© 2017. Published by The Company of Biologists Ltd | Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

RESEARCH ARTICLE PIGN prevents aggregation in the endoplasmic reticulum independently of its function in the GPI synthesis Shinji Ihara1,2,*, Sohei Nakayama1, Yoshiko Murakami3, Emiko Suzuki2,4, Masayo Asakawa1, Taroh Kinoshita3 and Hitoshi Sawa1,2

ABSTRACT et al., 1999). The GPI-anchor is a post-translational modification ∼ Quality control of in the endoplasmic reticulum (ER) is found in 60 proteins in yeast and 150 proteins in humans. GPI is essential for ensuring the integrity of secretory proteins before their synthesized and transferred to proteins in 12 stepwise reactions, release into the extracellular space. Secretory proteins that fail to pass which are catalyzed by ER-associated enzymes encoded by more – – quality control form aggregates. Here we show the PIGN-1/PIGN is than 20 the so-called called PIG genes including PIGN required for quality control in Caenorhabditis elegans and in (Ferguson et al., 2009; Kinoshita et al., 2008). PIGN catalyzes the mammalian cells. In C. elegans pign-1 mutants, several proteins fail transfer of phosphoethanolamine (EtNP) to the first mannose residue to be secreted and instead form abnormal aggregation. PIGN- at precursors of the GPI-anchor (Gaynor et al., 1999; Hong et al., knockout HEK293 cells also showed similar protein aggregation. 1999). Autosomal-recessive in the human PIGN were Although PIGN-1/PIGN is responsible for glycosylphosphatidylinositol identified in patients with multiple congenital anomalies-hypotonia- (GPI)-anchor biosynthesis in the ER, certain mutations in C. elegans seizures syndrome 1 (MCAHS1, Online Mendelian Inheritance in pign-1 caused protein aggregation in the ER without affecting GPI- Man, OMIM 614080) (Maydan et al., 2011). In addition to the anchor biosynthesis. These results show that PIGN-1/PIGN has a original mutations reported in patients with this disease, many conserved and non-canonical function to prevent deleterious protein recessive missense PIGN mutations have also been identified in aggregation in the ER independently of the GPI-anchor biosynthesis. patients with MCAHS1 (Brady et al., 2014; Couser et al., 2015; PIGN is a causative gene for some human diseases including Fleming et al., 2015; Khayat et al., 2015; Nakagawa et al., 2015; multiple congenital seizure-related syndrome (MCAHS1). Two pign-1 Ohba et al., 2014). Recently, some patients with Fryns syndrome mutations created by CRISPR/Cas9 that correspond to MCAHS1 also (OMIM 229850), which causes lethality during the neonatal period cause protein aggregation in the ER, implying that the dysfunction of through congenital diaphragmatic hernia (CDH) and other associated the PIGN non-canonical function might affect symptoms of MCAHS1 malformed features, have also been reported to harbor recessive and potentially those of other diseases. mutations in the PIGN gene (McInerney-Leo et al., 2016). By screening for mutants of Caenorhabditis elegans that KEY WORDS: GPI, PIGN, Endoplasmic reticulum, Protein conferred defects in protein secretion, we identified an os156 aggregation, MCAHS1 patients in the pign-1 gene encoding a PIGN homolog. Although the pign-1(os156) mutant possessed normal GPI-anchor INTRODUCTION biosynthetic activity, protein aggregation was observed in the ER. In eukaryotic cells, newly synthesized secretory proteins must pass Two other mutations equivalent to those observed in patients with quality control by chaperones for proper folding in the endoplasmic MCAHS1 and that affect the GPI anchor biosynthesis also caused reticulum (ER) before they are secreted and become functional. similar protein aggregation. In a mammalian cell line that lacks the Misfolded proteins that fail to pass this quality control form PIGN gene, accumulation of protein aggregates – as in C. elegans – aggregates that cause ER stress. To mitigate this ER stress, the was observed Our findings demonstrate a conserved function for unfolded protein response (UPR) attenuates translation and induces PIGN-1/PIGN in the regulation of protein quality control in the ER, expression of ER residential chaperones (Schröder and Kaufman, and the lack of it might be involved in symptoms of patients with 2005; Walter and Ron, 2011). If the ER stress is not resolved within PIGN mutations. a certain time period, cells undergo apoptosis (Tabas and Ron, 2011; Walter and Ron, 2011). Therefore, aggregated proteins in the RESULTS ER are harmful for cellular viability. Identification of a C. elegans mutant with intracellular The PIGN protein was originally identified as an enzyme that accumulation of secretory proteins synthesizes the glycosylphosphatidylinositol (GPI)-anchor (Gaynor To identify genes required for the efficient secretion of basement membrane (BM) proteins, we used a transgenic line of C. elegans that expresses type IV collagen::mCherry (emb-9::mCherry) (Ihara 1Multicellular Organization Laboratory, National Institute of Genetics, Mishima 411- 8540, Japan. 2Department of Genetics, School of Life Science, Sokendai, Mishima et al., 2011). EMB-9::mCherry is secreted from body wall muscle 411-8540, Japan. 3Department of Immunoregulation, Research Institute for (BWM) cells and localizes to the BM in vivo (Fig. 1A, Fig. S1A). Microbial Diseases, and WPI Immunology Frontier Research Center, Osaka We conducted a forward genetic screen using EMB-9::mCherry and University, Suita 565-0871, Japan. 4Gene Network Laboratory, National Institute of Genetics, Mishima 411-8540, Japan. succeeded in isolating the recessive mutant os156, which exhibits a fully penetrant phenotype that is characterized by accumulation of *Author for correspondence ([email protected]) EMB-9::mCherry in BWM cells (Fig. 1B). In wild-type animals, S.I., 0000-0002-8352-6889 EMB-9 is continuously secreted from BWM cells (Graham et al., 1997) and localizes primarily to the BM covering most tissues

Received 15 August 2016; Accepted 2 December 2016 throughout larval development (Fig. 1A,C; Fig. S1A,B; Movie 1). Journal of Cell Science

602 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

Fig. 1. Accumulation of secretory protein in BWM cells of the os156 mutant. (A–D) Shown are the localization of EMB-9::mCherry (right), and differential interference contrast (DIC) images (left) of wild type (A,C) and the os156 mutant (B,D) at L1 (A and B) and L3 stages (C,D). Asterisks indicate coelomocytes, scavenger cells that uptake secreted proteins from the pseudocoelom. The insets show enlarged images of the pharynx. Scale bars: 50 μm. (E–H) Localization of other secreted proteins expressed in body wall muscle (BWM) cells (black box in Fig. S1A) in wild type (E,G) and the os156 mutant (F,H). Anterior and ventral surfaces are to the left and bottom, respectively. Asterisks indicate coelomocytes. Animals expressing both EMB-9::mCherry and ssGFP (E,F) or both EMB-9::mCherry and LAM-1::GFP (G,H). DIC, mCherry, GFP, and overlay (mCherry and GFP) images are shown. Scale bars: 10 μm. Accumulation of EMB-9::mCherry colocalizes with ssGFP (F) and LAM-1::GFP (H) in the os156 mutant (arrowheads). (I) Quantification of EMB-9::mCherry puncta in BWM cells around the pharynx (n≥15 worms for each stage). Asterisks denote statistically significant differences (P<0.01; Student’s t-test). Error bars represent the standard error of the mean (±s.e.m.). See also Fig. S1.

