© 2019. Published by The Company of Biologists Ltd | Journal of Cell Science (2019) 132, jcs235176. doi:10.1242/jcs.235176

CORRECTION Correction: VAPB depletion alters neuritogenesis and phosphoinositide balance in motoneuron-like cells: relevance to VAPB-linked amyotrophic lateral sclerosis (doi:10.1242/ jcs.220061) Paola Genevini, Maria Nicol Colombo, Rossella Venditti, Stefania Marcuzzo, Sara Francesca Colombo, Pia Bernasconi, Maria Antonietta De Matteis, Nica Borgese and Francesca Navone

There were errors in J. Cell Sci. (2019) 132, jcs220061 (doi:10.1242/jcs.220061).

In the graph of Fig. 6A, the data points of the control (C16) cells inadvertently replaced those of the third experiment relative to the silenced, #1, clone. This error resulted in an apparent bimodal distribution of the data in the silenced clone. The corrected graph is shown below. The error was not introduced into the ANOVA analysis carried out on the means of the three experiments, so the significance values, figure legend and conclusions from the experiments remain unaltered.

The online full-text and PDF versions of the paper have been updated.

Fig. 6A (corrected panel). Increase in peripheral staining of PI4P-positive structures revealed by the PI4P-specific probe GFP-P4CSidC. (A) VAPB- depleted and vector-only clones, transfected with the GFP-P4CSidC probe, were incubated with Lysotracker before live imaging. Single confocal planes are shown. The percentages of Lysotracker-containing structures positive for the probe are shown on the right (median values±interquartile range; datafromn=150 cells pooled from three separate experiments; symbols indicate values for single cells). Statistical significance was tested by one-way ANOVA followed by Bonferroni’s post test. ***P<0.001. Journal of Cell Science

1 CORRECTION Journal of Cell Science (2019) 132, jcs235176. doi:10.1242/jcs.235176

Fig. 6A (orginal panel). Increase in peripheral staining of PI4P-positive structures revealed by the PI4P-specific probe GFP-P4CSidC. (A) VAPB-depleted and vector-only clones, transfected with the GFP-P4CSidC probe, were incubated with Lysotracker before live imaging. Single confocal planes are shown. The percentages of Lysotracker-containing structures positive for the probe are shown on the right (median values±interquartile range; data from n=150 cells pooled from three separate experiments; symbols indicate values for single cells). Statistical significance was tested by one-way ANOVA followed by Bonferroni’s post test. ***P<0.001.

The authors apologise to readers for this error. Journal of Cell Science

2 © 2019. Published by The Company of Biologists Ltd | Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

RESEARCH ARTICLE VAPB depletion alters neuritogenesis and phosphoinositide balance in motoneuron-like cells: relevance to VAPB-linked amyotrophic lateral sclerosis Paola Genevini1,*,‡, Maria Nicol Colombo1,‡, Rossella Venditti2,‡, Stefania Marcuzzo3, Sara Francesca Colombo1, Pia Bernasconi3, Maria Antonietta De Matteis2,4, Nica Borgese1,§ and Francesca Navone1,§

ABSTRACT sALS is less understood than that of fALS; however, several VAPB and VAPA are ubiquitously expressed endoplasmic reticulum observations have suggested that the two forms of the disease share membrane that play key roles in lipid exchange at membrane at least some pathogenic pathways. For instance, intracellular contact sites. A mutant, aggregation-prone, form of VAPB (P56S) is inclusions of the transactive response DNA-binding ∼ linked to a dominantly inherited form of amyotrophic lateral sclerosis; (TARDBP), of which the corresponding is mutated in 5% of however, it has been unclear whether its pathogenicity is due fALS cases, are observed in motoneurons of both fALS and sALS to toxic gain of function, to negative dominance, or simply to subjects (Neumann et al., 2006). Another example is given by the insufficient levels of the wild-type protein produced from a single GGGGCC-hexanucleotide repeat expansion in 9 open – allele (haploinsufficiency). To investigate whether reduced levels of reading frame 72 (C9Orf72), which are responsible for up to 40 45% – functional VAPB, independently from the presence of the mutant of fALS cases, but are also present in 5 7% of sALS patients (van form, affect the physiology of mammalian motoneuron-like cells, we Blitterswijk et al., 2012). As a final example, the protein investigated generated NSC34 clones, from which VAPB was partially or nearly in the present study, vesicle-associated membrane protein (VAMP)- completely depleted. VAPA levels, determined to be over fourfold associated protein B (VAPB), which is linked to a late-onset, slowly higher than those of VAPB in untransfected cells, were unaffected. progressing form of fALS [known as ALS8 (Nishimura et al., 2004)], Nonetheless, cells with even partially depleted VAPB showed an is also reported to be partially depleted in the motoneurons of sALS increase in Golgi- and acidic vesicle-localized phosphatidylinositol-4- patients and in those of a murine ALS model (Anagnostou et al., 2010; phosphate (PI4P) and reduced neurite extension when induced Kimetal.,2016a;Teulingetal.,2007). to differentiate. Conversely, the PI4 kinase inhibitors PIK93 and VAPB, and its homologue VAPA, are conserved and ubiquitously IN-10 increased neurite elongation. Thus, for long-term survival, expressed adaptor proteins of the endoplasmic reticulum (ER). motoneurons might require the full dose of functional VAPB, which The two proteins have high (63%) sequence identity and similar may have unique function(s) that VAPA cannot perform. domain organization, comprising a cytosolic N-terminal domain, homologous to the nematode major sperm protein (MSP), a central KEY WORDS: Endolysosomes, Neuritogenesis, NSC34 cells, coiled-coil domain and a C-terminal transmembrane domain bearing Neurodegenerative diseases, Phosphatidylinositol-4-phosphate a dimerization motif, which allow VAP proteins to undergo homo- and heterodimerization (Lev et al., 2008; Murphy and Levine, 2016). INTRODUCTION A large number of cytosolic proteins, many of which play a major Amyotrophic lateral sclerosis (ALS) is a complex and fatal adult-onset role in lipid dynamics, are recruited to the surface of the ER via disease, defined by the degeneration of both lower and upper motor interaction between the FFAT (two phenylalanines in an acidic tract) neurons. In some cases, ALS is inherited [familial ALS (fALS)], but motif contained within them and a binding pocket in the MSP domain more often it arises in individuals without a family history of the of the VAPs (Kaiser et al., 2005; Loewen and Levine, 2005; Murphy disease [sporadic ALS (sALS)]. Over 20 have been causally and Levine, 2016). The large number and diversity of VAP binding linked to fALS, many of which are involved in common cellular partners underlies the multiple functions of the two proteins in pathways whose function is thought to be perturbed in affected motor phenomena such as interorganellar lipid exchange (Holthuis and neurons (Nguyen et al., 2018; Taylor et al., 2016). The aetiology of Menon, 2014; Kim et al., 2016b; Yamaji and Hanada, 2015), membrane traffic (Kuijpers et al., 2013b), Ca2+ homeostasis (De Vos

1Consiglio Nazionale delle Ricerche Neuroscience Institute and BIOMETRA et al., 2012; Gomez-Suaga et al., 2017), cytoskeleton organization Department, Universitàdegli Studi di Milano, Milan 20129, Italy. 2Telethon Institute (Amarilio et al., 2005; Dong et al., 2016), autophagy (Gomez-Suaga of Genetics and Medicine, Pozzuoli 80078, Italy. 3Neurology IV – Neuroimmunology et al., 2017; Zhao et al., 2018), mitochondrial function (De Vos et al., and Neuromuscular Diseases Unit, Fondazione Istituto Neurologico ‘Carlo Besta’, Milan 20133, Italy. 4Department of Molecular Medicine and Medical Biotechnology, 2012; Han et al., 2012) and neurite extension (Raiborg et al., 2015; University of Naples Federico II, Naples 80133, Italy. Saita et al., 2009). Importantly, VAP proteins are involved in the *Present address: Diagenode s.a., Liege,̀ Belgium. genesis of membrane contact sites between the ER and other ‡These authors contributed equally to this work organelles. Within these structures, the VAPs, by providing §Authors for correspondence ([email protected]; [email protected]) anchoring sites for lipid exchange proteins, participate in the non- vesicular transport of sterols, sphingolipids and phospholipids (De S.F.C., 0000-0001-9714-3961; P.B., 0000-0003-0869-2104; N.B., 0000-0002- 7537-3974; F.N., 0000-0002-3616-829X Matteis and Rega, 2015; Lahiri et al., 2015; Phillips and Voeltz, 2016; Kumar et al., 2018). In particular, they are important players in

Received 7 May 2018; Accepted 1 February 2019 the regulation and intracellular distribution of phosphoinositides, a Journal of Cell Science

