Identification of two novel endoplasmic reticulum body-specific integral membrane proteins.

The endoplasmic reticulum (ER) body, a large compartment specific to the Brassicales, accumulates β-glucosidase and possibly plays a role in the defense against pathogens and herbivores. Although the ER body is a subdomain of the ER, it is unclear whether any ER body-specific membrane protein exists. In this study, we identified two integral membrane proteins of the ER body in Arabidopsis (Arabidopsis thaliana) and termed them MEMBRANE PROTEIN OF ENDOPLASMIC RETICULUM BODY1 (MEB1) and MEB2. In Arabidopsis, a basic helix-loop-helix transcription factor, NAI1, and an ER body component, NAI2, regulate ER body formation. The expression profiles of MEB1 and MEB2 are similar to those of NAI1, NAI2, and ER body β-glucosidase PYK10 in Arabidopsis. The expression of MEB1 and MEB2 was reduced in the nai1 mutant, indicating that NAI1 regulates the expression of MEB1 and MEB2 genes. MEB1 and MEB2 proteins localize to the ER body membrane but not to the ER network, suggesting that these proteins are specifically recruited to the ER body membrane. MEB1 and MEB2 physically interacted with ER body component NAI2, and they were diffused throughout the ER network in the nai2 mutant, which has no ER body. Heterologous expression of MEB1 and MEB2 in yeast (Saccharomyces cerevisiae) suppresses iron and manganese toxicity, suggesting that MEB1 and MEB2 are metal transporters. These results indicate that the membrane of ER bodies has specific membrane proteins and suggest that the ER body is involved in defense against metal stress as well as pathogens and herbivores.

Proteins entering the secretory pathway are synthesized on the rough endoplasmic reticulum (ER). COPII vesicles, which are approximately 50 nm in diameter, represent the majority of transport vesicles moving from the ER (Alberts et al., 2002). However, plants differ from animals in that they also produce different types of ER-derived vesicles, and these are involved in the accumulation of proteins (Galili, 2004;Hara-Nishimura et al., 2004;Herman and Schmidt, 2004;Herman, 2008). Protein bodies (PBs) store seed storage proteins in the endosperm of maize (Zea mays) and rice (Oryza sativa; Herman and Larkins, 1999;Satoh-Cruz et al., 2010;Nagamine et al., 2011). Precursoraccumulating vesicles accumulate seed protein precursors in the maturing cotyledons of pumpkin (Cucurbita maxima; Hara-Nishimura et al., 1998). Similar structures occur in the seeds of maigo mutants that have a defect in the transport of seed storage proteins between the ER and Golgi in Arabidopsis (Arabidopsis thaliana; Takahashi et al., 2010). KDEL-tailed proteinase-accumulating vesicles and ricinosomes accumulate a papain-type proteinase in dying tissues in mung bean (Vigna mungo) and castor bean (Ricinus communis) seedlings (Toyooka et al., 2000;Schmid et al., 2001). Interestingly, these unique vesicles are larger (approximately 500 nm) than COPII vesicles and appear at specific stages of the plant life cycle.
In Arabidopsis, several research groups identified another type of ER structure, designated the ER body/ fusiform body (Gunning, 1998;Ridge et al., 1999;Matsushima et al., 2003a). The spindle-shaped, 5-to 10-mm-long ER bodies can be easily detected in Arabidopsis expressing an ER-targeted GFP (Hayashi et al., 2001). Electron microscopy revealed that ER bodies have a fibrous structure, have a connection with the ER network, and are surrounded by a single ribosomebearing membrane (Gunning, 1998;Hayashi et al., 2001). In young seedlings, ER bodies are uniformly distributed throughout the epidermis of cotyledons and hypocotyls but subsequently disappear as the plant grows (Matsushima et al., 2002). In contrast, most root tissues constitutively express ER bodies (Matsushima et al., 2003a). Interestingly, wounding or treatment with the wound hormone jasmonate induces ER body accumulation in adult leaves (Matsushima et al., 2002), suggesting that the ER body is involved in pest/ pathogen resistance in Arabidopsis (Hara-Nishimura and Matsushima, 2003;Yamada et al., 2011). In Arabidopsis seedlings, PYK10, a b-glucosidase with the ER-retention signal KDEL, is the major component of the ER body (Matsushima et al., 2003b). Recently, PYK10 was identified as a b-glucosidase that cleaves scopolin, a major glucoside in the root (Ahn et al., 2010). PYK10 is also involved in resistance against the mutualistic root fungus Piriformospora indica (Sherameti et al., 2008). PYK10 deficiency causes hyperinfection by the fungus, which ends the mutualistic interaction and suggests that the ER body is involved in resistance to fungal infection. Structures similar to the ER bodies of Arabidopsis have been reported as dilated ER cisternae in cells of various organs in other Brassicales (Iversen, 1970a(Iversen, , 1970bBehnke and Eschlbeck, 1978).