In contrast, in the os156 mutant, accumulation of EMB-9::mCherry transmembrane proteins (MOM-5/Frizzled::GFP and PAT-3/ β- fluorescence was observed in all BWM cells (Fig. 1B, D; Fig. S1C; integrin::GFP) showed normal localization on the cell surface in the Movie 2). Compared to wild-type animals, the os156 mutant had os156 mutant (Fig. S1G–J), suggesting that the os156 mutation three times more EMB-9::mCherry localized in the BWM cells in causes a global defect in the secretion of soluble but not the head region (Fig. 1I). In addition, the fluorescence intensity of transmembrane proteins – at least in BWM cells. EMB-9::mCherry at a region of the BM in the pharynx (along the dashed lines in Fig. S1B,C) was decreased ∼30% in the os156 The mutation in the C. elegans homolog of PIGN causes mutant compared to wild type (Fig. S1D), indicating defects in defects in secretion protein secretion. Transgenic animals expressing secreted soluble We mapped the os156 mutation between SNPs located at 991k and GFP (ssGFP) (Fares and Greenwald, 2001), laminin::GFP (lam-1:: 2819k of C. elegans I. Whole-genome sequencing GFP) (Kao et al., 2006) and secreted metalloprotease::Venus (mig- revealed that os156 harbors a missense mutation of the arginine 17::venus) (Nishiwaki et al., 2000) from BWM cells showed similar residue at position 78 of Y54E10BR.1, which we named pign-1 (see dot-like localizations (Fig. 1E–H and Fig. S1E,F). In contrast, below) (Fig. S2). Similar to the os156 mutant, protein accumulation is Journal of Cell Science

603 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717 also present in the pign-1 deletion mutants (ok1118 and os162)that MIG-17) that showed abnormal accumulation in the pign-1(os156) additionally show a larval-arrest phenotype, in contrast to os156 with mutant are non-GPI-anchored proteins, these results indicate that the missense mutation (Fig. 2A,C). The expression of the GFP:: pign-1 is important for secretion of non-GPI-anchored proteins in PIGN-1 fusion protein in the BWM cells (myo-3p::GFP::PIGN-1) the ER. rescued the phenotype of the os156 and ok1118 mutants, confirming that pign-1 is responsible for the phenotype and also showing that PIGN-1 is expressed in various tissues and localizes to the ER PIGN-1 functions cell autonomously (Fig. 2B,C). The predicted membrane amino acid sequence of PIGN-1 indicates that it is homologous to A pign-1 reporter ( pign-1p::GFP) containing a 1.9 kb promoter human PIGN, with 14 transmembrane domains and a large sequence upstream of the pign-1 start codon was expressed at the hydrophilic ER luminal domain containing three conserved motifs highest levels in the pharynx and BWM cells and at lower levels in (Fig. 2D; Fig. S2). The PIGN enzyme adds phosphoethanolamine the intestine and hypodermis (Fig. 2F). The subcellular localization (EtNP) to the first mannose residue of GPI precursors (Fig. 2E) (Hong of PIGN-1 in BWM cells was analyzed by using myo-3p::GFP:: et al., 1999). Because the proteins (ssGFP, EMB-9, LAM-1 and PIGN-1. Consistent with previous studies that have used yeast and

Fig. 2. Molecular cloning and structure of PIGN-1. (A,B) Localization of EMB-9:: mCherry (top) and its overlay images with corresponding DIC images (bottom) at the L3 stage. Scale bars: 10 μm. (A) In the pign-1 (ok1118) mutant, accumulation of EMB-9::mCherry was observed in the BWM cells. (B) The expression of GFP:: PIGN-1 in BWM cells (myo-3p::GFP:: PIGN-1) rescues accumulation of EMB-9:: mCherry in the pign-1 (ok1118) mutant. (C) Quantification of EMB-9::mCherry puncta in wild type, pign-1 mutants and rescued animals (n≥15 animals at the L3 stage). Asterisks indicate statistically significant differences (P<0.01, Student’s t-test). (D) Membrane topology of PIGN-1 in the ER membrane (green). Positions of the pign-1(os156) mutation and those equivalent to MCAHS1 patients are indicated. See also Fig. S2. (E) The reaction catalyzed by PIGN-1/PIGN. PIGN-1/PIGN transfers EtNP to the first mannose of GPI-anchors. (F) Fluorescence image (top) of pign-1 reporter ( pign-1p::GFP) expression and its overlay with the corresponding DIC image (bottom). GFP is expressed at the highest level in the pharynx and BWM cells (arrows). Scale bar: 50 μm. (G,H) Confocal images of BWM cells expressing the ER marker RFP::SP12 together with either (G) GFP::PIGN-1 (WT) or (H) GFP::PIGN-1(os156:R78K). GFP, RFP and the merged images are shown. The merged image (lower) shows the colocalization of GFP::PIGN with RFP::SP12. N, nuclei. Scale bar: 10 μm. Journal of Cell Science