1 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061 class of lipid molecules that play key roles in the establishment of motoneuron-enriched spinal cord cells with neuroblastoma cells membrane identity and in membrane trafficking (Di Paolo and De (Cashman et al., 1992). These cells express many of the Camilli, 2006). To be noted, the sets of binding partners of VAPA and morphological and physiological properties of motoneurons VAPB overlap to a large extent (Huttlin et al., 2015; Murphy and (Cashman et al., 1992; Maier et al., 2013), and have often been Levine, 2016), and specific roles that functionally distinguish the two used for in vitro studies on ALS. VAP isoforms have not been identified yet. We induced these cells to acquire a motoneuron-like phenotype The VAPB-linked form of fALS (ALS8) is due to a dominant by applying the differentiation protocol developed by Maier et al. missense mutation that results in the substitution of a conserved (2013), consisting of serum deprivation combined with retinoic acid proline with serine at position 56, in the MSP domain (P56S-VAPB) (RA) treatment. In agreement with published work, we found (Nishimura et al., 2004). As a consequence, the three-dimensional that this treatment induced the cells to generate long, generally structure of VAPB is disrupted, and the mutant protein is prone to unbranched, processes (Fig. 1A, top row). As shown by the aggregation. Accordingly, P56S-VAPB, when transiently expressed quantitative analysis in Fig. 1B, the percentage of cells with long in cultured cells, forms intracellular inclusions that dramatically (100–200 μm) and very long (>200 μm) neurites increased over alter the structure of the ER (Teuling et al., 2007; Tsuda et al., 2008; time, with a concomitant decrease in those with short ones (0– Fasana et al., 2010; Papiani et al., 2012). 50 μm). The synaptic vesicle protein synaptobrevin 2 (also known Because of the tendency of the mutant protein to aggregate, ALS8 as VAMP2) was seen to concentrate in growth cones at the distal has usually been considered a protein misfolding disorder (Murphy extremities of the long neurites or in varicosities along their length and Levine, 2016; Navone et al., 2015), and, in addition to a (Fig. 1A, right column), indicating an accumulation of maturing or possible toxic gain of function of the inclusions, a dominant- mature synaptic vesicles at the physiologically expected locations. negative mechanism has been thought to cause the disease in the We also analysed, by immunoblotting, several neuronal markers, heterozygote. Indeed, wild-type VAPB, as well as VAPA and other including GAP43, a protein expressed at high levels during neurite functionally important interacting proteins, are sequestered in the elongation, MAP2, a dendrite marker, and choline acetyl transferase inclusions generated by overexpressed P56S-VAPB (Darbyson and (ChAT), the key enzyme for the production of acetylcholine, the Ngsee, 2016; Ratnaparkhi et al., 2008; Suzuki et al., 2009; Teuling neurotransmitter of secondary motoneurons. As shown in Fig. 1C, et al., 2007). Thus, it is thought that, in ALS8 motoneurons, the and in agreement with previous reports (Maier et al., 2013), the product of the wild-type VAPB allele co-aggregates with the mutant expression levels of all three of these proteins increased in cells protein and is therefore unable to function. grown in differentiating medium. Similar results were obtained An alternative hypothesis on ALS8 pathogenesis posits that the with two proteins specifically involved in the process of neurite mutant protein itself is harmless to mammalian motoneurons, and that extension, the small GTPase Rab7a and the ER protein protrudin the dominant inheritance is due to simple haploinsufficiency (Raiborg et al., 2015; Saita et al., 2009; Shirane and Nakayama, (Genevini et al., 2014; Kabashi et al., 2013; Navone et al., 2015). 2006) (Fig. 1C). This hypothesis is based on the observation that P56S-VAPB Because of our interest in the VAPs, we also analysed the levels of inclusions are rapidly cleared from cells (Papiani et al., 2012), and both VAPA and VAPB upon differentiation. Interestingly, VAPB that expression of P56S-VAPB at moderate levels does not interfere increased during differentiation in a statistically highly significant with proteostasis or intracellular transport along the secretory pathway manner (Fig. 1D), in agreement with a previous observation on (Genevini et al., 2014). In agreement with these observations, P56S- iPSCs differentiated into motoneurons (Mitne-Neto et al., 2011). VAPB inclusions do not accumulate in induced pluripotent stem cell VAPA also increased, although to a lesser extent (Fig. 1E). (iPSC)-derived motoneuronal cells of ALS8 patients (Mitne-Neto et al., 2011), and a lack of pathogenicity of the mutant protein is Establishment and characterization of VAPB-downregulated suggested by studies on transgenic mice (Aliaga et al., 2013; Kuijpers NSC34 cell lines et al., 2013a; Qiu et al., 2013; Tudor et al., 2010). Thus, it seems Having characterized our NSC34 cells, we proceeded to generate possible that the P56S mutation is – in itself – not harmful, and that its stable VAPB-silenced lines. To this end, undifferentiated cells were dominant inheritance is due only to the insufficient levels of the wild- transfected with a GFP-encoding plasmid carrying the sequence for a type protein produced from a single allele (haploinsufficiency). VAPB-targeted short hairpin RNA (shRNA). Control clones were Here, to explore the haploinsufficiency hypothesis, we have generated by transfection with the empty vector. After selection by analysed, in a motoneuronal cell line, the consequences of reduced antibiotic resistance, we obtained two VAPB-downregulated clones levels of wild-type VAPB in the absence of expression of the P56S that had ∼10% (clone #1) and ∼40% (clone #6) residual VAPB mutant. Although these cells express VAPA in over a fourfold excess expression compared to untransfected cells or controls transfected compared to VAPB, we find that even partial VAPB depletion, while with the empty vector (clones C11 and C16; Fig. 2A,B). A similar having only a minor effect on the size of the total VAP pool, alters trend was also seen in the transcript levels, with clone #6 displaying phosphoinositide balance and significantly retards neurite extension in levels intermediate between clone #1 and the control clones (Fig. 2C). differentiating cells. These results are consistent with the hypothesis Downregulation of VAPB in clones #1 and #6 was stable over time; that ALS8 is caused by insufficient VAPB expression, and are relevant reduced levels of VAPB could be observed after several weeks in also to the more common sporadic ALS forms, in which motoneuronal culture (data not shown). Importantly, VAPA levels were not affected VAP is reduced. Our results also suggest that, in motoneurons, VAPB by VAPB downregulation (Fig. 2B). Thus, neither a compensatory has unique function(s) that cannot be performed by VAPA. increase in VAPA, nor a decrease caused by disruption of VAPA- VAPB dimers could be observed in the silenced cells. RESULTS NSC34 cells extend long neurites and express neuronal Quantification of VAPA and VAPB absolute amounts markers upon differentiation in NSC34 cells To investigate the role of VAPB in motoneuronal physiology, Before further characterizing the VAPB-downregulated clones, we used the murine cell line NSC34, generated by fusion of we were interested in determining the relative amounts of the two Journal of Cell Science

2 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 1. Analysis of undifferentiated and differentiated NSC34 cells. (A) Top row shows synaptobrevin 2 immunofluorescence (red) superimposed on 4′,6- diamidino-2-phenylindole (DAPI) staining (blue) and on the differential interference contrast (DIC) image. Bottom row shows higher-magnification wide-field immunofluorescence images of synaptobrevin 2 staining. In the lower-right image, the cells are outlined in white. CB, cell body. Arrowheads indicate synaptobrevin 2 accumulation at neurite terminals or varicosities. Scale bars: 50 μm (top); 20 μm (bottom). Insets, ×2. (B) Classification of cells during differentiation according to the length of the longest neurite (means+s.e.m. of three biological replicates). (C) Immunoblot analysis of undifferentiated (UD) and 6- day-differentiated (Diff) NSC34 cells. Actin, tubulin or calnexin (CNX) were used as loading controls. The means±s.e.m. of each analysed protein in the differentiated cells, normalized to the values before differentiation, are indicated below the lanes (n=4, 3, 2, 3 and 2 biological replicates for GAP43, ChAT, MAP2, Rab7a and protrudin, respectively). (D,E) Analysis of VAPB and VAPA in 5-day-differentiated versus UD cells. The amounts of loaded lysate protein are indicated below the lanes. The vertical line separates lanes deriving from the same blot exposure. The graphs show means+s.d. from eight and six biological replicates for VAPB and VAPA, respectively. Significance was assessed by paired Student’s t-test. The numbers on the right of the blots in panels C–E show the positions of the indicated size markers (kDa).

VAP isoforms in untransfected cells. We reasoned that, if the weight sequence, we used two different antibodies that target the of the contribution of VAPB to the VAPA+VAPB pool were high, N-terminal region, which is identical in mouse and human its depletion might have effects related to a reduction in the entire (Kabashi et al., 2013; Kanekura et al., 2006; see Table S1 for VAP pool; on the other hand, if it were a minority component of the details). We also considered possible interference of the lysate VAP pool, any effect caused by its depletion might be attributed to proteins with the detection of the target band, as these are not loss of a specific function not carried out by the VAPA isoform. present in the recombinant protein sample. To address this issue, we To determine the absolute amounts of the two VAP proteins in constructed a VAPB standard curve using the recombinant protein NSC34 lysates, we compared immunoblot band intensities with mixed with the lysate of clone #1, which has only a minor residual those of known amounts of recombinant VAP proteins used as VAPB expression. The estimated concentration of VAPB, based on standards (Fig. 3A). After constructing calibration curves with the the standard curve constructed with the recombinant protein alone recombinant proteins (Fig. 3B,C), we were able to quantify the or after adding clone #1 lysate, was similar with both antibodies amount of each VAP in the lysate (Fig. 3D). In the case of VAPB, (Fig. 3D), indicating that lysate proteins do not interfere with the since our standard recombinant protein was based on the human detection of the target polypeptide. Journal of Cell Science

3 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

(Fig. S1) increased in the control clones after differentiation, and treatment with RA also upregulated VAPB in the partially silenced clone #6. Thus, the silenced clones appear to be able to respond to treatment with RA. In agreement with the biochemical data, we found that the silenced clones were able to extend neurites when grown in differentiating medium (Fig. 4B). However, a quantitative analysis revealed that both clones had a significantly reduced rate of neurite extension compared to untransfected cells and to cells transfected with the empty vector (Fig. 4C; see Figs S2–S4 for examples of analysed fields). A more detailed analysis was obtained by classifying cells in four distinct populations according to the length of their longest neurite at day 6 of differentiation (Fig. 4D). The two silenced clones showed fewer cells in the long (100–200 μm) and very long (>200 μm) classes, and more in the short (0–50 μm) ones, than control cells. Although there was quite some biological variability in the rate and extent of neurite extension of the cell lines and in the degree of delay of the #1 and #6 silenced versus the control clones (e.g. compare Fig. 4C and D with Fig. 8A), we consistently observed that both silenced clones had a reduced rate of neuritogenesis compared to controls grown in parallel, indicating that even mild VAPB downregulation slows neurite growth. To assess whether this phenomenon is due specifically to VAPB depletion, we performed rescue experiments using either a plasmid-encoding human VAPB (which is resistant to shRNA against mouse VAPB) or VAPA, both carrying a myc tag at the N-terminus. Transfection with the VAPB-encoding plasmid caused an increase in the percentage of silenced cells with Fig. 2. Generation of stable VAPB-downregulated NSC34 cell lines. μ (A) Fluorescence staining of GFP expressed by UD cells stably transfected neurites longer than 100 m, in comparison with cells transfected with VAPB-targeting pSUPER.neo.GFP (clones #1 and #6) or with the empty with the empty vector, demonstrating that the delay in vector (clones C11 and C16). Nuclei were counterstained with DAPI. Images neuritogenesis in the silenced cells was not due to an off-target were acquired with identical parameters. Scale bar: 20 μm. (B) VAPA and effect (Fig. 4E). Instead, transfection of VAPB did not enhance VAPB protein expression in NSC34 clones and untransfected (Wt) cells, neurite extension in a control clone, indicating that endogenous assayed by immunoblotting. The line separates lanes from the same blot VAPB is present at a concentration sufficient for its maximal acquired together. The graph shows average values of VAPB protein in the NSC34 clones compared to those in untransfected cells from three biological stimulatory effect on neuritogenesis (Fig. 4E). Interestingly, replicates. By one-way paired ANOVA, clone #1 was different from clone #6 myc-VAPA was less effective than myc-VAPB at restoring (P<0.01) and from C11, C16 and Wt cells (P<0.001), clone #6 was different neuritogenesis in the silenced clone (Fig. 4F). from C11, C16, and Wt cells (P<0.01), but there was no difference between the control clones and the Wt cells. (C) qPCR analysis of VAPB transcript in control Alterations in PI4P in the Golgi complex and in LAMP1- and silenced clones. VAPB transcript levels were normalized to 18S RNA. The positive vesicles of VAPB-downregulated cells symbols indicate results of two separate determinations, each carried out in The VAPs play key roles in establishing/maintaining ER– duplicate or triplicate. The horizontal lines indicate the mean of the two – determinations. endolysosome and ER Golgi contact sites (Dong et al., 2016; Venditti et al., 2019), and in the trafficking and metabolism of the phosphoinositide phosphatidylinositol-4-phosphate (PI4P) at these The results of the quantification experiments, summarized in sites. In particular, they recruit members of the oxysterol binding Fig. 3D and E, revealed that, both in undifferentiated and differentiated protein (OSBP)/OSBP-related protein (ORP) family to the ER NSC34 cells, VAPA is the predominant VAP isoform, in an over membrane, from which they mediate the transport of cholesterol to fourfold excess over VAPB. the Golgi and to endosomes, and back transfer of PI4P from these compartments to the ER (Dong et al., 2016; Mesmin et al., 2013). VAPB downregulation delays neurite extension during On the ER, PI4P is hydrolysed to phosphatidylinositol (PI) by the NSC34 differentiation Sac1 (encoded by Sacm1l) phosphatase, which also requires the As no obvious growth defects or gross morphological changes VAPs for its activity (Forrest et al., 2013; Stefan et al., 2011), and were observed in the two selected silenced clones under normal the function of which is required to drive the PI4P/cholesterol culture conditions, we investigated whether VAPB exchange cycle (Zewe et al., 2018). downregulation could alter the differentiation process. First, the Presumably because of the role of the VAPs in the OSBP/ORP clones were cultured in differentiating medium and analysed by cycle, their combined depletion, or depletion of the unique orthologue immunoblotting (Fig. 4A). As observed for untransfected cells of Drosophila or Caenorhabditis elegans, led to an increase in the (Fig. 1), at 6 days of differentiation we observed increased levels of intracellular PI4P and to alterations in its distribution (Dong expression of both ChAT and GAP43, although the basal et al., 2016; Forrest et al., 2013; Venditti et al., 2019; Zhang et al., expression of GAP43 in undifferentiated cells varied between 2017). To investigate whether downregulation of VAPB in the different clones. Expression of both VAPB (Fig. 4A) and VAPA presence of normal levels of VAPA had an impact on the PI4P Journal of Cell Science

