suggesting protein stabilisation via FFAT motif-binding (Prosser et al., 2008). Additionally, VAPB P56S induces clustering of mitochondria and peroxisomes that colocalise with the VAPB aggregates (De Vos et al., 2012; Hua et al., 2017). The clustering of peroxisomes was dependent on the presence of ACBD5, suggesting that the mutant VAPB can sequester FFAT-motif containing proteins such as peroxisomal ACBD5 and possibly mitochondrial PTPIP51 (De Vos et al., 2012; Hua et al., 2017). However, FFAT-proteins ORP9 and NIR2 were not detectable in the VAPB aggregates (Kuijpers et al., 2013a). In summary, the P56S mutation causes conformational changes in the MSP domain and although this does not affect FFAT-binding to the domain on its own, in the presence of full-length VAPB, exposed hydrophobic patches cause enhanced oligomerization of the protein, which seems to reduce accessibility to the FFAT-binding site.

The disordered MSP P56S domain, but not the wild-type MSP domain, is able to interact with dodecylphosphocholine, a lipid commonly used to resemble membrane lipids, transforming the domain into a highly helical conformation (Qin et al., 2013b). This allows MSP P56S to be inserted into membrane environments (Qin et al., 2013a). Therefore, the interaction of VAPB P56S with lipids from the ER membranes could provide a mechanism for the formation of the membranous aggregates. The presence of membrane structures within the aggregates could also be attributed to VAPB being a tail-anchored protein. VAPB P56S has been shown to be efficiently post-translationally inserted into the ER membrane, after which it rapidly clusters (Fasana et al., 2010). This is further confirmed by a study showing that co-expression of a FFAT-containing peptide with VAPB P56S, partially restored the characteristic reticular ER pattern of VAPB (Prosser et al., 2008), suggesting that FFAT binding can maybe stabilise the mutant MSP structure and that the MSP domain/FFAT-interaction plays an important role in the formation of (membranous) aggregates.

Overall, VAPB P56S seems prone to aggregation due to instability of its MSP structure caused by the mutation. Because of its unaffected TMD, the mutant protein is still targeted to the ER membrane (Fasana et al., 2010), where it forms clusters, recruits wild-type VAPA/B and VAP interactors, and induces the formation of membranous clusters (Teuling et al., 2007; Fasana et al., 2010). Conceivably, newly synthesized mutant VAPB could also aggregate in the cytosol before its insertion into the ER membrane. This may depend on the rate of protein synthesis and levels/capacity of the chaperone machinery, which may differ between different cell types, but could then give rise to two different types of aggregates, cytosolic and membranous, which may explain some discrepancy between studies.

While overexpression studies show that VAPB P56S is aggregation prone, aggregate formation was also induced at physiological conditions at low levels of mutant VAPB, comparable to endogenous wild-type protein, in HeLa cells (Fasana et al., 2010; Papiani et al., 2012). However, more research is required to clarify in what extend these aggregates form in patients; it has been shown that iPSC-derived motor neurons from ALS8 patients have reduced levels of VAPB and no signs of aggregate accumulation (Mitne-Neto et al., 2011; Oliveira et al., 2020), while ALS8 patient-derived muscle biopsy and fibroblasts revealed VAPB aggregates (Tripathi et al., 2021). As HeLa cells also displayed aggregated VAPA P56S (Teuling et al., 2007), which was not observed in other cell types, it seems likely that different cell types show altered VAP aggregate accumulation.

Discrepancy in detection of aggregates in patients might be due to differences in clearance of mutant VAPB. VAPB P56S has been reported to be less stable than the wild-type protein in both cultured cells and transgenic mice (Papiani et al., 2012; Aliaga et al., 2013; Genevini et al., 2014). VAPB P56S was polyubiquitinated shortly after synthesis and degraded by the proteasome in inducibly-expressing HeLa and NSC34 (motoneuronal) cells, with no evident involvement of basal autophagy (although it can be targeted by stimulated autophagy) (Kanekura et al., 2006; Papiani et al., 2012; Genevini et al., 2014). Ubiquitination of VAPB P56S has also been observed in motor neurons and muscle of transgenic mice and flies (Ratnaparkhi et al., 2008; Tsuda et al., 2008; Tudor et al., 2010). The data further indicates that in the HeLa cells and transgenic mice, the mutant protein initially avoids degradation, clusters and is then cleared by the proteasome (Papiani et al., 2012; Kuijpers et al., 2013a). This comprises the involvement of ER membrane chaperone BAP31 and the ATPase chaperone p97/VCP, proteins involved in ER-associated protein degradation (ERAD), likely by extracting mutant VAPB from the ER membrane. However, a study reported that overexpression of both wild-type and mutant VAPB impaired proteasome activity, possibly by inducing ER stress (see below) (Moumen et al., 2011), although this might be attributed to the high levels of expressed VAPB in comparison to the inducible system. Interestingly, it has also been reported that VAPB P56S is resistant to proteolysis by an unidentified protease that releases the MSP domain from wild-type VAPB (Gkogkas et al., 2011).