604 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717 human cells (Gaynor et al., 1999; Hong et al., 1999), GFP::PIGN-1 protein aggregation (Fig. S3F). Even upon UPR activation, protein localized to the ER membrane (as determined by colocalization with aggregation is not resolved in the pign-1 mutant. These results show an ER marker, RFP::SP12) (Fig. 2G). GFP::PIGN-1 with the R78K that PIGN-1 is required for quality control to prevent accumulation mutation as in the os156 mutant also localized normally to the ER of protein aggregates in the ER. membrane (Fig. 2H), indicating that the os156 mutation affects the function but not localization of PIGN-1. The role of PIGN in protein quality control is conserved in humans Protein aggregation in the ER of the pign-1 mutant To determine whether the role of pign-1 in protein quality control is The accumulation of secretory proteins (EMB-9::mCherry, ssGFP, conserved in mammalian cells, we expressed ssGFP in wild-type laminin::GFP, and MIG-17::venus) in BWM cells in pign-1(os156) and PIGN-knockout HEK293 cells (Ohba et al., 2014). High levels led us to hypothesize that proteins can amass in organelles, such as of ssGFP puncta were observed in PIGN-knockout HEK293 cells lysosomes, ER or the Golgi complex. To determine these but not in wild-type cells (Fig. 4A, B). These puncta colocalized organelles, we established transgenic animals that express EMB-9 with mCherry-ER-3 (calreticulin) – an ER luminal chaperone fused with the pH-sensitive green fluorescent protein pHluorin, (Fig. 4F) – indicating that the puncta represent aggregation of ssGFP whose fluorescence is quenched at vesicular pH, e.g. pH 5.5 in the ER. Transient transfection of HEK293 cells with vector (Dittman and Kaplan, 2006). In the pign-1 (os156) mutant, expressing wild-type human PIGN abrogated the protein colocalization of the EMB-9::mCherry and EMB-9::pHluorin was aggregation in PIGN-knockout HEK293 cells (Fig. 4C–E). These detected in BWM cells (arrowheads in Fig. 3A) but not in observations strongly suggest that the regulation of protein quality coelomocytes (arrows), i.e. scavenger cells with acidic organelles. control by PIGN is evolutionarily conserved from C. elegans to These results indicate that those proteins accumulated in organelles, humans. such as the ER or Golgi complex, that are near neutral pH. To To exclude the possibility that the aggregation of ssGFP is caused examine organelles with accumulated proteins, we simultaneously by exit defects from ER rather than defects of protein quality visualized EMB-9::mCherry, the ER marker TRAM::YFP (Rolls control, we examined colocalization of Sec-31A-TagRFP-T (a et al., 2002) or the Golgi marker MIG-23::GFP (Nishiwaki et al., marker for ER exit sites and COPII vesicles) (Shibata et al., 2010) 2004). Although the ER in wild type showed the typical and ssGFP in PIGN-knockout HEK293 cells that stably expressed morphology of the reticular network, the ER in the pign-1(os156) ssGFP. Punctate signals of Sec-31A-TagRFP-T did not colocalize mutant did not show normal network morphology and the with ssGFP (Fig. 4G). Furthermore, in the C. elegans sec-16 fluorescence of EMB-9::mCherry was confined to the lumen of (tm5375) deletion mutant, which is defective in the initiation of the vesicle-like exaggerated ER membranes (Fig. 3B,C). In contrast, the coat protein II (COP-II) assembly that is required for transport of morphology of the Golgi complex, which was observed by using vesicles from the ER (Espenshade et al., 1995; Supek et al., 2002), MIG-23::GFP, was not significantly different between wild type EMB-9::mCherry localized normally in the basement membrane and pign-1(os156), and the marker did not colocalize with the around the pharynx (Fig. S3G). These results indicate that the EMB-9::mCherry signals (Fig. S3A,B). protein aggregation is caused by defects in quality control within the Transmission electron microscopy of cross-sections of BWM ER rather than the disruption of COP-II vesicle function. cells of the pign-1(os156) mutant expressing EMB-9::mCherry The phenotype of the pign-1 mutants suggested that mutations in showed large electron-dense structures surrounded by membrane other genes involved in GPI biosynthesis cause similar phenotypes. studded with ribosomes that are likely to be protein aggregations in We analyzed deletion mutants of enzymes involved in the GPI the rough ER (Fig. 3D–F). The largest of these expanded vesicles in biosynthesis: piga-1(tm2939), pigp-1(ok3557), pigo-1(tm1906), our observation was ∼2.5 µm (Fig. S3C). Similar enlarged pigk-1(tm2739) and pgap-2(gk285) (Murata et al., 2012). These structures with ribosomes were observed without the EMB-9:: mutants shared the sterile phenotype, which is likely to be caused by mCherry transgene (Fig. 3G), indicating that protein aggregation defects in the GPI-anchor synthesis. Unexpectedly, in all of these within the ER lumen of the pign-1 mutants was not caused by the deletion mutants, EMB-9::mCherry did not form dot-like overexpression of EMB-9::mCherry. aggregates and was normally detected in the BM of the pharynx It has been reported that protein aggregates within the ER often (Fig. 5A–F). These results indicate that only PIGN-1 among the GPI contain aberrantly formed disulfide bonds (Braakman et al., 1992; biosynthetic enzymes regulates protein quality control in the ER. Marquardt and Helenius, 1992). Therefore, we examined the Because PIGN functions in the middle of the GPI synthesis formation of aberrant disulfide crosslinks in EMB-9::mCherry by pathway (Kinoshita et al., 2008), it is possible that incorrectly SDS-PAGE under non-reducing conditions. Western blot analyses formed GPI-anchor intermediates disrupt protein quality control in with an anti-mCherry antibody showed bands of high molecular the pign-1 mutants. If so, mutation of piga-1, which catalyzes the mass (>500 kDa) at the top of the gel in a sample from the pign-1 first step of the GPI synthesis, should suppress the pign-1 (os156) mutant but not from wild type (Fig. S3D). Under reducing phenotype. However, the pign-1(os156); piga-1(tm2939) double conditions, the bands of high molecular mass were not detected and mutant showed aggregation of EMB-9::mCherry, similar to the a major band of nearly 250 kDa corresponding to full-length EMB- pign-1(os156) single mutant (Fig. 5G, H). These results strongly 9::mCherry was observed in both samples (Fig. S3E) (Graham et al., suggest that PIGN-1 regulates protein quality control independently 1997). These observations confirmed protein aggregation with of the GPI synthesis. aberrant disulfide crosslinking in the ER lumen of the pign-1 To further examine the relationship between the GPI-anchor and (os156) mutant. protein quality control, we analyzed the effects of the os156 Protein aggregation in the ER should activate the UPR pathway. mutation on the GPI-anchor proteins (GPI-AP). We first analyzed In fact, the expression of HSP-4, a homolog of a human chaperone the localization of EGFP::WRK-1 (GPI-AP) in pign-1 mutants. and the UPR stress sensor BiP/GRP78 (Shen et al., 2001), was Consistent with previous reports (Murata et al., 2012), EGFP:: increased in the pign-1(os156) mutant, as determined by hsp-4p:: WRK-1 expressed in BWM cells by the myo-3 promoter (myo-3p::

GFP, indicating the activation of the UPR pathway in response to EGFP::WRK-1) localized to the plasma membrane in wild type Journal of Cell Science

605 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

Fig. 3. The pign-1 (os156) mutant accumulates protein aggregates in the ER. (A) Localizations of EMB-9 fused with the pH-sensitive reporter pHluorin in pign- 1(os156). Arrowheads indicate colocalization of EMB-9::pHlurion and EMB-9::mCherry in BWM cells (right). Arrows indicate coelomocytes. The fluorescence of EMB-9::pHluroin was quenched in coelomocytes but not in BWM cells. (B,C) EMB-9::mCherry (left), ER membrane (TRAM::YFP, center), and merged images (right) in wild type (B) and pign-1 (os156) mutants (C). Arrows indicate nuclei with no protein accumulations. (C′) Magnified images of the boxed area in the bottom right image of C. EMB-9::mCherry accumulates in the lumen of the ER (arrowheads). Scale bars: 10 μm. (D) Illustrations of BWM cells and the BM of the pharynx. A transverse cross-section indicated by a dashed line at the left side is shown on the right side and was observed by using electron microscopy. (E–G) Electron micrographs of BWM cell sections (boxed region in D) in wild type (E) and pign-1 (os156) mutants with (F) or without (G) the emb-9::mCherry transgene at the L3 stage. Asterisks and arrows indicate myofibrils in BWMs and BM, respectively. Middle and right panels in F and G are high-magnification

images of the boxed regions in the left panel. See also Fig. S3. Journal of Cell Science

606 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

Fig. 4. Evolutionary conserved functions of PIGN-1 in protein quality control. (A–D) Localization of ssGFP in wild type (A), PIGN-knockout (B) and PIGN-knockout HEK293 cells transfected with a control plasmid (C) or pME 3-HA:: PIGN (D). (E) Quantification of ssGFP puncta (n=20). Asterisks indicate statistically significant differences (P<0.01, Student’s t-test). (F,G) Localizations of ER luminal marker (F) and ER exit site marker (G). PIGN-knockout HEK293 cells stably expressing ssGFP were transiently transfected with mcherry-ER-3 (F) or Sec31A-TagRFP-T (G) and observed by using confocal laser scanning microscope after transfection. All scale bars: 10 μm.