4 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 3. Determination of VAPA/VAPB ratio in NSC34 cells. (A) Coomassie-stained gel of recombinant proteins used for the construction of standard curves. In addition to the full-length polypeptides, indicated by the arrows, degradation products are present. The percentage of the total protein in the preparation contributed by the full-length product was determined with ImageJ software. (B) Determination of VAPB concentration in differentiated and UD NSC34 cells by immunoblotting with antibody #1 (Kabashi et al., 2013) or #2 (Kanekura et al., 2006). In the illustrated examples, lanes were loaded with the indicated amounts of recombinant standard or of cell lysates from two separate cultures of 5-day-differentiated (Diff) or UD cells. (C) Determination of VAPA concentration in NSC34 cells, as in B. (D) Summary of the quantification of VAPA and VAPB concentration. To exclude interference of lysate proteins in the determination of NSC34 VAPB, two determinations for each antibody were carried out after dilution of the recombinant protein in 15 μg protein VAPB-downregulated lysate (shRNA-VAPB clone #1); in this case, the values for the recombinant protein were corrected for the residual amount of VAPB in the silenced lysate. Means±s.e.m. from the indicated number of biological replicates are shown. (E) Cartoon illustrating the quantitative relationship between VAPA and VAPB based on the data in D. distribution in NSC34 cells, we used an anti-PI4P antibody with an illumination microscopy (SIM), revealed, in addition to the many immunofluorescence protocol designed to detect the phosphoinositide PI4P-positive elements distinct from LAMP1-containing structures, in intracellular membranes (Hammond et al., 2009). We first analysed the presence of partly overlapping, closely juxtaposed domains PI4P fluorescence intensity in the Golgi complex, defined by enriched in each of the two components (Fig. 5F). This observation positivity for an anti-giantin (encoded by Golgb1) antibody suggests that PI4P and LAMP1 may occupy the same structure, (Fig. 5A), and found that Golgi PI4P revealed by the antibody was within which they would segregate into spatially distinct domains significantly increased in shRNA clone #1 (∼90% silenced) compared (Fig. 5F,c). to clone #6 (∼60% silenced) or control clones (Fig. 5B). Since antibody binding to PI4P can be influenced by competition Qualitative observations suggested that, in addition to Golgi with endogenous PI4P binding proteins (Goto et al., 2016; PI4P, there was also a striking increase in PI4P-positive vesicles Hammond et al., 2009), we also probed cells by transfection with dispersed throughout the cytoplasm (Fig. 5A). We investigated the an unbiased GFP-conjugated PI4P probe, P4CSidC (Luo et al., relation of the PI4P-positive vesicles to endosomal compartments, 2015). The increase of PI4P on acidic (Lysotracker-positive) and found that colocalization of PI4P with a number of endosome/ vesicles of the silenced clones was also confirmed by this method trans-Golgi network (TGN) proteins – including mannose-6- (Fig. 6A,B). By time-lapse imaging, we could observe Lysotracker phosphate receptor, EEA1, synaptobrevin 2 and APPL1 – was and the GFP probe moving together on rectilinear trajectories, very low (R.V., M.A.D.M., unpublished work). A stronger, although confirming that they were contained on the same structures (Fig. 6B; only partial, colocalization of PI4P was observed with the lysosome Movie 1). late-endosome marker lysosome-associated membrane protein 1 To investigate whether the rescue of neuritogenesis of the (LAMP1). As shown in Fig. 5C and D, in both shRNA clones, PI4P silenced cells by myc-VAPB reported in Fig. 4E correlated with a staining in LAMP1-positive compartments was significantly diminution in the levels of PI4P, we quantified the PI4P content of increased over that in control cells, with clone #6 showing levels LAMP1-positive elements in transfected silenced (#1) and control intermediate between clone #1 and control. PI4P-positive vesicles (C16) cells. As shown in Fig. 7, transfection of myc-VAPB strongly were also observed in the neurites of the differentiated silenced reduced the PI4P content in the silenced cells, compared to those NSC34 cells, and were often seen to concentrate below the plasma transfected with the empty vector, while not affecting the control membrane of growth cones (Fig. 5E). clone; this observation is in agreement with the lack of effect of A higher-resolution analysis of the spatial relationship between myc-VAPB overexpression on the neuritogenesis of the cells of this

PI4P and LAMP1-positive structures, carried out by structured clone (Fig. 4E). Journal of Cell Science

5 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 4. See next page for legend.

Reduction of PI4P levels by phosphoinositide kinase and Godi, 2004; Di Paolo and De Camilli, 2006), we hypothesized inhibitors increases neuritogenesis in NSC34 clones that the delay in neuritogenesis of the VAPB-silenced clones As the membrane expansion occurring during neurite elongation could be related to the observed altered intracellular PI4P levels. To requires exocytosis of vesicle carriers (Sann et al., 2009), and because test this hypothesis, we attempted to pharmacologically reduce the phosphinositides are key regulators of membrane traffic (De Matteis intracellular levels of the phosphoinositide in the clones by treating Journal of Cell Science