The P56S mutant VAPB causes ER stress (Aliaga et al., 2013; Larroquette et al., 2015), altered ER domain properties (Fasana et al., 2010; Papiani et al., 2012; Yamanaka et al., 2020) and malfunction of the unfolded protein response (UPR), a physiological reaction to suppress accumulation of misfolded proteins in the ER (Kanekura et al., 2006; Suzuki et al., 2009). In mammalian cells, the three main signalling pathways of UPR are IRE1, ATF6, and PERK–with all three shown to be affected by mutant VAPB. VAPB P56S suppress the IRE1-XBP1 pathway that activates expression of UPR target genes, such as chaperones and ERAD components (Kanekura et al., 2006; Suzuki et al., 2009; Tokutake et al., 2015b). VAPB directly interacts with the ER-localized transcription factor ATF6 which, by acting as an ER stress sensor, regulates the transcription of genes encoding chaperones and other UPR transcription factors (Gkogkas et al., 2008). VAPB P56S was shown to attenuate the ATF6-mediated UPR transcription. On the other hand, VAPB P56S activates UPR via PERK-ATF4 which, by promoting the expression of the pro-apoptotic gene CHOP, initiates the cell apoptotic pathway under prolonged ER stress (Aliaga et al., 2013; Tokutake et al., 2015a). Increased basal ER stress and UPR activation has also been reported in the ALS8 patient-derived fibroblasts (Guber et al., 2018). Overall, if the UPR impacted by VAPB P56S cannot restore proteostasis, it might lead to apoptosis.

Mutant VAPB has been linked with dysfunctional autophagy (Zhao et al., 2018; Tripathi et al., 2021). The P56S mutation reduced VAPB’s interaction with early autophagy proteins ULK1 and FIP200, impairing autophagosome biogenesis (see Table 1) (Zhao et al., 2018). Additionally, VAPB P56S accumulates in autophagosomes and impairs their clearance, showing that VAPB acts at different stages of autophagy (Larroquette et al., 2015; Tripathi et al., 2021). An accumulation and sequestering of autophagic markers p62 and LC3 at VAPB P56S aggregates was also observed in ALS8 patient fibroblasts and muscle biopsies (Tripathi et al., 2021). Impairment of autophagy by mutant VAPB can result in the aggregation of FUS, TDP-43 and Matrin 3 – mutations in which are associated with familial ALS–leading to the formation of stress granules (Tudor et al., 2010; Tripathi et al., 2021). Overexpression of FUS and TDP-43 have both been linked with disruption of the PTPIP51-VAPB association and hence, mitochondria-ER contacts (Stoica et al., 2014, 2016) (Figure 1). Loosening mitochondria-ER contacts via PTPIP51 or VAPB knockdown has been shown to stimulate autophagosome formation by disrupting the Ca²⁺ delivery to mitochondria from ER stores (Gómez-Suaga et al., 2017). VAPB P56S and TDP-43 may also co-operate in the pathogenesis of ALS by activating the mitochondrial apoptotic pathway (Suzuki and Matsuoka, 2011). VAPB is also involved in ER-phagy, a selective form of autophagy for degradation of the ER, via interaction with the soluble ER-phagy receptor CALCOCO1 which, via ATG8 binding, connects the ER and autophagosome membranes (see Table 1) (Nthiga et al., 2020).