(Fig. 5I). In the pign-1(ok1118) or piga-1(tm2739) deletion The function of PIGN in protein quality control does not mutants, as expected, EGFP::WRK-1 was not detected on the require its phosphoethanolamine transferase activity plasma membrane (Fig. 5J, L). Surprisingly, in the pign-1(os156, The above results show that the GPI synthesis activity is R78K) mutant, EGFP::WRK-1 localization was normal (Fig. 5K). independent from its role in protein quality control. However, it is We next used PIGN-knockout HEK293 cells expressing C. elegans possible that PIGN catalyzes substrates other than GPI to regulate PIGN-1 with or without the os156 mutation (R78K) and measured protein quality control. PIGN has a large loop between the first and the cell surface expression of CD59 (GPI-AP) as a surrogate for the second transmembrane domain that contains three conserved motifs enzymatic activities by using flow cytometry. Consistent with the present in phosphodiesterases, nucleotide pyrophosphatases and normal localization of EGFP::WRK-1 in the pign-1(os156) mutant, sulfatases, each of which is believed to be important for the PIGN expression of C. elegans PIGN-1 (R78K) and wild-type PIGN-1, as catalytic activity in the EtNP transferase reaction (Fig. 2D; Fig. S2) well as human PIGN restored CD59 location to the plasma (Gaynor et al., 1999). Mutations in each putatively conserved motif membrane at similar levels (arrows in Fig. 5M). These of human PIGN were generated by replacing conserved histidine observations suggest that the PIGN-1 (R78K) protein possesses residues with alanine. The mutations strongly (H263A in motif 3) or normal enzymatic activity (EtNP transferase) despite its moderately (H218A in motif 2) affected GPI synthesis of PIGN, as compromised function in protein quality control. determined by the surface expression of CD59 (GPI-AP) in PIGN- Journal of Cell Science

607 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

Fig. 5. PIGN mediates protein quality control in the ER independently of GPI biosynthesis. (A–H) Localization patterns of EMB-9::mCherry in mutants of genes that are required for GPI-AP biosynthesis. DIC (left) and fluorescence (right) images of animals with the indicated genotypes expressing EMB-9::mCherry. Scale bars: 50 μm. (I–L) Confocal images of animals with the indicated genotypes expressing EMB-9::mCherry and EGFP::WRK-1 (GPI-AP). Cell-surface expression of EGFP::WRK-1, as indicated by arrows, was undetectable in 0/30 wild-type, 1/30 pign-1(os156), 19/30 pign-1(ok1118), and 26/30 piga-1(tm2939) animals. Dashed lines indicate the boundary between muscle cells and hypodermal cells recognized in the corresponding DIC images (not shown). Scale bars: 10 μm. (M) Cell-surface expression of GPI-AP (CD59) in HEK293 cells was determined by flow cytometry. PIGN-knockout HEK293 cells were transiently transfected with an empty vector (blue) or expression constructs (red) of wild-type human PIGN (left), C. elegans PIGN-1 (WT) (center), or mutated PIGN-1 (R78K) (right). PIGN and PIGN-1 (WT and R78K) restored cell surface expression of CD59 (arrows indicate cell populations with high cell surface levels of CD59). Green line indicate isotype control. knockout HEK293 cells (Fig. 6A). Unexpectedly, the H98A phenotype in PIGN-knockout HEK293 cells (Fig. 6B,C). mutation in motif 1 had normal enzymatic activity. Surprisingly, Furthermore, C. elegans PIGN-1 constructs with equivalent all of the PIGN mutants efficiently rescued the protein aggregation mutations (H97A, H214A, H258A) efficiently rescued protein Journal of Cell Science

608 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

Fig. 6. Enzymatic activity of PIGN is not required to regulate protein quality control. (A) Expression of wild-type and mutated (H98A in motif 1, H218A in motif 2 or H263A in motif 3) PIGN fully (wild type and H98A), partially (H218A) and slightly (H263A) restored cell surface expression of CD59 in PIGN-knockout HEK293 cells. Black arrows indicate population of cells with high cell surface levels of CD59. (B) Expression of PIGN (H98A), PIGN (H218A) and PIGN (H263A) fully rescued the accumulation of ssGFP aggregates in PIGN-knockout HEK293 cells. Scale bars: 10 μm. (C) Quantification of puncta in each transfectant (n=20). Asterisks indicate statistically significant differences compared to PIGN-knockout cells (P<0.01, Student’s t-test). See also Fig. S4. aggregation in the C. elegans pign-1(ok1118) mutant (Fig. S4). indistinguishable from that of pign-1(os156) (Fig. 7D–F). To These results clearly show that PIGN-1/PIGN mediates efficient exclude the possibility that the phenotype of the pign-1 (os163 and quality control and secretion of proteins from the ER, independently os164) mutants is caused by abnormal localizations or unstable of their catalytic activities, including that of EtNP transferase but expression of the mutated proteins rather than their abnormal through a so-far-unknown activity, which we named and hereafter functions, we expressed GFP::PIGN-1 constructs with the refer to as the ‘non-canonical function’ of PIGN-1/PIGN. corresponding mutations (S265P and R679Q) (Fig. 7G–I). Although some signals of GFP::PIGN-1(S265P) and GFP::PIGN- The non-canonical function of PIGN is also affected by 1(R679Q) showed abnormal localization that did not overlap with mutations equivalent to those in MCAHS1 patients the ER marker (arrows in Fig. 7H′,I′), the majority of these proteins Multiple congenital anomalies-hypotonia-seizures syndrome 1 still localized to the ER (Fig. 7H, I). Therefore, the protein (MCAHS1) is caused by recessive missense mutations in the aggregation phenotype in these pign-1 mutants is likely to be caused PIGN locus (Maydan et al., 2011). Patients with MCAHS1 exhibit by defects in the non-canonical function rather than localization. developmental delay, hypotonia and epilepsy combined with Taken together, our observations raise the possibility that some multiple congenital anomalies that may lead to early death symptoms of MCAHS1 are caused by defects of the non-canonical (Maydan et al., 2011; Ohba et al., 2014). Although these PIGN function of PIGN for protein quality control in the ER. mutations affect the GPI-anchor biogenesis in mammalian cells (Maydan et al., 2011; Ohba et al., 2014), they may also affect the DISCUSSION non-canonical function of PIGN-1 in protein quality control. We In this study, we revealed an evolutionarily conserved and used the CRISPR/Cas9 system to generate missense mutations in unexpected role of PIGN-1/PIGN in protein quality control in the the C. elegans pign-1 gene (R679Q and S265P) that are identical to ER, which we designated as the non-canonical function. In the those present in MCAHS1 patients (designated as os163 and os164, C. elegans pign-1 mutants, but not in those of other genes belonging respectively) (Fig. 2D, Fig. S2). In the pign-1(os163 and os164) and to the pig family, several proteins failed to be secreted and, instead, the deletion mutants, we observed the defect in EGFP::WRK-1 formed abnormal aggregates in the ER. PIGN-1/PIGN with (GPI-AP) localization (Fig. 7A–C), confirming that these mutations mutations in the catalytic motifs that disrupt its enzymatic activity affect GPI biosynthesis. In terms of aggregation of EMB-9:: for the GPI-anchor biogenesis, efficiently rescued protein mCherry, the phenotypes of pign-1(os163 and os164) were aggregation in PIGN-knockout HEK293 cells and in pign-1 Journal of Cell Science

609 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

Fig. 7. pign-1 mutations corresponding to those in MCAHS1 patients cause protein aggregation in the ER. (A–C) Confocal images of pign-1(os163: R679Q) and pign-1(os164:S265P) mutants expressing EGFP::WRK-1 (GPI-AP). Cell-surface expression of EGFP::WRK-1, as indicated by arrows, was undetectable in 0/30 wild-type, 14/30 pign-1(os163::R679Q), and 17/30 pign-1 (os164:S265P) animals. Dashed white lines indicate the boundaries between muscle cells and hypodermal cells recognized in the corresponding DIC images (not shown). Scale bars: 10 μm. (D,E) Aggregated EMB-9::mCherry in pign-1(os163:R679Q) and pign-1(os164: S265P) mutants that correspond to mutations found in patients with MCAHS1. Scale bars: 50 μm. Bottom panels show magnification of the boxed areas in the top panels. (F) Quantification of puncta in each strain (n≥15 animals at L3 stage). Asterisks denote statistically significant differences compared to wild type (P<0.01, Student’s t-test). (G–I) Confocal images of BWM cells expressing ER marker RFP::SP12 together with either (G) GFP::PIGN-1 (WT), (H) GFP::PIGN-1(R679Q) or (I) GFP::PIGN-1(S265P). GFP, RFP and merged images are shown. N indicates nuclei. Scale bars: 10 μm. (G′–I′) Magnification of the boxed areas in the left panels of G–I. Arrows indicate abnormal punctate localization of mutated GFP::PIGN-1 that does not overlap with the localization of the ER marker.