6 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 4. VAPB-downregulated clones show reduced neurite extension. DISCUSSION (A) Expression of neuronal markers and VAPB in UD or 6-day-differentiated Because ALS8 is transmitted by autosomal-dominant inheritance (Diff) shRNA- or vector-only VAPB clones. GAPDH, tubulin or calnexin were (Nishimura et al., 2004), many investigations on the pathogenic used as loading controls. The levels of each protein in the differentiated cells, normalized to the values before differentiation, are indicated below the lanes. mechanism of the P56S mutant of VAPB have either analysed (B) Representative images of untransfected and shRNA-VAPB clone #1 cells cultured cells overexpressing the mutant protein, or have been in basal conditions and after 3 and 6 days of differentiation. Cells were stained carried out on transgenic animals carrying the ALS8 gene (reviewed with anti-tubulin antibodies (red) and DAPI (blue). Scale bar: 100 μm. in Navone et al., 2015). The effects that have been observed in (C) Percentage of cells with at least one neurite longer than 100 μm [mean±s.e.m. acutely overexpressing cells – such as delayed transport along the of two (clone C11) or three (all other cell lines) independent experiments) in secretory pathway (Kuijpers et al., 2013b), dysregulation of the each of which ∼250 cells were analysed]. Two-way ANOVA on the 3- and 6-day unfolded protein response (Kanekura et al., 2006), sensitivity of samples revealed highly significant differences between cell lines and between – time points (P<0.0001). Statistical significance, by Bonferroni’s post test, of neurons to stress (Suzuki et al., 2009; Teuling et al., 2007) silenced clones versus untransfected cells at 6 days of differentiation, is combined with the observation that wild-type VAPB, as well as indicated by the asterisks. Differences between silenced clones and control VAPA and other interacting proteins, are sequestered within clone C16 were also statistically significant at 3 and 6 days (#6: P<0.05 and P56S-VAPB inclusions (Darbyson and Ngsee, 2016; Kanekura P<0.001, respectively; #1: P<0.05 and P<0.01, respectively), and clone C11 et al., 2006; Suzuki et al., 2009; Teuling et al., 2007), have led to was significantly different from clone #6 and #1 at 6 days (P<0.001 and P<0.05, the hypothesis that mutant VAPB acts by a dominant-negative respectively). (D) Classification of cells according to the length of the longest neurite [means+s.e.m. of two (C11) or three (all other cell lines) independent mechanism. Results in transgenic mice have not, however, lent experiments]. The column representing the untransfected cells was presented support to this hypothesis: of the four transgenic lines reported so in Fig. 1B and is shown again here for comparison. (E,F) VAPB rescues neurite far (Aliaga et al., 2013; Kuijpers et al., 2013a; Qiu et al., 2013; elongation. Cells were co-transfected with siRNA-resistant myc-tagged VAPB Tudor et al., 2010), only one, in which the mutant protein was highly (E) or VAPA (F) or the corresponding empty plasmid, together with soluble overexpressed, developed mild motor abnormalities and loss of BFP. 24 h after transfection, the cells were transferred to complete cortical, but not spinal, motoneurons (Aliaga et al., 2013). differentiating medium and then fixed after 3 days. Means of three (E) or two (F) independent experiments are shown. Statistical significant was assessed by The discrepancy between the cell and animal studies might be repeated measures one-way ANOVA. myc-VAPA was less efficient than myc- explained by the ease with which cells are able to clear mutant VAPB at rescuing neuritogenesis (F). An immunoblot of lysates prepared from VAPB; indeed, our previous studies showed that, in HeLa cells, parallel, transfected, cultures grown in normal medium, is displayed below P56S-VAPB, although aggregated, is less stable than the wild-type each graph (E,F). The vertical lines separate lanes deriving from the same blot protein, and degraded by the proteasome with a half-life of 7–8h exposure. (Genevini et al., 2014; Papiani et al., 2012). Therefore, in vivo, mutant VAPB, expressed from a single allele, may not accumulate them with the PI4 kinase (PI4K) IIIβ inhibitor PIK93 (Knight et al., to levels sufficient to inflict damage to motoneurons. In agreement, 2006). PI4KIIIβ is known to produce PI4P at the Golgi complex VAPB levels in the brains of P56S transgenic mice are considerably (Godi et al., 1999) and has more recently been reported to operate also lower than those in mice transduced with the wild-type protein, on lysosomes (De Leo et al., 2016; Sridhar et al., 2013). although the corresponding mRNA is expressed equally in the In preliminary experiments, we found that PIK93, administered at two strains (Aliaga et al., 2013). Similarly, and importantly, concentrations close to or above its half-maximal inhibitory motoneurons generated from ALS8 patients’ iPSCs do not have concentration (IC50) (19 nM), did not permit survival of the cells P56S-VAPB inclusions but do have reduced levels of VAPB. To be for the minimum 2- to 3-day period required to observe noted, the situation is quite different in a Drosophila model: flies neuritogenesis. When chronically exposed to 10 nM of the drug, expressing mutant Drosophila VAP (Vap33) do indeed develop the cells remained viable for 2 days, and responded to RA and reduced motoneuron symptoms that recapitulate the major hallmarks of ALS serum treatment. We thus compared the extent of neuritogenesis (Chai et al., 2008; Forrest et al., 2013; Ratnaparkhi et al., 2008). between the treated and untreated clones grown under the same Although the basis of the difference between mice and flies is conditions. As illustrated in Fig. 8A (see Figs S2–S4 for examples of presently not understood, a possible explanation could be a reduced analysed fields), both clone #1 and #6 increased neurite extension in efficiency of Drosophila in eliminating the mutant protein. response to the drug. Unexpectedly, we also observed a response to On the basis of the results summarized in the preceding paragraphs, PIK93 by the control (C11) clone, suggesting that physiological levels we previously hypothesized that the dominant inheritance of the of intracellular PI4P might control the timing of neuritogenesis. ALS8 mutation could be due to haploinsufficiency alone. If this were Since PIK93 targets some lipid kinases other than PI4KIIIβ the case, one might expect motor deficit to develop in knockout mice (Knight et al., 2006), we repeated these experiments with the use of (Kabashi et al., 2013) or in P56S-VAPB knock-in mice, in which one a recently developed more specific inhibitor, IN-10 (Rutaganira of the wild-type alleles is replaced by the mutant one (Larroquette et al., 2016). This inhibitor could be used at concentrations above its et al., 2015). Although mild motor defects were observed in these IC50 (3.6 nM) for as long as 12 h without apparent toxicity. mouse models, the animals did not develop full-blown ALS, maybe Examination of cells cultured for 48 h in differentiation medium and because of the slow progression of the VAPB-dependent form of then treated for 12 h with 20 nM IN-10 in the same medium showed the disease, combined with the short lifespan of mice. Hence, we that, under this condition, the inhibitor was effective at lowering reasoned that, for a full understanding of ALS8 pathogenesis, the PI4P in the Golgi complex in both the control and silenced clone. In effects of VAPB depletion, in the absence of expression of the the latter cells, the phosphoinositide was also decreased in LAMP1- mutant, require further investigation in appropriate cell models. positive structures, whereas in the vector-only clone, the low The effects of VAP downregulation in mammalian cultured cells amount of LAMP1-colocalizing PI4P was not affected by the drug have, with few exceptions (Darbyson and Ngsee, 2016; Saita et al., (Fig. 8B; Figs S5 and S6). We then analysed neurite lengths of cells 2009), been investigated by silencing both isoforms together (Dong grown in parallel with those used for the colocalization analyses; as et al., 2016; Kuijpers et al., 2013b; Teuling et al., 2007; Wakana et al., illustrated in Fig. 8C, both the control (C11) and the silenced (#1) 2015; Zhao et al., 2018). In these models, strong phenotypic effects clone exhibited increased neurite extension in response to the drug. have been observed in the distribution of phosphoinositides and Journal of Cell Science

7 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 5. See next page for legend. Journal of Cell Science

8 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 5. Increase in PI4P content in the Golgi complex and in LAMP1- Instead, overexpressed myc-VAPB was without effect on both the positive vesicles in VAPB-silenced cells. (A) Representative images neuritogenesis and vesicular PI4P levels of the control cell line, (maximum intensity projections of z-stacks) of UD VAPB-downregulated and suggesting that the endogenous concentrations of VAPB, while vector-only clones stained for PI4P (red) and giantin (green). Cell boundaries are outlined in white in the merged image. (B) Median values±interquartile required to support the rate of RA-induced neuritogenesis and PI4P ranges of PI4P/giantin ratio in shRNA-VAPB and vector-only clones. For each homeostasis, are not limiting for either of these processes. This contrasts sample, 35–50 cells were analysed. The symbols indicate values for single with the situation in Drosophila, in which endogenous VAP levels are cells. Statistical significance of clone #1 versus the other three clones was limiting for nerve terminal sprouting; indeed, overexpression of VAP in determined by one-way ANOVA. (C) Representative images of UD VAPB- the wild-type background enhances the number and reduces the size of downregulated and vector-only clones stained for PI4P (green) and LAMP1 boutons at neuromuscular junctions (Pennetta et al., 2002). (red). Single sections, chosen on the basis of the highest intensity of LAMP1 staining, are shown. The red arrows and arrowheads in the single channel To seek further support for a causal link between PI4P imbalance insets indicate examples of structures in which PI4P and LAMP1 colocalize. and neuritogenesis, we pharmacologically altered levels of the (D) Median values±interquartile range of PI4P staining in LAMP1-positive phosphoinositide in intracellular compartments. Of the three PI4Ks structures normalized to LAMP1 intensity in UD cells of shRNA-VAPB and acting at intracellular compartments of the secretory pathway, one, vector-only clones (data from n=150 cells pooled from three separate PI4KIIIβ, which is active at the Golgi complex (Godi et al., 1999) and experiments; symbols indicate values for single cells). Values were normalized lysosomes (De Leo et al., 2016; Sridhar et al., 2013), can be to the average of clone C16 in each experiment. Statistical significance was specifically targeted by small molecule inhibitors (Knight et al., tested by one-way ANOVA followed by Bonferroni’s post test. (E) In 7-day- differentiated VAPB shRNA clone #1, PI4P-positive structures are present in 2006; Rutaganira et al., 2016). We treated differentiating NSC34 cells neurites and in growth cones, where they accumulate beneath the plasma with two related PI4KIIIβ inhibitors and found that the drugs membrane (red arrowhead). (F) SIM analysis of VAPB-depleted accelerated neurite outgrowth of VAPB-downregulated as well as undifferentiated NSC34 cells (clone #1), stained as in C. The boxed area in a is control clones. In agreement with our results, improvement of magnified fivefold in b and c. The brightest point projection of b is rotated neuronal survival in Drosophila and C. elegans models of ALS8 was μ counterclockwise on the z-axis by 45° degrees in c. Scale bars: 10 m (A,E,F); obtained by downregulation of PI4KIIIα or PI4KIIIβ (Forrest et al., 20 μm (C). Insets in A and C, ×2 and ×4. 2013) or by PIK93 treatment (Zhang et al., 2017), respectively. In the latter study, however, the much higher concentrations of the inhibitor sphingolipids, and in processes such as response to stress, autophagy, (250 nM versus 10 nM in the current study) could also have affected intracellular transport, delivery of membrane to growing dendrites other kinases (Knight et al., 2006). and cytoskeletal organization. To investigate the effects of VAP We have not, until now, been able to test the contribution of the depletion in a context more relevant to ALS8 pathology, we generated other two PI4Ks acting on intracellular compartments, as specific NSC34 cell lines, in which only VAPB was downregulated. inhibitors are not available, and we have failed so far to establish a We found that, in untransfected NSC34 cells, VAPAwas in four- to possible time window in which to analyse neurite elongation in five-fold excess over VAPB, an observation consistent with the cells silenced for specific PI4K isoforms. Nevertheless, the effect reported relative levels of the transcripts for the two isoforms in obtained with PIK93 and IN-10 supports the hypothesis that VAPB neurons (Zhang et al., 2014). Notwithstanding the modest deficit negatively affects neurite outgrowth, at least in part, by contribution of VAPB to the total VAP pool, even its partial altering intracellular PI4P levels. depletion caused a significant delay in neurite elongation. This What mechanism(s) could underlie the inhibitory effect on reduced neuritogenesis could correspond, in vivo, to insufficient neuritogenesis of excess PI4P at the Golgi and in acidic vesicles? It replacement of the molecular components of motor neuron axons and is well known that PI4P, by recruiting adaptors and regulators at the dendrites in the face of ongoing membrane turnover and, with time, TGN, has a crucial role in the generation of secretory vesicles (De could lead to neurodegeneration. Matteis et al., 2013). Very interestingly, however, the maturation of yeast post-Golgi vesicles to become competent for polarized PI4P imbalance and neuritogenesis exocytosis involves the reduction of their PI4P content, a process In addition to the delay in neurite elongation, the silenced cells showed requiring the yeast OSBP homologue Osh4p and the Sac1 an increase in immunostained PI4P both in the Golgi complex and in phosphatase (Ling et al., 2014; Mizuno-Yamasaki et al., 2010). peripheral vesicles. An increase in peripheral staining was observed Hence, we propose that a PI4P build-up on post-Golgi vesicles of also with the unbiased probe P4CSidC, indicating that the augmented differentiating NSC34 cells may diminish their capacity to undergo PI4P immunostaining corresponds to a genuine increase in PI4P exocytosis, thus delaying the process of plasma membrane expansion content. This observation is in agreement with previous studies in required for neuritogenesis. It is currently thought that a population of other systems, which involved depletion of both VAP isoforms (Dong late endosomes is involved in delivering membrane to the cell surface et al., 2016), of VAPB alone (Darbyson and Ngsee, 2016) or of the during neuritogenesis (Raiborg et al., 2015, 2016; Sann et al., 2009); unique VAP homologue in invertebrates (Forrest et al., 2013; Zhang thus, the PI4P-containing acidic vesicles observed in the silenced et al., 2017). At variance with observations in VAPA/B double- clones could represent stalled intermediates on their way to fusion knockout HeLa cells (Dong et al., 2016), however, we observed very with the plasma membrane. Whether the PI4P build-up in these minor colocalization of peripheral PI4P with classical endosomal vesicles is a consequence of the PI4P overload in the TGN, or a direct markers. We also failed to observe the dramatic changes in actin consequence of the absence of VAPB at ER–endolysosome contact organization reported by Dong et al. (2016) (P.G., F.N., N.B., sites, remains to be established. unpublished work). Instead, in our system, a portion of the PI4P- While the phosphinositide imbalance that we describe here may positive vesicles localized to LAMP1-positive, acidic compartments. directly affect vesicle fusion as suggested above, it is likely that it To investigate the relationship between PI4P imbalance and delayed acts in addition by indirect mechanisms. Given the pivotal role of neuritogenesis, we carried out rescue experiments, monitoring both PI4P in lipid exchange reactions (Wong et al., 2019), alterations in neurite elongation and intracellular PI4P levels. The results its distribution are predicted to impact the lipid composition of the demonstrated that transfection of the silenced clone with myc-VAPB bilayers that delimit compartments of the secretory pathway. rescued both these parameters, suggesting a causal link between them. Changes in sterol and sphingolipid content at the TGN and in Journal of Cell Science