PYK10 is the major component of the ER body in the seedlings and roots of Arabidopsis (Matsushima et al., 2003b). Wound-induced ER bodies accumulate BGLU18, a b-glucosidase with the ER-retention signal REEL . Structural changes to PYK10 via the mutation or down-regulation of PYK10 and BGLU21 (the closest homolog of PYK10) results in the elongation of the ER body . These findings suggest that ER body contents regulate ER body morphology. Most ER-soluble proteins contain a Cterminal ER-retention signal that is required for retention in the ER. The presence of KDEL on PYK10 and of REEL on BGLU18 suggests that ER-retention mechanisms are used to localize these proteins in the ER body.
In a previous study, Arabidopsis nai1 and nai2 mutants were found to lack ER bodies; NAI1 was found to encode a basic helix-loop-helix-type transcription factor, while NAI2 was found to encode a novel ER body protein Yamada et al., 2008). NAI1 regulates the expression of NAI2 and PYK10. NAI2 is an ER body component involved in ER body formation and PYK10 accumulation (Yamada et al., 2008(Yamada et al., , 2011. NAI1 also regulates the expression of JACALIN-RELATED LEC-TIN genes (JAL22, JAL23, JAL31, JAL33, and PBP1/ JAL30) and GDSL-LIPASE-LIKE PROTEIN genes (GLL23 and GLL25; Matsushima et al., 2004;Nagano et al., 2005Nagano et al., , 2008. PBP1 localizes to the cytosol, while PYK10 localizes to the ER body (Nagano et al., 2005). When cells are broken, PYK10 forms large complexes with JALs and GLLs that regulate the size of the PYK10 complex and may regulate substrate specificity . Although soluble ER body components have been identified, including NAI2, PYK10, and BGLU18, the components of the ER body membrane remain unknown. This study focuses on two novel ER body membrane proteins, the localization of which is regulated by the ER body component NAI2.

Identification of MEB1 and MEB2 by in Silico Analysis
The failure of Arabidopsis nai1 mutants to form ER bodies suggests that the transcription factor NAI1 regulates the expression of ER body genes. Microarray analysis of a nai1-1 mutant showed that the expression of several genes was reduced compared with wildtype plants , including two homologous genes, At4g27860 and At5g24290, that encode integral membrane proteins (Table I). The ATTED-II database showed that the expression profile of At4g27860 and At5g24290 was similar to that of PYK10, NAI1, and NAI2 (Table I), suggesting that these genes encode proteins related to the ER body. We named these genes MEMBRANE PROTEIN OF EN-DOPLASMIC RETICULUM BODY1 (MEB1; At4g27860) and MEB2 (At5g24290). Cell fractionation and proteomic analysis showed that the MEB2 protein cofractionates with NAI2, but the localization of both proteins could not be clearly resolved (Dunkley et al., 2006). These findings suggest that MEB1 and MEB2 localize to the ER body, which includes NAI2 in Arabidopsis.