mutants, indicating that the enzymatic activity of PIGN-1 is not is the case, PIGN might sense ER stress through dysfunction of the required to prevent protein aggregation. In contrast, the os156 GPI pathway to switch its function to the non-canonical role (R78K) mutation in pign-1, which causes the protein aggregation preventing protein aggregation. phenotype in C. elegans, does not affect EtNP transferase activity of How does the non-canonical function of PIGN-1/PIGN prevent the GPI biogenesis pathway. These results clearly show that PIGN- protein aggregation in the ER? One possibility is that activities of 1/PIGN is a bi-functional protein: an EtNP transferase enzyme chaperones are regulated directly or indirectly by PIGN-1/PIGN. (canonical function) and a mediator of protein quality control within BiP, one of the most abundant ER chaperone that refolds misfolded the ER (non-canonical function). proteins is transcriptionally activated following activation of the The PIGN-1 protein possesses 14 transmembrane domains in UPR pathway in response to ER stress (Hendershot et al., 1995; addition to a large first ER luminal loop. This loop contains Zhao et al., 2005). In pign-1 mutants, we observed enhanced residues that are required for canonical and non-canonical reporter expression of HSP-4/BiP, indicating that BiP is activated functions: R78 mutated in os156 is required only for the non- but fails to ameliorate protein aggregation in the pign-1 mutants, canonical function, H214 and H258 in the catalytic motifs are only suggesting that the function of BiP is abrogated. Although required for canonical functions, whereas S265 associated with involvements of other chaperons regulated by PIGN-1/PIGN are MCAHS1 is required for both of the functions (Fig. 2D). possible, PIGN-1/PIGN might be required for correct activity of BiP Therefore, it is plausible that PIGN cannot fulfill both the by directly regulating its activity, its cofactors or its correct functions simultaneously and that these functions might be localization in the ER. Interestingly, it has been reported that the competitive. For example, the canonical activity might inhibit its expression of an ATPase-inactive BiP mutant in cultured cells or the non-canonical function. In addition, it appeared to be reasonable disruption of mouse Sil1 (also known as BAP), which encodes an that ER stress causes dysfunction of PIG proteins (e.g. PIGA) in adenine nucleotide exchange factor for BiP, result in an expanded the ER, resulting in the decrease of PIGN substrates (intermediates ER filled with protein aggregates (Hendershot et al., 1995; Zhao of the GPI synthesis). The lack of the substrates of the canonical et al., 2005). Protein aggregates in the ER and a block of protein activity might stimulate the non-canonical activity of PIGN. If this secretion were also observed following artificial depletion of ATP in Journal of Cell Science