9 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 6. Increase in peripheral staining of PI4P-positive structures revealed by the PI4P-specific probe GFP-P4CSidC. (A) VAPB-depleted and vector-only clones, transfected with the GFP-P4CSidC probe, were incubated with Lysotracker before live imaging. Single confocal planes are shown. The percentages of Lysotracker-containing structures positive for the probe are shown on the right (median values±interquartile range; data from n=150 cells pooled from three separate experiments; symbols indicate values for single cells). Statistical significance was tested by one-way ANOVA followed by Bonferroni’s post test. (B) Selected still images from Movie 1 of UD VAPB-depleted NSC34 cells (clone #1), treated as in A. Cells were imaged for 100 sequential frames, at 3.5-s intervals. The image on the left shows a cell at the beginning of the recording. Boxed areas 1 and 2, magnified on the right (×4.7 and ×4), illustrate two structures positive for both probes that move together. The numbers indicate the time (in seconds) of the sequential frames from the start of the recording. Scale bars: 10 μm. post-Golgi vesicles could interfere with the generation of the lipid 2007), our results, which demonstrate a four- to five-fold excess of microdomains required for polarized secretion (Ledesma et al., VAPA over VAPB, indicate that the VAPB requirement in NSC34 1998; Simons and Ikonen, 1997). cells cannot be explained by a major contribution of VAPB to the total VAP pool. Rather, they suggest that VAPB carries out A specific function for VAPB in motor neurons function(s) that cannot be replaced by VAPA. This hypothesis is A delay in neuritogenesis had previously been observed in PC12 supported by the rescue experiments, carried out with myc-tagged neuroendocrine cells silenced for VAPA (Saita et al., 2009). VAPA and VAPB constructs: overexpression of myc-VAPB However, in contrast to the results reported here for motor neuron- restored neuritogenesis nearly to the levels of control cells, while like cells, the selective knockdown of VAPB in PC12 cells was the corresponding VAPA construct was less effective. These results without effect. Thus, our results appear to reflect a specific raise the question as to the specific molecular interactions that requirement for VAPB in motor neuron physiology. Although the distinguish the two VAP isoforms. vulnerability of spinal motoneurons to VAPB depletion has Although our initial investigation on intracellular PI4P pools in the previously been attributed to the high expression levels of the silenced cells was based on the known role of the VAPs in the OSBP

VAPB isoform in these cells (Larroquette et al., 2015; Teuling et al., cycle (Mesmin et al., 2013), it must be noted that OSBP and other Journal of Cell Science

10 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 7. myc-VAPB transfection reduces PI4P levels in LAMP1-positive elements of VAPB-deleted clone. (A) 24 h after transfection of clones shRNA #1 and control C16 with myc- VAPB, cells were fixed and immunostained with anti-LAMP1, anti- PI4P and anti-myc antibodies. In the merged LAMP1/PI4P images on the right, transfected and untransfected cells, as assessed by myc staining, are indicated by asterisks and rectangles, respectively. Scale bar: 10 μm. Inset, ×4. (B) Median values±interquartile range of PI4P staining in LAMP1-positive structures normalized to LAMP1 intensity in myc-positive cells transfected with myc-VAPB or empty vector (data from n=150 cells pooled from three separate experiments; the symbols indicate values for single cells). Values were normalized to the average of clone C16, transfected with empty vector, in each experiment. Statistical significance was tested by repeated measures one-way ANOVA followed by Bonferroni’s post test.

ORPs can associate with both VAPs (Murphy and Levine, 2016). overexpression of VAPB in the G93A-SOD1 mouse model slowed Nevertheless, a careful comparison of ORP binding affinities for the progression of the disease (Kim et al., 2016a). Thus, the molecular two VAP isoforms, and investigation of possible differential characterization of pathways dependent on the full complement of interactions in vivo have not, to our knowledge, been carried out. A functional VAPB in appropriate motor neuron cellular models will number of studies have focused on interactions between VAPB and be important both for unravelling specific roles of the two VAP binding partners involved in lipid exchange, generation of membrane isoforms and for shedding light on the molecular pathogenesis of contact sites or ER-associated degradation (Amarilio et al., 2005; familial and sporadic ALS. Baron et al., 2014; Costello et al., 2017; Darbyson and Ngsee, 2016; Gomez-Suaga et al., 2017); however, for none of these interacting MATERIALS AND METHODS proteins has a preferential interaction with VAPB over VAPA been Plasmids, antibodies and recombinant proteins demonstrated. In summary, as there are currently no characterized The sources of plasmids and antibodies used in this study are listed in VAPB-specific interactors (Murphy and Levine, 2016), the results Table S1. Recombinant full-length rat VAPA and fragment 1–225 of human reported here should spur research aimed at uncovering differences in VAPB, expressed as glutathione S-transferase (GST) fusion proteins in the binding preferences of the two VAP isoforms. Given the very Escherichia coli BL21, were purified by glutathione-Sepharose 4B affinity chromatography (GE Healthcare Life Sciences, Pittsburgh, PA), and excised large number (>100) (Murphy and Levine, 2016) of VAP interactors, from GST by thrombin digestion. The concentration of purified proteins was it is possible that more than one is involved in the motoneuronal determined with the Bradford Protein Assay (Bio-Rad, Hercules, CA) and damage caused by selective VAPB deficit. corrected for the presence of contaminating polypeptides revealed by Coomassie staining of SDS-polyacrylamide gels (see Fig. 3A). Conclusions Regardless of the molecular interactions underlying the observed Cell culture, differentiation and transfection phenotype of the VAPB-downregulated cells, our results support the NSC34 cells were obtained from Lavinia Cantoni (Mario Negri Institute for hypothesis that motoneuronal cells need the full complement of Pharmacological Research, Milan, Italy) and tested to be mycoplasma free. functional VAPB, expressed from two alleles, for normal They were maintained in Dulbecco’s modified Eagle medium (DMEM) physiology, and that the dominant inheritance of the P56S VAPB supplemented with 10% fetal bovine serum (FBS), penicillin/streptomycin + allele is caused mainly by haploinsufficiency and not by toxicity of and Na pyruvate. To generate stable VAPB-downregulated cell lines, NSC34 the easily cleared mutant protein aggregates. A reduction in VAP were transfected with pSuper.neo.GFP, coding for a VAPB-targeted shRNA, or with the empty plasmid, using JetPEI transfection agent (Polyplus levels has been reported in the spinal cords of human sALS patients Transfection, Illkirch, France). The transfected cells were subjected to and of a transgenic ALS mouse model expressing a mutant protein selection with 500 μg/ml G418 (Gibco/Thermo Fisher Scientific, Waltham, not related to VAPB (G93A-SOD1) (Anagnostou et al., 2010; Kim MA). After ∼4 weeks of growth in selection medium, individual clones were et al., 2016a; Teuling et al., 2007), and a contribution of VAPB collected, amplified and screened for expression of GFP by fluorescence deficiency to ALS has been supported by the observation that microscopy. GFP-expressing clones were then tested for VAPB expression by Journal of Cell Science

11 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Fig. 8. Effect of PI4KIIIβ inhibition on neurite elongation in NSC34 clones. (A) Effect of PIK93: VAPB-downregulated (#1 and #6) and vector-only clones (C11), grown in differentiation medium for 3 days, were treated with 10 nM PIK93 during the last 48 h of the differentiation protocol or left untreated. Left: classification of neurites of untreated and PIK93-treated cells (means±half-range of two independent experiments). Right: median and interquartile ranges of neurite lengths of all the cells pooled from the same two experiments. Statistical significance was tested by the Mann–Whitney test for each treated sample against its untreated control. 500–800 cells were analysed for each sample. See Figs S2–S4 for examples of analysed fields. (B) Effect of PI4KIIIβ-IN-10 on PI4P in the Golgi complex and in LAMP1-positive compartments of NSC34 clones. Cells, grown in differentiation medium for 48 h, were further incubated for 12–14 h in the same medium in the presence or absence of 20 nM IN-10. Representative cells double stained for PI4P (green) and giantin or LAMP1 (red); the corresponding single channel images are presented in Figs S5 and S6. Single confocal sections are shown. Scale bars: 20 μm. Insets, ×2. The median±interquartile range of the PI4P/giantin or PI4P/LAMP1 ratios are shown below. Each symbol represents the ratio in a single field, containing 1–4 cells. Statistical significance for the difference between treated and untreated cells was assessed by the Mann–Whitney test. (C) Neurite length, assessed in cells treated or not treated with IN-10 as in B. Left: classification of neurites as in A (means±s.e.m. of three biological replicates). Right: the median and interquartile ranges of neurite length of all the cells pooled from the same three experiments. Statistical significance was tested by the Mann–Whitney test for each treated sample against its untreated control. ∼400 cells were analysed in each experiment. Journal of Cell Science