Although VAPB is ubiquitously expressed and hence disruption caused by the P56S mutation would affect all cells in the body, ALS8 patients mainly present with (lower) motor neuron dysfunction and neurodegeneration. The large size and complex morphology of motor neurons make the maintenance of protein homeostasis and the distribution of organelles a greater challenge. Hence, motor neurons may be more vulnerable to the overall homeostatic disruption caused by aberrant VAPB. Several studies illustrate how VAPB P56S can affect neuron-specific processes and morphology. For instance, the mutant VAPB disrupts anterograde mitochondrial axonal transport by disrupting Ca²⁺ homeostasis in neurons (Mórotz et al., 2012). Peroxisomal movement in hippocampal neurones has also been shown to resemble that of mitochondria and be altered by levels of the peroxisome-ER tethering protein ACBD5 (Wang et al., 2018). However, unlike for mitochondria, this did not appear to be dependent upon VAPB interaction. A loss of the VAPB orthologue in Drosophila also resulted in abnormal organelle distribution in neuronal axons and dendrites, including mitochondria and the Golgi apparatus, which may have contributed to the altered dendrite morphology (Kamemura et al., 2021), indicating the importance of ER-tethering in organelle distribution. Furthermore, mitochondria-ER contacts, mediated by the PTPIP51-VAPB interaction, are present at synapses and regulate synaptic function (Gómez-Suaga et al., 2019). Loss of PTPIP51 or VAPB reduced synaptic function and altered dendritic morphology. VAPB P56S also sequesters VAP-interactor YIF1A (Table 1), which regulates membrane trafficking into dendrites and dendritic morphology (Kuijpers et al., 2013b). VAPB is also important for neurite extension of motor neurons (Genevini et al., 2014), possibly via its interaction with protrudin (Table 1) (Saita et al., 2009). Additionally, VAPB P56S led to a loss of HCN channel activity, important for neuronal and cardiac pacemaker currents (Silbernagel et al., 2018). These alterations in motor neurons may partly explain the neurodegeneration and muscle-related symptoms observed in ALS8 patients. VAPB P56S may also affect muscle cells more directly; the VAPB mutation disrupted the formation of multinuclear myotubes (muscle fibres) by mouse skeletal muscle cells (Tokutake et al., 2015b) and caused accumulation of ER Ca²⁺ sensor STIM1 at neuromuscular junctions (NMJ) in muscle fibres of ALS8 patients, suggesting altered intracellular Ca²⁺ homeostasis (Goswami et al., 2015). Interestingly, in Drosophila, VAPB regulates the number and size of synaptic boutons at NMJ (Pennetta et al., 2002; Chai et al., 2008). Additionally, VAPB deficient mice showed abnormal skeletal muscle energy metabolism upon fasting (Han et al., 2013). Impaired degradation pathways, accumulation/aggregation of misfolded proteins and disrupted Ca²⁺ homeostasis in motor neurons and muscle fibres may all contribute to ALS8 pathogenesis.

Acknowledging the various roles VAPB plays in many important physiological pathways, it is not surprising that disruption of the protein has a major effect on cellular homeostasis. Nevertheless, it is still under debate whether the P56S mutation in VAPB induces the symptoms of ALS8 patients by a loss of function (lost/reduced protein interactions), a toxic gain of function (aggregate formation, protein sequestering), or a dominant negative effect (wild-type VAP recruitment). VAPB P56S aggregates in the nervous system of transgenic mice did not cause motor neuron dysfunction, suggesting that aggregates are not sufficient to initiate pathogenesis (Tudor et al., 2010; Qiu et al., 2013), although, with a higher fold increase of VAPB P56S protein expression, mice developed abnormal motor behaviour and progressive degeneration of corticospinal motor neurons (Aliaga et al., 2013). A study using both homozygous and heterozygous VAPB P56S knock-in mice showed defects in motor behaviours, with accumulation of cytoplasmic inclusions selectively in motor neurons before onset of the defects, though the homozygous knock-in mice presented with a more severe phenotype, reflecting a dose-dependent effect of the mutant protein (Larroquette et al., 2015). On the other hand, VAPB knockdown was sufficient to lead to motor deficits in zebrafish and mild, late-onset motor deficits were observed in VAPB knockout mice, however, VAPB depletion was unable to induce a complete ALS phenotype (Kabashi et al., 2013). Thus, VAPB P56S abnormalities might be a combination of gained and lost functions, in a dominant and dose-dependent manner.