610 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717 cultured cells (Braakman et al., 1992; de Silva et al., 1993; Dorner Plasmid construction and production of transgenic animals et al., 1990; Verde et al., 1995). These studies emphasize the To construct myo-3p::gfp::pign-1, we first generated the gfp::pign-1 importance of ATP for chaperone functions in resolving protein plasmid. A cDNA fragment from yk573C9 corresponding to the entire aggregation within the ER. Interestingly, it has been reported that PIGN-1 protein (residues 1-906) was amplified and inserted immediately overexpression of the S. cerevisiae PIGN homolog MCD4 can after gfp in pPD95.75 (pPD vectors are gifts of A. Fire) in frame with the gfp sequence with a linker sequence (5′-GGATCCGGAGGAGGATCCGGA- facilitate ATP transport – mostly into the Golgi complex and weakly GGAGGATCC-3′) between gfp and pign-1. For the construction of myo- into the ER (Zhong et al., 2003). Therefore, it is tempting to 3p::gfp::pign-1,agfp::pign-1 fragment was amplified by PCR from the speculate that PIGN-1/PIGN facilitates transport of ATP into the gfp::pign-1 plasmid and inserted between the myo-3 promoter and the unc- ER, which is required for correct chaperone function in order to 54 3′ UTR of pPD136.64 by using the In-Fusion technique (Clontech). To prevent protein aggregation. Biochemical analyses of the roles create the mutations R78K, H97A, H214A and H258A in pign-1 cDNA, a PIGN has in the activity of chaperones and/or transport/metabolism QuikChange site-directed mutagenesis kit (Stratagene) was used with of ATP will be an important step toward a better understanding of its primers containing the corresponding nucleotide changes. To generate the non-canonical function. pign-1p::gfp plasmid, the putative promoter sequence encompassing Recently, in addition to the original mutation (Maydan et al., 1960 bp immediately upstream the first methionine codon was PCR- 2011), many new recessive PIGN mutations, mostly missense amplified and inserted into the SphI-PstI sites of pPD95.75. The emb-9p:: emb-9::pHluorin plasmid was generated by replacing mCherry sequences mutations, have been reported in patients with MCAHS1 (Brady with pHluorin (Dittman and Kaplan, 2006) in emb-9p::emb-9:mCherry et al., 2014; Couser et al., 2015; Fleming et al., 2015; Khayat et al., (Ihara et al., 2011) by using the In-Fusion technique. For myo-3p::egfp:: 2015; Nakagawa et al., 2015; Ohba et al., 2014). In addition to wrk-1,aegfp::wrk-1 fragment was amplified by PCR from the egfp::wrk-1 MCAHS1, PIGN mutations with premature stop codons have been plasmid (Murata et al., 2012) and inserted between the myo-3 promoter and identified in patients suffering from Fryns syndrome, a condition the unc-54 3′ UTR of pPD136.64 by using the In-Fusion technique. For that shares some symptoms, e.g. congenital anomalies, with rescue experiments, plasmids myo-3p::gfp::pign-1, myo-3p::gfp::pign-1 MCAHS1 but also shows more-severe symptoms, such as (H97A), myo-3p::gfp::pign-1 (H214A) or myo-3p::gfp::pign-1 (H258A) congenital diaphragmatic hernia and neonatal lethality were injected into pign-1 (os156) or pign-1 (ok1118)/hT2 hermaphrodite (McInerney-Leo et al., 2016). Since mutations of other ‘PIG’ gonads at 10 µg/ml with 50 µg/ml of unc-119 plasmid pDP#MM016B genes involved in the GPI-anchor biosynthesis have been identified (Maduro and Pilgrim, 1995) and 90 µg/ml of digested salmon sperm DNA. Plasmids pign-1p::gfp, myo-3p::gfp::pign-1, myo-3p::gfp::pign-1 (R78K), in patients with similar symptoms (e.g. PIGA mutations in myo-3p::egfp ::wrk-1 (10 µg/ml each) and myo-3p::emb-9::pHluorin MCAHS2 or PIGT mutations in MCAHS3), most symptoms for (2 µg/ml) were each injected into the gonads of unc-119 (ed4) with MCAHS1 seem to be explained by defects in the canonical PIGN 50 µg/ml of pDP#MM016B and 95 µg/ml of digested salmon sperm DNA. function. Nonetheless, the defects in the non-canonical function Microinjection was performed as previously described (Mello et al., 1991). might affect some symptoms of MCAHS1 and Fryns syndrome. Since it is known that protein aggregation decreases cell viability in Microscopy, image acquisition, and analysis several cultured cells (Bucciantini et al., 2002; Rao and Bredesen, Images of C. elegans were acquired by using a Yokogawa spinning disk 2004), disruption of the non-canonical PIGN function might affect confocal scan head mounted on a Zeiss Axioplan 2 microscope equipped symptom of other diseases. Future research on the non-canonical with a 63× Plan Apochromat oil objective lens that was controlled by using function of PIGN/PIGN-1 and the regulation of its function is likely Andor iQ (Andor Technology). Images were also acquired by using a to contribute to the development of novel therapeutic strategies in a confocal laser scanning microscope (Olympus FV1200) equipped with 20× variety of diseases. and 40× objective lenses. Images were optimized and superimposed using Photoshop CS5 Extended (Adobe Systems), and three-dimensional (3D) projections were generated using IMARIS 7.4 (Bitplane). The dots MATERIALS AND METHODS representing accumulated protein aggregates (fluorescence signals of Strains and genetic analysis EMB-9::mCherry in C. elegans and those of secreted soluble GFP The N2 Bristol strain of C. elegans was used as wild-type strain (Brenner, (ssGFP) in HEK293 cells) were counted by using the MeasurementPro 1974). Animals were raised at 20°C. The following alleles and transgenes function of Imaris. All images were acquired by using identical settings and were used in this study: osEx454 [mig-17p:: MIG-17::Venus; unc-119(+)], counted under the same threshold conditions. osEx502 [myo-3p::GFP::PIGN-1; unc-54p::RFP::SP12; unc-119(+)], osEx508 [emb-9p::EMB-9::pHluorin; emb-9p::EMB-9::mCherry; unc- Transmission electron microscopy (TEM) 119(+)], osEx512 [myo-3p::GFP::PIGN-1(R78K); unc-54p::RFP::SP12; Animals were grown at 20°C and prepared for TEM as previously described unc-119(+)], osEx517 [myo-3p::GFP::PIGN-1; sur-5::gfp(+)], osEx524 (Hall et al., 2012). We prepared cross-sections of the pharyngeal region of [mom-5p::MOM-5::GFP; unc-76(+)], osIs54 [pign-1p::GFP; unc-119 L3-stage animals. Fig. 2D shows the region of the cross-sections that was (+)], osIs60[unc-54p::MIG-23::GFP; unc-119(+)], osIs64[myo-3p::YFP:: observed. Samples were observed by using a JEM1010 electron microscope TRAM; unc-119(+)], osIs66[myo-3p:: EGFP::WRK-1; unc-119(+)], (JEOL, Tokyo, Japan). osIs77 [unc-54p::RFP::SP12; unc-119(+)]; arIs37 [myo-3p::ssGFP], pign-1&arx-7 (ok1118), pign-1(os156), pign-1(os162), pign-1(os163), Western blot analysis pign-1(os164), pigo-1(tm1906), piga-1(tm2939), pigp-1 (ok3557), pgap-2 Mixed populations of transgenic worms expressing emb-9::mCherry were (gk285), unc-119(ed4), pigk-1(tm2739), sec-16 (tm5375), qyIs10 [lam-1p:: collected and then disrupted with glass beads by using Micro Smash MS-100 LAM-1::GFP], qyIs42 [pat-3::GFP;genomic ina-1], qyIs44 [emb-9p:: (Tomy) in 100 mM Tris–HCl pH 7.4, 150 mM NaCl, protease inhibitor EMB-9::mCherry], and zcIs4 (hsp-4p::GFP). The pign-1(os156) mutant cocktail (Roche) and 1% (w/v) Triton X-100. After disruption, the lysates was identified through the abnormal localization of EMB-9::mCherry after were rotated at 4°C for 30 min, followed by centrifugation at 17,400 g for mutagenesis of qyIs44 animals with ethyl methanesulfonate. The pign-1 20 min at 4°C. The supernatants were boiled in SDS–PAGE sample buffer (os156) and pign-1(os163) mutants were viable and did not exhibit any with or without 100 mM DTT, separated by 4-12% SDS–PAGE and then obvious phenotype, although the growth rate was less and brood numbers transferred to nitrocellulose membranes. After blocking with PBS were slightly lower than those of wild-type animals. The pign-1(ok1118 and (supplemented with 3% skimmed milk) for 2 h at room temperature, os162), and the deletion mutants of other ‘PIG’ genes (tm1906, tm2939, membranes were immunoblotted with 1:1000 diluted rabbit anti-RFP (which ok3557, gk285, and tm2739) shared the sterile phenotype, which is likely to reacts with mCherry, MBL) for 2 h at room temperature. The filter was be caused by defects in GPI-anchor synthesis. washed three times with Tris-buffered saline containing 0.05% Tween 20 for Journal of Cell Science