MEB1 and MEB2 have homologous C-terminal regions designated as DOMAIN OF UNKNOWN FUNCTION125 (DUF125) in the Pfam database (Finn et al., 2010). MEB1 and MEB2 also contain four hypothetical transmembrane domains (Fig. 1A). It is possible that the two transmembrane domains of MEB1 are not actual transmembrane regions, because Table I. Expression levels of MEB1 and MEB2 in the nai1 mutant and correlation with the expression levels of NAI1, NAI2, and PYK10 Fold indicates fold change determined by microarray analysis of Arabidopsis nai1-1 and the Columbia wild type. Cor indicates Pearson's correlation coefficient for gene expression patterns. These correlation coefficients are listed in the ATTED-II database (http://atted.jp; Obayashi et al., 2011). Rank orders of correlation coefficients for each gene are shown in parentheses.  1B). The N-terminal regions of these proteins are enriched in hydrophilic amino acids, but their amino acid sequences diverged highly. Interestingly, MEB1 has two characteristic hydrophobic N-terminal regions, presumably transmembrane domains, which are absent in MEB2, At4g27870, CAO17728, Os06g0103800, and XP_001760792 (Fig.  1A). Instead of these hydrophobic regions, MEB2, At4g27870, Os06g0103800, and XP_001760792 have a short homologous region that we have named the MEB2 domain (Fig. 1C). GenBank/EMBL/DNA Data Bank of Japan was queried using the unique N-terminal regions of MEB1 and MEB2 to identify MEB1-like and MEB2-like proteins, respectively. The potential homologs of MEB1 were found only in the Brassicaceae. In Arabidopsis lyrata, a single protein (EFH45796) appeared to be a homolog of MEB1. In Brassica napus, a tentative consensus sequence (TC11332) was assembled from ESTs (http://compbio.dfci.harvard.edu/tgi/plant.html) that appeared to encode a homolog of MEB1 (Supplemental Fig. S1A). These genes encode proteins that have two transmembrane regions in their N-terminal regions, which is a characteristic feature of MEB1 (Supplemental Fig. S1B). These findings suggest that MEB1-like proteins diverged during the evolution of Brassicales, which have ER bodies. Several proteins with both the MEB2 domain and a partial DUF125 were identified in various plant species (Fig. 1, B and C), suggesting that MEB2-like proteins are widely distributed. Antibodies were generated against the unique Nterminal regions of MEB1 and MEB2 (Fig. 1A). T-DNA insertion mutants of MEB1 and MEB2 were obtained ( Fig. 2A). All mutants have a T-DNA insertion in the protein-coding region and therefore may produce truncated proteins. The anti-MEB1 antibody recognized an approximately 84-kD polypeptide absent in all meb1 mutants, and the anti-MEB2 antibody recognized an approximately 82-kD polypeptide absent in all meb2 mutants (Fig. 2B), indicating that these bands represent the MEB1 and MEB2 polypeptides and showing that full-length MEB1 and MEB2 are expressed in Arabidopsis.

NAI1 Regulates the Expression of MEB1 and MEB2
NAI1 is a basic helix-loop-helix transcription factor that regulates the expression of ER body-related genes in Arabidopsis . The microarray data showed that the expression of MEB1 and MEB2 is reduced in the nai1 mutant (Table I). To determine the expression level of MEB1, MEB2, and At4g27870 in Arabidopsis, quantitative reverse transcription (qRT)-PCR was performed in nai1 mutant and wild-type Arabidopsis seedlings (Fig. 3). The expression of MEB1 and MEB2 was reduced by approximately two-thirds in nai1 mutant seedlings compared with wild-type seedlings, indicating that these genes are regulated by NAI1. This result is consistent with the absence of MEB1 and MEB2 protein in the nai1 mutant (Fig. 2B). The expression of At4g27870 is not reduced by the nai1 mutation, suggesting that NAI1 is not a transcription factor for At4g27870.

MEB1 and MEB2 Are ER Body Integral Membrane Proteins
To examine the localization of MEB1 and MEB2, we generated expression vectors encoding fusion proteins between the fluorescent protein tdTomato (tdTOM) and MEB1 or MEB2. These proteins were transiently  To confirm the localization of MEB1 and MEB2, ER body-rich (P1) and ER network-rich (P100) fractions were isolated using a subcellular fractionation method established previously (Matsushima et al., 2003b). These fractions were subjected to immunoblot analysis using specific antibodies against MEB1 and MEB2. Binding protein (BiP) and GFP-HDEL were detected in both the P1 and P100 fractions (Fig. 4B), indicating that these proteins accumulate in both the ER and the ER body. By contrast, MEB1 and MEB2 were concentrated in the P1 fraction along with the ER body resident proteins NAI2 and PYK10. These data confirm that MEB1 and MEB2 are ER body proteins.