12 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061 quantitative PCR (qPCR) and by immunoblotting. Stable clones were a SIM grating of 34 µm was applied. Image stacks were processed and maintained in culture with 250 μg/ml G418. reconstructed with Zen software, using the SIM tool in the Processing palette. To induce NSC34 differentiation, the cells were plated onto Matrigel (BD Bioscience, Franklin Lakes, NJ)-coated plates and grown in differentiating Image analysis medium, consisting of DMEM/F12 supplemented with antibiotics and Na+ Images were analysed with ImageJ software. For all image analyses, randomly pyruvate as above, but containing reduced FBS (1%) and 1 μM all-trans RA selected fields were acquired with the same parameters, taking care to remain (Sigma-Aldrich, St. Louis, MO) (Maier et al., 2013). The medium was below saturation in both channels. For each set of analysed fields, the figures replaced with fresh, RA-containing medium every 2 days. illustrate either the original acquisitions or intensity-adjusted images for easier Transfections of the NSC34 clones were carried out with the use of visualization (ImageJ). In the latter cases, identical adjustment is applied to Lipofectamine 2000 or Lipofectamine LTX with Plus agent (Thermo Fisher the illustrated fields presented for comparison. Scientific), according to the manufacturer’s protocols. Golgi PI4P Treatment of NSC34 clones with inhibitors of PI4KIIIβ inhibitors To quantify the amount of PI4P in the Golgi region, fixed cells were PIK93 (Sigma-Aldrich) was dissolved in dimethyl sulfoxide (DMSO) at a subjected to double staining with anti-PI4P and anti-giantin antibodies, and concentration of 50 mM. Cells were exposed to 10 nM of the drug 24 h after 1.6 μm thick (0.4 μm interval) z-stacks (∼30 cells for each sample) were their transfer to differentiation medium, and incubated in differentiation acquired, centred around the plane with maximum giantin staining. For each medium containing the inhibitor for another 2 days. section, a region of interest corresponding to giantin staining was outlined; PI4KIIIβ-IN-10 (MedChem Express, Sollentuna, Sweden) was dissolved the integrated PI4P fluorescence intensity of this region was determined over in DMSO at a concentration of 10 mM. Cells were allowed to differentiate the entire stack. To avoid artefacts due to differing efficacy of digitonin for 48 h and then incubated for a further 12–14 h in differentiation medium permeabilization of different cells, this value was normalized to that of the in the presence of 20 nM of the drug. giantin staining summed over the entire stack (Fig. 5A). Alternatively, a The stock solutions of both inhibitors were stored in one-use aliquots at single confocal plane, selected on the basis of maximum giantin staining −20°C. Control cells were incubated with an equal volume of the vector. intensity, was acquired with the 40× objective and 0.5× zoom, with an open pinhole (4 airy units) to acquire greater field depth. After delimitation of the SDS–PAGE immunoblotting giantin-positive area by automatic thresholding, the PI4P mean intensity Freshly collected cells or frozen cell pellets were lysed with pre-heated lysis within the thresholded area was determined and normalized to the intensity buffer (2% SDS, 50 mM Tris-HCl pH 6.8), and analysed by SDS–PAGE and of giantin staining, as above (Fig. 8B). immunoblotting, using standard procedures. Protein content was assayed with a BCA Protein Assay Kit (Euroclone, Pero, Italy). Because of cross-reactivity PI4P in LAMP1-positive elements of the anti-protrudin antibody with tubulin, we eliminated the cytosol from the To quantify the amount of PI4P in LAMP1-positive structures, fixed cells samples to be analysed for this protein. In this case, total membrane fractions, were subjected to double staining with anti-PI4P and anti-LAMP1 prepared by ultracentrifugation (100,000 g for 1 h) from postnuclear antibodies. Fifty cells for each condition in three independent experiments supernatants of cell homogenates, were loaded onto the gels. were imaged on the focal plan with the highest LAMP1 signal. A mask Most blots were developed with infrared dye-conjugated secondary corresponding to LAMP1 staining was generated automatically with ImageJ antibodies, scanned with the Odyssey CLx Infrared Imaging System, and software, and the ratio between PI4P and LAMP1 mean intensities in the analysed with Image Studio software (LI-COR Biosciences, Lincoln, NE). mask was determined. Alternatively, peroxidase-conjugated secondary reagents were revealed by enhanced chemiluminescence (Perkin Elmer, Waltham, MA). The films were P4CSidC in Lysotracker-positive elements digitized and band intensities were determined with ImageJ software Individual Lysotracker-positive structures were counted and the mean (National Institutes of Health, Bethesda, MD), after calibration with an intensity in the GFP channel was individually determined. The structure was optical density calibration step table (Stouffer Graphics Arts, Mishawaka, IN). considered positive when its mean GFP fluorescence was higher than that of a plasma membrane region of similar area, determined for each cell individually. The percentage of double-positive Lysotracker and P4C-GFP Fluorescence microscopy structures was calculated over the total Lysotracker-positive structures. For synaptobrevin 2 and tubulin staining, cells grown on coverslips were fixed with 4% paraformaldehyde+4% sucrose, and processed for indirect immunofluorescence after permeabilization with Triton X-100, by standard Neurite length procedures (Papiani et al., 2012). To reveal PI4P on intracellular To clearly visualize cell bodies and neurites, fixed cells were stained with membranes, cells were fixed with 2% paraformaldehyde, permeabilized anti-tubulin antibody. Random fields were acquired at low magnification ∼ μ with digitonin and processed according to the protocol of Hammond et al. (10× objective). In most experiments, fields of 3000×3000 m, obtained ‘ ’ (2009). For live-cell imaging, cells, transfected with the GFP-P4C probe by applying the splice command of Volocity software (Improvision, SidC – (Luo et al., 2015), were incubated with 2 μM LysoTracker Red (Thermo Coventry, UK), were used for analysis (see Figs S2 S4), and the length of Fisher Scientific) for 20 min at 37°C. The cells were then washed twice with the longest neurite of each cell was determined manually with ImageJ – complete medium and imaged using the LSM800 confocal system equipped software. Generally, 200 400 cells/clone/experiment were analysed. with an on-stage incubator (37°C, 5% CO2). Wide-field images were acquired with a Nikon E600 microscope qPCR equipped with an Axiocam HRM digital camera (Carl Zeiss, Oberkochen, Total RNA, extracted from frozen pellets of NSC34 control or VAPB- Germany). Confocal images were acquired with the Zeiss LSM Meta 510 or downregulated cells by the TRIzol method, was retrotranscribed using the LSM 800 confocal system; alternatively, images were acquired with a SuperScript Vilo cDNA Synthesis kit (Thermo Fisher Scientific). cDNA Hamamatsu Orca AG C4742-80-12AG camera in conjunction with a Perkin deriving from 100 ng total RNAwas amplified in triplicate, using predesigned Elmer Ultraview spinning disk acquisition system (Yokogawa CSUX1-A3) functionally tested Taqman gene expression assays specific for the Vapb gene, attached to a Zeiss Axiovert 200 M microscope. The microscopes were on the ViiA7 Real-time PCR system (Thermo Fisher Scientific), using 18S equipped with 20×/0.5, 25×/0.8 oil, 40×/1.3 oil Plan-Neofluar, and 10×/0.3 ribosomal RNA as an internal control. Transcriptional levels of the target gene and 63×/1.4 oil Plan-Apochromat, objectives. were expressed as values normalized to 18S RNA (2−ΔCt). For SIM, cells were imaged with a Plan-Apochromat 63×/1.4 oil objective on a Zeiss LSM880 confocal system. Image stacks of 0.88 μmthicknesswere Statistical analysis taken with 0.126 μm z-steps and 15 images (three angles and five phases per Statistical analyses were performed either with two-tailed Student’s t-test, angle) per z-section. The field of view was ∼75×75 μmat0.065μm/pixel, and Mann–Whitney non-parametric test or ANOVA, followed by Bonferroni’s Journal of Cell Science