611 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

10 min each and then incubated with 1:1000 diluted horseradish-peroxidase- Programs; P40 OD010440), the C. elegans Gene Knockout Consortium and conjugated anti-rabbit IgG (Santa Cruz) at room temperature for 1 h. National Bioresource Project for reagents and strains. We thank Dr Masato Kanemaki (National Institute of genetics) for advice on imaging and the luciferase assay. We thank Akane Oishi for technical assistant for Transmission electron pign-1 deletion in C. elegans strains by using CRISPR/Cas9 microscopy. We thank Dr Akira Nozawa (Ehime Univeristy) for technical advice. We The pign-1 deletion (os162) was generated from the EMB-9::mCherry thank Dr David R. Sherwood (Duke University) for comments on the manuscript. transgenic line using the CRISPR/Cas9 method (Dickinson et al., 2013). We used the Streptococcus pyogenes cas9 gene with its codons optimized Competing interests for C. elegans and the single-guide RNA (sgRNA) expression plasmid The authors declare no competing or financial interests. pDD162 (Addgene, Cambridge, MA). The pDD162 (Peft-3::Cas9+Empty sgRNA) was a gift from Bob Goldstein (Addgene plasmid # 47549). The Author contributions 5′-ACGGGTCCATATCTTGTGG-3′ sequence was used to identify the S.I. designated the experiments. S.I. and H.S. designed mutant screening. S.I., S.N., and M.A. performed the experiments. Y.M. and T.K. performed flow cytometry and sgRNA target sequence. We selected candidate animals for each mutation analyzed the data. E.S. performed electron microscopy and analyzed the based on abnormal localization of EMB-9::mCherry and verified the micrographs. S.I. and H.S. wrote the manuscript. mutations by determining the nucleotide sequences of the all exons of the pign-1 genomic locus. Funding This work was supported by research grants from Astellas Foundation for Research Phosphoethanolamine transferase assay on Metabolic Disorders, the Nakajima Foundation, the Princess Takamatsu Cancer Research Fund, Takeda Science Foundation and the Tomizawa Jun-ichi & Keiko To measure the phosphoethanolamine transferase activity of C. elegans Fund of the Molecular Biology Society of Japan for Young Scientists, JSPS pign-1, pME 3HA::PIGN-1 and pME 3HA::PIGN-1(os156:R78K) were KAKENHI [grant numbers 23111527, 2465715, 24657159, 25113722, 25111731, constructed by replacing the human PIGN cDNA with the corresponding 15K07063 and 22127005]. cDNAs amplified by PCR from myo-3p::gfp::pign-1 and myo-3p::gfp:: pign-1(R78K), respectively, in the pME 3HA::PIGN plasmid in which Supplementary information PIGN cDNA comprising three hemagglutinin (HA) tags at the N terminus is Supplementary information available online at driven by the simian virus 40 (SV40)-derived SRα promoter (Ohba et al., http://jcs.biologists.org/lookup/doi/10.1242/jcs.196717.supplemental 2014). To generate site-specific mutations in the PIGN conserved domains for pME 3HA-human PIGN, the QuikChange Site-Directed Mutagenesis kit References (Agilent) was used with primers containing the appropriate nucleotide Braakman, I., Helenius, J. and Helenius, A. (1992). Role of ATP and disulphide changes (H98A, H218A, and H263A). PIGN-knockout HEK293 cells bonds during protein folding in the endoplasmic reticulum. Nature 356, 260-262. were transiently transfected with wild-type and various mutants of Brady, P. D., Moerman, P., De Catte, L., Deprest, J., Devriendt, K. and PIGN-containing plasmids, and flow cytometric analyses of GPI-AP Vermeesch, J. R. (2014). Exome sequencing identifies a recessive PIGN splice expressed on the cell surface were performed as previously described (Ohba site mutation as a cause of syndromic congenital diaphragmatic hernia. et al., 2014). Eur. J. Med. Genet. 57, 487-493. Brenner, S. (1974). The genetics of Caenorhabditis elegans. Genetics 77, 71-94. Bucciantini, M., Giannoni, E., Chiti, F., Baroni, F., Formigli, L., Zurdo, J., Taddei, Analyses of ssGFP localization in HEK293 cells N., Ramponi, G., Dobson, C. M. and Stefani, M. (2002). Inherent toxicity of ’ ’ aggregates implies a common mechanism for protein misfolding diseases. Nature HEK293 cells were cultured in Dulbecco s modified Eagle s medium 416, 507-511. (DMEM) containing 10% fetal bovine serum and antibiotics. To express Couser, N. L., Masood, M. M., Strande, N. T., Foreman, A. K., Crooks, K., Weck, secreted soluble GFP (ssGFP) in HEK293 cells, the signal sequence of K. E., Lu, M., Wilhelmsen, K. C., Roche, M., Evans, J. P. et al. (2015). The CD59 (1-30 aa) was PCR-amplified and ligated to the CMV promoter- phenotype of multiple congenital anomalies-hypotonia-seizures syndrome 1: driven expression vector of EGFP using the In-Fusion technique to report and review. Am. J. Med. Genet. A 167A, 2176-2181. produce CMVp::ssGFP, which was then transfected into wild-type and de Silva, A., Braakman, I. and Helenius, A. (1993). Posttranslational folding of PIGN-knockout HEK293 cells by using Lipofectamine (Invitrogen). Three vesicular stomatitis virus G protein in the ER: involvement of noncovalent and covalent complexes. J. Cell Biol. 120, 647-655. clones each for wild-type and PIGN-knockout HEK293 cells, which stably Dickinson, D. J., Ward, J. D., Reiner, D. J. and Goldstein, B. (2013). Engineering expressed ssGFP, were selected in a medium containing 500 µg/ml of the Caenorhabditis elegans genome using Cas9-triggered homologous G418 (Sigma-Aldrich). The results of only one clone each for wild-type recombination. Nat. Methods 10, 1028-1034. and PIGN-knockout cells are shown in Fig. 4, and equivalent results were Dittman, J. S. and Kaplan, J. M. (2006). Factors regulating the abundance and obtained using the other clones. The fluorescence of ssGFP expressed in localization of synaptobrevin in the plasma membrane. Proc. Natl. Acad. Sci. USA HEK293 cells was captured by using an Olympus IX71 microscope 103, 11399-11404. Dorner, A. J., Wasley, L. C. and Kaufman, R. J. (1990). Protein dissociation from equipped with a 40× objective lens. To analyze the rescue of aggregated GRP78 and secretion are blocked by depletion of cellular ATP levels. Proc. Natl. ssGFP, PIGN-knockout HEK293 cells were transiently transfected with the Acad. Sci. USA 87, 7429-7432. SRα-promoter-driven expression plasmid pME 3HA::PIGN (Ohba et al., Espenshade, P., Gimeno, R. E., Holzmacher, E., Teung, P. and Kaiser, C. A. 2014). The localization of ssGFP was determined 3 days after transfection. (1995). Yeast SEC16 gene encodes a multidomain vesicle coat protein that To express ER luminal marker and ERES marker, PIGN-knockout interacts with Sec23p. J. Cell Biol. 131, 311-324. HEK293 cells, which stably express ssGFP, were transiently transfected Fares, H. and Greenwald, I. (2001). Regulation of endocytosis by CUP-5, the with CMVp::mCherry-ER-3 or CMVp::Sec31A-TagRFP-T by using Caenorhabditis elegans mucolipin-1 homolog. Nat. Genet. 28, 64-68. Ferguson, M. A. J., Kinoshita, T. and Hart, G. W. (2009). Glycosylphosphatidylinositol Lipofectamine. mCherry-ER-3 was a gift from Michael Davidson Anchors. In Essentials of Glycobiology (ed. A. Varki, R. D. Cummings, J. D. Esko, (Addgene plasmid # 55041). These localizations were analyzed using a H. H. Freeze, P. Stanley, C. R. Bertozzi, G. W. Hart and M. E. Etzler). Chapter 11. confocal laser scanning microscope (Olympus FV1200). NY: Cold Spring Harbor. (https://www.ncbi.nlm.nih.gov/books/NBK1966/) Fleming, L., Lemmon, M., Beck, N., Johnson, M., Mu, W., Murdock, D., Bodurtha, J., Hoover-Fong, J., Cohn, R., Bosemani, T. et al. (2015). Genotype- Acknowledgements phenotype correlation of congenital anomalies in multiple congenital anomalies We thank Dr Yuji Kohara (National Institute of Genetics) for yk573C9; Dr Kazuya hypotonia seizures syndrome (MCAHS1)/PIGN-related epilepsy. Am. J. Med. Nomura (Kyusyu University) for the WRK-1::GFP and mCherry::SP12 vectors; Genet. A 170A, 77-86. Dr Kiyoji Nishiwaki (Kwansei Gakuin University) for the transgenic lines MIG-17:: Gaynor, E. C., Mondesert, G., Grimme, S. J., Reed, S. I., Orlean, P. and Emr, venus, mig-23::GFP, and myo-3p::TRAM::YFP; Dr Sawako Yoshina and Dr Shohei S. D. (1999). MCD4 encodes a conserved endoplasmic reticulum membrane Mitani (Tokyo Women’s Medical University) for the venus::SP12 and aman-2::venus protein essential for glycosylphosphatidylinositol anchor synthesis in yeast. Mol. vectors; Dr Joshua M. Kaplan (Massachusetts General Hospital) for pHluroin vector; Biol. Cell 10, 627-648. Dr Hideki Shibata (Nagoya university) for CMVp::Sec31A-TagRFP-T vector; the Graham, P. L., Johnson, J. J., Wang, S., Sibley, M. H., Gupta, M. C. and Kramer,

Caenorhabditis Genetic Center (funded by the NIH Office of Research Infrastructure J. M. (1997). Type IV collagen is detectable in most, but not all, basement Journal of Cell Science

612 RESEARCH ARTICLE Journal of Cell Science (2017) 130, 602-613 doi:10.1242/jcs.196717

membranes of Caenorhabditis elegans and assembles on tissues that do not Nakagawa, T., Taniguchi-Ikeda, M., Murakami, Y., Nakamura, S., Motooka, D., express it. J. Cell Biol. 137, 1171-1183. Emoto, T., Satake, W., Nishiyama, M., Toyoshima, D., Morisada, N. et al. Hall, D. H., Hartwieg, E. and Nguyen, K. C. Q. (2012). Modern electron microscopy (2015). A novel PIGN mutation and prenatal diagnosis of inherited methods for C. elegans. Methods Cell Biol. 107, 93-149. glycosylphosphatidylinositol deficiency. Am. J. Med. Genet. A 170A, 183-188. Hendershot, L. M., Wei, J. Y., Gaut, J. R., Lawson, B., Freiden, P. J. and Murti, Nishiwaki, K., Hisamoto, N. and Matsumoto, K. (2000). A metalloprotease K. G. (1995). In vivo expression of mammalian BiP ATPase mutants causes disintegrin that controls cell migration in Caenorhabditis elegans. Science 288, disruption of the endoplasmic reticulum. Mol. Biol. Cell 6, 283-296. 2205-2208. Hong, Y., Maeda, Y., Watanabe, R., Ohishi, K., Mishkind, M., Riezman, H. and Nishiwaki, K., Kubota, Y., Chigira, Y., Roy, S. K., Suzuki, M., Schvarzstein, M., Kinoshita, T. (1999). Pig-n, a mammalian homologue of yeast Mcd4p, is involved Jigami, Y., Hisamoto, N. and Matsumoto, K. (2004). An NDPase links ADAM in transferring phosphoethanolamine to the first mannose of the protease glycosylation with organ morphogenesis in C. elegans. Nat. Cell Biol. 6, glycosylphosphatidylinositol. J. Biol. Chem. 274, 35099-35106. 31-37. Ihara, S., Hagedorn, E. J., Morrissey, M. A., Chi, Q., Motegi, F., Kramer, J. M. and Ohba, C., Okamoto, N., Murakami, Y., Suzuki, Y., Tsurusaki, Y., Nakashima, M., Sherwood, D. R. (2011). Basement membrane sliding and targeted adhesion Miyake, N., Tanaka, F., Kinoshita, T., Matsumoto, N. et al. (2014). PIGN remodels tissue boundaries during uterine-vulval attachment in Caenorhabditis mutations cause congenital anomalies, developmental delay, hypotonia, elegans. Nat. Cell Biol. 13, 641-651. epilepsy, and progressive cerebellar atrophy. Neurogenetics 15, 85-92. Kao, G., Huang, C.-C., Hedgecock, E. M., Hall, D. H. and Wadsworth, W. G. Rao, R. V. and Bredesen, D. E. (2004). Misfolded proteins, endoplasmic reticulum (2006). The role of the laminin beta subunit in laminin heterotrimer assembly and stress and neurodegeneration. Curr. Opin. Cell Biol. 16, 653-662. basement membrane function and development in C. elegans. Dev. Biol. 290, Rolls, M. M., Hall, D. H., Victor, M., Stelzer, E. H. and Rapoport, T. A. (2002). 211-219. Targeting of rough endoplasmic reticulum membrane proteins and ribosomes in Khayat, M., Tilghman, J. M., Chervinsky, I., Zalman, L., Chakravarti, A. and invertebrate neurons. Mol. Biol. Cell 13, 1778-1791. Schröder, M. and Kaufman, R. J. (2005). The mammalian unfolded protein Shalev, S. A. (2015). A PIGN mutation responsible for multiple congenital response. Annu. Rev. Biochem. 74, 739-789. anomalies-hypotonia-seizures syndrome 1 (MCAHS1) in an Israeli-Arab family. Shen, X., Ellis, R. E., Lee, K., Liu, C.-Y., Yang, K., Solomon, A., Yoshida, H., Am. J. Med. Genet. A 170A, 176-182. Morimoto, R., Kurnit, D. M., Mori, K. et al. (2001). Complementary signaling Kinoshita, T., Fujita, M. and Maeda, Y. (2008). Biosynthesis, remodelling and pathways regulate the unfolded protein response and are required for C. elegans functions of mammalian GPI-anchored proteins: recent progress. J. Biochem. development. Cell 107, 893-903. 144, 287-294. Shibata, H., Inuzuka, T., Yoshida, H., Sugiura, H., Wada, I. and Maki, M. (2010). Maduro, M. and Pilgrim, D. (1995). Identification and cloning of unc-119, a gene The ALG-2 binding site in Sec31A influences the retention kinetics of Sec31A at expressed in the Caenorhabditis elegans nervous system. Genetics 141, the endoplasmic reticulum exit sites as revealed by live-cell time-lapse imaging. 977-988. Biosci. Biotechnol. Biochem. 74, 1819-1826. Marquardt, T. and Helenius, A. (1992). Misfolding and aggregation of newly Supek, F., Madden, D. T., Hamamoto, S., Orci, L. and Schekman, R. (2002). synthesized proteins in the endoplasmic reticulum. J. Cell Biol. 117, 505-513. Sec16p potentiates the action of COPII proteins to bud transport vesicles. J. Cell Maydan, G., Noyman, I., Har-Zahav, A., Neriah, Z. B., Pasmanik-Chor, M., Biol. 158, 1029-1038. Yeheskel, A., Albin-Kaplanski, A., Maya, I., Magal, N., Birk, E. et al. (2011). Tabas, I. and Ron, D. (2011). Integrating the mechanisms of apoptosis induced by Multiple congenital anomalies-hypotonia-seizures syndrome is caused by a endoplasmic reticulum stress. Nat. Cell Biol. 13, 184-190. mutation in PIGN. J. Med. Genet. 48, 383-389. Verde, C., Pascale, M. C., Martire, G., Lotti, L. V., Torrisi, M. R., Helenius, A. and McInerney-Leo, A. M., Harris, J. E., Gattas, M., Peach, E. E., Sinnott, S., Bonatti, S. (1995). Effect of ATP depletion and DTT on the transport of membrane Dudding-Byth, T., Rajagopalan, S., Barnett, C. P., Anderson, L. K., Wheeler, proteins from the endoplasmic reticulum and the intermediate compartment to the L. et al. (2016). Fryns syndrome associated with recessive mutations in PIGN in Golgi complex. Eur. J. Cell Biol. 67, 267-274. two separate families. Hum. Mutat. 37, 695-702. Walter, P. and Ron, D. (2011). The unfolded protein response: from stress pathway Mello, C. C., Kramer, J. M., Stinchcomb, D. and Ambros, V. (1991). Efficient gene to homeostatic regulation. Science 334, 1081-1086. transfer in C.elegans: extrachromosomal maintenance and integration of Zhao, L., Longo-Guess, C., Harris, B. S., Lee, J.-W. and Ackerman, S. L. (2005). transforming sequences. EMBO J. 10, 3959-3970. Protein accumulation and neurodegeneration in the woozy mutant mouse is Murata, D., Nomura, K. H., Dejima, K., Mizuguchi, S., Kawasaki, N., Matsuishi- caused by disruption of SIL1, a cochaperone of BiP. Nat. Genet. 37, 974-979. Nakajima, Y., Ito, S., Gengyo-Ando, K., Kage-Nakadai, E., Mitani, S. et al. Zhong, X., Malhotra, R. and Guidotti, G. (2003). ATP uptake in the Golgi and (2012). GPI-anchor synthesis is indispensable for the germline development of extracellular release require Mcd4 protein and the vacuolar H+-ATPase. J. Biol. the nematode Caenorhabditis elegans. Mol. Biol. Cell 23, 982-995. Chem. 278, 33436-33444. Journal of Cell Science

613