13 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061 post test (GraphPad Prism), as indicated in the figure legends. P-values are De Vos, K. J., Mórotz, G. M., Stoica, R., Tudor, E. L., Lau, K.-F., Ackerley, S., indicated in the figures as follows: *P<0.05; **P<0.01; ***P<0.001. For Warley, A., Shaw, C. E. and Miller, C. C. J. (2012). VAPB interacts with the analysis of immunoblotting experiments, statistical analyses were carried mitochondrial protein PTPIP51 to regulate calcium homeostasis. Hum. Mol. Genet. 21, 1299-1311. out on band intensities subjected to logarithmic transformation. Di Paolo, G. and De Camilli, P. (2006). Phosphoinositides in cell regulation and membrane dynamics. Nature 443, 651-657. Acknowledgements Dong, R., Saheki, Y., Swarup, S., Lucast, L., Harper, J. W. and De Camilli, P. In addition to those who kindly supplied reagents (listed in Table S1), we (2016). Endosome-ER contacts control actin nucleation and retromer function acknowledge Maura Francolini for her support with confocal microscopy. through VAP-dependent regulation of PI4P. Cell 166, 408-423. Fasana, E., Fossati, M., Ruggiano, A., Brambillasca, S., Hoogenraad, C. C., Competing interests Navone, F., Francolini, M. and Borgese, N. (2010). A VAPB mutant linked to The authors declare no competing or financial interests. amyotrophic lateral sclerosis generates a novel form of organized smooth endoplasmic reticulum. FASEB J. 24, 1419-1430. Author contributions Forrest, S., Chai, A., Sanhueza, M., Marescotti, M., Parry, K., Georgiev, A., Conceptualization: P.G., M.N.C., R.V., S.F.C., M.A.D.M., F.N., N.B.; Methodology: Sahota, V., Mendez-Castro, R. and Pennetta, G. (2013). Increased levels of P.G., M.N.C., R.V., S.M., S.F.C., P.B., M.A.D.M., F.N., N.B.; Validation: M.A.D.M., phosphoinositides cause neurodegeneration in a Drosophila model of F.N., N.B.; Formal analysis: P.G., M.N.C., R.V., S.M., P.B., N.B.; Investigation: P.G., amyotrophic lateral sclerosis. Hum. Mol. Genet. 22, 2689-2704. M.N.C., R.V., S.M.; Writing - original draft: P.G., F.N., N.B.; Writing - review & editing: Genevini, P., Papiani, G., Ruggiano, A., Cantoni, L., Navone, F. and Borgese, N. P.G., S.F.C., M.A.D.M., F.N., N.B.; Visualization: P.G., M.N.C., R.V., N.B.; (2014). Amyotrophic lateral sclerosis-linked mutant VAPB inclusions do not interfere with protein degradation pathways or intracellular transport in a cultured Supervision: S.F.C., P.B., M.A.D.M., F.N., N.B.; Project administration: F.N.; Funding cell model. PLoS ONE 9, e113416. acquisition: M.A.D.M., F.N. Godi, A., Pertile, P., Meyers, R., Marra, P., Di Tullio, G., Iurisci, C., Luini, A., Corda, D. and De Matteis, M. A. (1999). ARF mediates recruitment of PtdIns-4- Funding OH kinase-beta and stimulates synthesis of PtdIns(4,5)P2 on the Golgi complex. This research was funded by the PNR-Consiglio Nazionale delle Ricerche [Ageing Nat. Cell Biol. 1, 280-287. – Program 2012 2014 to N.B. and F.N.], Fondazione Regionale per la Ricerca Gomez-Suaga, P., Paillusson, S. and Miller, C. C. J. (2017). ER-mitochondria Biomedica [TRANS-ALS grant no. 2015-0023 to F.N. and P.B.], Fondazione signaling regulates autophagy. Autophagy 13, 1250-1251. Telethon, the European Research Council [Advanced Investigator grant no. 670881 Goto, A., Charman, M. and Ridgway, N. D. (2016). Oxysterol-binding protein (SYSMET)] and Associazione Italiana per la Ricerca sul Cancro [to M.A.D.M.]. P.G. activation at endoplasmic reticulum-golgi contact sites reorganizes was a doctoral student supported by Universitàdegli Studi di Milano, and phosphatidylinositol 4-phosphate pools. J. Biol. Chem. 291, 1336-1347. subsequently the recipient of a fellowship from Associazione Italiana per la Hammond, G. R., Schiavo, G. and Irvine, R. F. (2009). Immunocytochemical Miastenia. The Zeiss LSM 510 Meta and Perkin Elmer spinning disk confocal techniques reveal multiple, distinct cellular pools of PtdIns4P and PtdIns(4,5)P(2). microscopes were generous gifts from Fondazione Antonio Carlo Monzino. The Biochem. J. 422, 23-35. purchase of the Zeiss LSM 800 confocal system was co-funded by Regione Han, S. M., Tsuda, H., Yang, Y., Vibbert, J., Cottee, P., Lee, S.-J., Winek, J., Lombardia [Amanda Project CUP_B42F16000440005] and Fondazione Antonio Haueter, C., Bellen, H. J. and Miller, M. A. (2012). Secreted VAPB/ALS8 major Carlo Monzino. sperm protein domains modulate mitochondrial localization and morphology via growth cone guidance receptors. Dev. Cell 22, 348-362. Supplementary information Holthuis, J. C. M. and Menon, A. K. (2014). Lipid landscapes and pipelines in Supplementary information available online at membrane homeostasis. Nature 510, 48-57. http://jcs.biologists.org/lookup/doi/10.1242/jcs.220061.supplemental Huttlin, E. L., Ting, L., Bruckner, R. J., Gebreab, F., Gygi, M. P., Szpyt, J., Tam, S., Zarraga, G., Colby, G., Baltier, K. et al. (2015). The BioPlex network: a systematic exploration of the human interactome. Cell 162, 425-440. References Kabashi, E., El Oussini, H., Bercier, V., Gros-Louis, F., Valdmanis, P. N., Aliaga, L., Lai, C., Yu, J., Chub, N., Shim, H., Sun, L., Xie, C., Yang, W.-J., Lin, X., McDearmid, J., Mejier, I. A., Dion, P. A., Dupre, N., Hollinger, D. et al. (2013). O’Donovan, M. J. et al. (2013). Amyotrophic lateral sclerosis-related VAPB P56S Investigating the contribution of VAPB/ALS8 loss of function in amyotrophic lateral mutation differentially affects the function and survival of corticospinal and spinal sclerosis. Hum. Mol. Genet. 22, 2350-2360. motor neurons. Hum. Mol. Genet. 22, 4293-4305. Kaiser, S. E., Brickner, J. H., Reilein, A. R., Fenn, T. D., Walter, P. and Brunger, Amarilio, R., Ramachandran, S., Sabanay, H. and Lev, S. (2005). Differential A. T. (2005). Structural basis of FFAT motif-mediated ER targeting. Structure 13, regulation of endoplasmic reticulum structure through VAP-Nir protein interaction. 1035-1045. J. Biol. Chem. 280, 5934-5944. Kanekura, K., Nishimoto, I., Aiso, S. and Matsuoka, M. (2006). Characterization Anagnostou, G., Akbar, M. T., Paul, P., Angelinetta, C., Steiner, T. J. and de of amyotrophic lateral sclerosis-linked P56S mutation of vesicle-associated Belleroche, J. (2010). Vesicle associated membrane protein B (VAPB) is decreased in ALS spinal cord. Neurobiol. Aging 31, 969-985. membrane protein-associated protein B (VAPB/ALS8). J. Biol. Chem. 281, Baron, Y., Pedrioli, P. G., Tyagi, K., Johnson, C., Wood, N. T., Fountaine, D., 30223-30233. Wightman, M. and Alexandru, G. (2014). VAPB/ALS8 interacts with FFAT-like Kim, J.-Y., Jang, A., Reddy, R., Hee Yoon, W. and Jankowsky, J. L. (2016a). proteins including the p97 cofactor FAF1 and the ASNA1 ATPase. BMC Biol. 12,39. Neuronal overexpression of human VAPB slows motor impairment and Cashman, N. R., Durham, H. D., Blusztajn, J. K., Oda, K., Tabira, T., Shaw, I. T., neuromuscular denervation in a mouse model of ALS. Hum. Mol. Genet. 25, Dahrouge, S. and Antel, J. P. (1992). Neuroblastoma×spinal cord (NSC) hybrid 4661-4673. cell lines resemble developing motor neurons. Dev. Dyn. 194, 209-221. Kim, Y.-J., Guzman-Hernandez, M. L., Wisniewski, E., Echeverria, N. and Balla, Chai, A., Withers, J., Koh, Y. H., Parry, K., Bao, H., Zhang, B., Budnik, V. and T. (2016b). Phosphatidylinositol and phosphatidic acid transport between the ER Pennetta, G. (2008). hVAPB, the causative gene of a heterogeneous group of and plasma membrane during PLC activation requires the Nir2 protein. Biochem. motor neuron diseases in humans, is functionally interchangeable with its Soc. Trans. 44, 197-201. Drosophila homologue DVAP-33A at the neuromuscular junction. Hum. Mol. Knight, Z. A., Gonzalez, B., Feldman, M. E., Zunder, E. R., Goldenberg, D. D., Genet. 17, 266-280. Williams, O., Loewith, R., Stokoe, D., Balla, A., Toth, B. et al. (2006). A Costello, J. L., Castro, I. G., Schrader, T. A., Islinger, M. and Schrader, M. (2017). pharmacological map of the PI3-K family defines a role for p110alpha in insulin Peroxisomal ACBD4 interacts with VAPB and promotes ER-peroxisome signaling. Cell 125, 733-747. associations. Cell Cycle 16, 1039-1045. Kuijpers, M., van Dis, V., Haasdijk, E. D., Harterink, M., Vocking, K., Post, J. A., Darbyson, A. and Ngsee, J. K. (2016). Oxysterol-binding protein ORP3 rescues the Scheper, W., Hoogenraad, C. C. and Jaarsma, D. (2013a). Amyotrophic lateral Amyotrophic Lateral Sclerosis-linked mutant VAPB phenotype. Exp. Cell Res. sclerosis (ALS)-associated VAPB-P56S inclusions represent an ER quality 341, 18-31. control compartment. Acta Neuropathol. Commun. 1, 24. De Leo, M. G., Staiano, L., Vicinanza, M., Luciani, A., Carissimo, A., Mutarelli, Kuijpers, M., Yu, K. L., Teuling, E., Akhmanova, A., Jaarsma, D. and M., Di Campli, A., Polishchuk, E., Di Tullio, G., Morra, V. et al. (2016). Hoogenraad, C. C. (2013b). The ALS8 protein VAPB interacts with the ER- Autophagosome-lysosome fusion triggers a lysosomal response mediated by Golgi recycling protein YIF1A and regulates membrane delivery into dendrites. TLR9 and controlled by OCRL. Nat. Cell Biol. 18, 839-850. EMBO J. 32, 2056-2072. De Matteis, M. A. and Godi, A. (2004). PI-loting membrane traffic. Nat. Cell Biol. 6, Kumar, N., Leonzino, M., Hancock-Cerutti, W., Horenkamp, F. A., Li, P. Q., Lees, 487-492. J. A., Wheeler, H., Reinisch, K. M. and De Camilli, P. (2018). VPS13A and De Matteis, M. A., Wilson, C. and D’Angelo, G. (2013). Phosphatidylinositol-4- VPS13C are lipid transport proteins differentially localized at ER contact sites. phosphate: the Golgi and beyond. Bioessays 35, 612-622. J. Cell Biol. 217, 3625-3639. De Matteis, M. A. and Rega, L. R. (2015). Endoplasmic reticulum-Golgi complex Lahiri, S., Toulmay, A. and Prinz, W. A. (2015). Membrane contact sites, gateways

membrane contact sites. Curr. Opin. Cell Biol. 35, 43-50. for lipid homeostasis. Curr. Opin. Cell Biol. 33, 82-87. Journal of Cell Science

14 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220061. doi:10.1242/jcs.220061

Larroquette, F., Seto, L., Gaub, P. L., Kamal, B., Wallis, D., Lariviere,̀ R., Vallée, Rutaganira, F. U., Fowler, M. L., McPhail, J. A., Gelman, M. A., Nguyen, K., J., Robitaille, R. and Tsuda, H. (2015). Vapb/Amyotrophic lateral sclerosis 8 Xiong, A., Dornan, G. L., Tavshanjian, B., Glenn, J. S., Shokat, K. M. et al. knock-in mice display slowly progressive motor behavior defects accompanying (2016). Design and structural characterization of potent and selective inhibitors of ER stress and autophagic response. Hum. Mol. Genet. 24, 6515-6529. phosphatidylinositol 4 kinase IIIbeta. J. Med. Chem. 59, 1830-1839. Ledesma, M. D., Simons, K. and Dotti, C. G. (1998). Neuronal polarity: essential Saita, S., Shirane, M., Natume, T., Iemura, S. and Nakayama, K. I. (2009). role of protein-lipid complexes in axonal sorting. Proc. Natl. Acad. Sci. USA 95, Promotion of neurite extension by protrudin requires its interaction with vesicle- 3966-3971. associated membrane protein-associated protein. J. Biol. Chem. 284, Lev, S., Ben Halevy, D., Peretti, D. and Dahan, N. (2008). The VAP protein family: 13766-13777. from cellular functions to motor neuron disease. Trends Cell Biol. 18, 282-290. Sann, S., Wang, Z., Brown, H. and Jin, Y. (2009). Roles of endosomal trafficking in Ling, Y., Hayano, S. and Novick, P. (2014). Osh4p is needed to reduce the level of neurite outgrowth and guidance. Trends Cell Biol. 19, 317-324. phosphatidylinositol-4-phosphate on secretory vesicles as they mature. Mol. Biol. Shirane, M. and Nakayama, K. I. (2006). Protrudin induces neurite formation by Cell 21, 3389-3400. directional membrane trafficking. Science 314, 818-821. Loewen, C. J. R. and Levine, T. P. (2005). A highly conserved binding site in Simons, K. and Ikonen, E. (1997). Functional rafts in cell membranes. Nature 387, vesicle-associated membrane protein-associated protein (VAP) for the FFAT motif 569-572. of lipid-binding proteins. J. Biol. Chem. 280, 14097-14104. Sridhar, S., Patel, B., Aphkhazava, D., Macian, F., Santambrogio, L., Shields, D. Luo, X., Wasilko, D. J., Liu, Y., Sun, J., Wu, X., Luo, Z.-Q. and Mao, Y. (2015). and Cuervo, A. M. (2013). The lipid kinase PI4KIIIbeta preserves lysosomal Structure of the legionella virulence factor, SidC reveals a unique PI(4)P-specific identity. EMBO J. 32, 324-339. binding domain essential for its targeting to the bacterial phagosome. PLoS Stefan, C. J., Manford, A. G., Baird, D., Yamada-Hanff, J., Mao, Y. and Emr, S. D. Pathog. 11, e1004965. (2011). Osh proteins regulate phosphoinositide metabolism at ER-plasma Maier, O., Böhm, J., Dahm, M., Brück, S., Beyer, C. and Johann, S. (2013). membrane contact sites. Cell 144, 389-401. Differentiated NSC-34 motoneuron-like cells as experimental model for Suzuki, H., Kanekura, K., Levine, T. P., Kohno, K., Olkkonen, V. M., Aiso, S. and cholinergic neurodegeneration. Neurochem. Int. 62, 1029-1038. Matsuoka, M. (2009). ALS-linked P56S-VAPB, an aggregated loss-of-function Mesmin, B., Bigay, J., Moser von Filseck, J., Lacas-Gervais, S., Drin, G. and Antonny, B. (2013). A four-step cycle driven by PI(4)P hydrolysis directs sterol/ mutant of VAPB, predisposes motor neurons to ER stress-related death by inducing PI(4)P exchange by the ER-Golgi tether OSBP. Cell 155, 830-843. aggregation of co-expressed wild-type VAPB. J. Neurochem. 108, 973-985. Mitne-Neto, M., Machado-Costa, M., Marchetto, M. C. N., Bengtson, M. H., Taylor, J. P., Brown, R. H., Jr. and Cleveland, D. W. (2016). Decoding ALS: from Joazeiro, C. A., Tsuda, H., Bellen, H. J., Silva, H. C. A., Oliveira, A. S. B., Lazar, genes to mechanism. Nature 539, 197-206. M. et al. (2011). Downregulation of VAPB expression in motor neurons derived Teuling, E., Ahmed, S., Haasdijk, E., Demmers, J., Steinmetz, M. O., from induced pluripotent stem cells of ALS8 patients. Hum. Mol. Genet. 20, Akhmanova, A., Jaarsma, D. and Hoogenraad, C. C. (2007). Motor neuron 3642-3652. disease-associated mutant vesicle-associated membrane protein-associated Mizuno-Yamasaki, E., Medkova, M., Coleman, J. and Novick, P. (2010). protein (VAP) B recruits wild-type VAPs into endoplasmic reticulum-derived Phosphatidylinositol 4-phosphate controls both membrane recruitment and a tubular aggregates. J. Neurosci. 27, 9801-9815. regulatory switch of the Rab GEF Sec2p. Dev. Cell 18, 828-840. Tsuda, H., Han, S. M., Yang, Y., Tong, C., Lin, Y. Q., Mohan, K., Haueter, C., Murphy, S. E. and Levine, T. P. (2016). VAP, a versatile access point for the Zoghbi, A., Harati, Y., Kwan, J. et al. (2008). The amyotrophic lateral sclerosis 8 endoplasmic reticulum: review and analysis of FFAT-like motifs in the VAPome. protein VAPB is cleaved, secreted, and acts as a ligand for Eph receptors. Cell Biochim. Biophys. Acta 1861, 952-961. 133, 963-977. Navone, F., Genevini, P. and Borgese, N. (2015). Autophagy and Tudor, E. L., Galtrey, C. M., Perkinton, M. S., Lau, K.-F., De Vos, K. J., Mitchell, neurodegeneration: insights from a cultured cell model of ALS. Cells 4, 354-386. J. C., Ackerley, S., Hortobágyi, T., Vámos, E., Leigh, P. N. et al. (2010). Neumann, M., Sampathu, D. M., Kwong, L. K., Truax, A. C., Micsenyi, M. C., Amyotrophic lateral sclerosis mutant vesicle-associated membrane protein- Chou, T. T., Bruce, J., Schuck, T., Grossman, M., Clark, C. M. et al. (2006). associated protein-B transgenic mice develop TAR-DNA-binding protein-43 Ubiquitinated TDP-43 in frontotemporal lobar degeneration and amyotrophic pathology. Neuroscience 167, 774-785. lateral sclerosis. Science 314, 130-133. van Blitterswijk, M., DeJesus-Hernandez, M. and Rademakers, R. (2012). How Nguyen, H. P., Van Broeckhoven, C. and van der Zee, J. (2018). ALS genes in the do C9ORF72 repeat expansions cause amyotrophic lateral sclerosis and genomic era and their implications for FTD. Trends Genet. 34, 404-423. frontotemporal dementia: can we learn from other noncoding repeat expansion Nishimura, A. L., Mitne-Neto, M., Silva, H. C. A., Richieri-Costa, A., Middleton, disorders? Curr. Opin. Neurol. 25, 689-700. S., Cascio, D., Kok, F., Oliveira, J. R. M., Gillingwater, T., Webb, J. et al. (2004). Venditti, R., Rega, L. R., Masone, M. C., Santoro, M., Polishchuk, E., Sarnataro, A mutation in the vesicle-trafficking protein VAPB causes late-onset spinal D., Paladino, S., D’Auria, S., Varriale, A., Di Tullio, G. et al. (2019). Molecular muscular atrophy and amyotrophic lateral sclerosis. Am. J. Hum. Genet. 75, determinants of ER-Golgi contacts identified through a new FRET-FLIM system. 822-831. J. Cell Biol. 218, 1055-1065. Papiani, G., Ruggiano, A., Fossati, M., Raimondi, A., Bertoni, G., Francolini, M., Wakana, Y., Kotake, R., Oyama, N., Murate, M., Kobayashi, T., Arasaki, K., Benfante, R., Navone, F. and Borgese, N. (2012). Restructured endoplasmic Inoue, H. and Tagaya, M. (2015). CARTS biogenesis requires VAP-lipid transfer reticulum generated by mutant amyotrophic lateral sclerosis-linked VAPB is protein complexes functioning at the endoplasmic reticulum-Golgi interface. Mol. cleared by the proteasome. J. Cell Sci. 125, 3601-3611. Biol. Cell 26, 4686-4699. Pennetta, G., Hiesinger, P. R., Fabian-Fine, R., Meinertzhagen, I. A. and Bellen, Wong, L. H., Gatta, A. T. and Levine, T. P. (2019). Lipid transfer proteins: the lipid H. J. (2002). Drosophila VAP-33A directs bouton formation at neuromuscular commute via shuttles, bridges and tubes. Nat. Rev. Mol. Cell Biol. 20, 85-101. junctions in a dosage-dependent manner. Neuron 35, 291-306. Yamaji, T. and Hanada, K. (2015). Sphingolipid metabolism and interorganellar Phillips, M. J. and Voeltz, G. K. (2016). Structure and function of ER membrane contact sites with other organelles. Nat. Rev. Mol. Cell Biol. 17, 69-82. transport: localization of sphingolipid enzymes and lipid transfer proteins. Traffic Qiu, L., Qiao, T., Beers, M., Tan, W., Wang, H., Yang, B. and Xu, Z. (2013). 16, 101-122. Widespread aggregation of mutant VAPB associated with ALS does not cause Zewe, J. P., Wills, R. C., Sangappa, S., Goulden, B. D. and Hammond, G. R. V. motor neuron degeneration or modulate mutant SOD1 aggregation and toxicity in (2018). SAC1 degrades its lipid substrate PtdIns4P in the endoplasmic mice. Mol. Neurodegener. 8,1. reticulum to maintain a steep chemical gradient with donor membranes. eLife Raiborg, C., Wenzel, E. M., Pedersen, N. M., Olsvik, H., Schink, K. O., Schultz, 7, e35588. ’ S. W., Vietri, M., Nisi, V., Bucci, C., Brech, A. et al. (2015). Repeated ER- Zhang, Y., Chen, K., Sloan, S. A., Bennett, M. L., Scholze, A. R., O Keeffe, S., endosome contacts promote endosome translocation and neurite outgrowth. Phatnani, H. P., Guarnieri, P., Caneda, C., Ruderisch, N. et al. (2014). An RNA- Nature 520, 234-238. sequencing transcriptome and splicing database of glia, neurons, and vascular Raiborg, C., Wenzel, E. M., Pedersen, N. M. and Stenmark, H. (2016). ER- cells of the cerebral cortex. J. Neurosci. 34, 11929-11947. endosome contact sites in endosome positioning and protrusion outgrowth. Zhang, W., Colavita, A. and Ngsee, J. K. (2017). Mitigating motor neuronal loss in Biochem. Soc. Trans. 44, 441-446. C. elegans model of ALS8. Sci. Rep. 7, 11582. Ratnaparkhi, A., Lawless, G. M., Schweizer, F. E., Golshani, P. and Jackson, Zhao, Y. G., Liu, N., Miao, G., Chen, Y., Zhao, H. and Zhang, H. (2018). The ER G. R. (2008). A Drosophila model of ALS: human ALS-associated mutation in contact proteins VAPA/B interact with multiple autophagy proteins to modulate VAP33A suggests a dominant negative mechanism. PLoS ONE 3, e2334. autophagosome biogenesis. Curr. Biol. 28, 1234-1245.e4. Journal of Cell Science

15