Since MEB1 and MEB2 contain putative transmembrane regions (Fig. 1), we investigated whether they were integral membrane proteins. The P1 fractions were treated with different buffers: a low-salt buffer to extract soluble proteins, a high-salt buffer and alkaline buffer to extract peripheral membrane proteins, or a detergent buffer to extract integral membrane proteins. Following centrifugation, the supernatants and precipitates were subjected to immunoblot analysis (Fig. 4C). MEB1 and MEB2 were not extracted in the low-salt, high-salt, and alkaline buffers but were extracted in the detergent buffer. These data show that MEB1 and MEB2 are integral membrane proteins of the ER body. Figure 3. MEB1, MEB2, and At4g27870 mRNA levels in the wild type and the nai1 mutant. Total RNA from 7-d-old seedlings was subjected to qRT-PCR analysis. The data were normalized to UBIQUITIN10 mRNA levels. The relative abundance of each mRNA was calibrated against the amount of MEB1 mRNA in wild-type plants. Relative quantities were calculated using the 2 2ΔΔCt method (Livak and Schmittgen, 2001). Error bars indicate SE of threshold cycle (Ct) values, which are calculated using the equation 2 2ΔΔCt 6 SE . These data represent the results of three independent biological replicates. B, Immunoblot analysis of subcellular fractions enriched for ER bodies (P1) or ER networks (P100) using antibodies against BiP, GFP (GFP-HDEL), NAI2, MEB1, and MEB2. The same sample volume was loaded in each lane. Fold enrichment in P1 indicates the ratio of P1 to P100 band intensities. C, Immunoblot analysis of the P1 fraction shown in B following buffer treatment and ultracentrifugation to separate soluble and membrane fractions. Low-salt (50 mM NaCl) buffer extracts soluble proteins, highsalt (500 mM NaCl) and alkaline buffers extract peripheral membrane proteins, and detergent (2% [v/v] Triton X-100) buffer extracts integral membrane proteins. ppt, Precipitate; sup, supernatant.

Accumulation of MEB1 and MEB2 on ER Bodies Produced by NAI2
In the Arabidopsis nai2 mutant, which has no ER body, PYK10 protein levels are reduced and PYK10 expression is diffuse throughout the ER network (Yamada et al., 2008). To examine the effect of NAI2 depletion on MEB1 and MEB2 protein levels, we performed immunoblot analysis of MEB1 and MEB2 with the nai2 mutants. Immunoblot analysis showed no reduction in the levels of MEB1 or MEB2 in the Arabidopsis nai2 mutants (Fig. 5A). Next, the localization of MEB1 and MEB2 was assessed in the nai2 mutant by the transient expression of tdTOM-MEB1 and tdTOM-MEB2. The fusion proteins colocalized with the ER GFP signal (Fig. 5B), indicating that NAI2 depletion results in the dispersion of MEB1 and MEB2 to the ER network. This result suggests that MEB1 and MEB2 accumulate on ER bodies produced by NAI2. Since NAI2 is an ER body protein, it is possible that NAI2 forms a complex with MEB1 and MEB2 on the ER body membrane. To examine this, proteins were extracted from Arabidopsis seedlings, and MEB1-and MEB2containing protein complexes were immunoprecipitated using specific antibodies. NAI2 was detected in MEB1and MEB2-containing protein complexes (Fig. 5C), indicating that NAI2 and MEB1 or MEB2 are located in the same complexes. These results suggest that NAI2 interacts directly or indirectly with MEB1 and MEB2 on the ER body membrane. MEB1 and MEB2 have DUF125 sequence. The proteins containing a DUF125 sequence are assigned as the VACUOLAR IRON TRANSPORTER (VIT) family, including CCC1 in yeast (Saccharomyces cerevisiae), AtVIT1 in Arabidopsis, TgVIT1 in tulip (Tulipa gesneriana), NODULIN21 in soybean (Glycine max), and LjSEN1 in Lotus japonicus (Delauney et al., 1990;Li et al., 2001;Kim et al., 2006;Momonoi et al., 2009;Hakoyama et al., 2012). The VIT family includes the iron/manganese uptake transporter. Although the protein structures of MEB1 and MEB2 are different from canonical VITs such as CCC1 in yeast and VIT1 in Arabidopsis, the DUF125 sequence of those proteins could be aligned during phylogenetic tree analysis (Fig. 6A). There are three subfamilies of plant DUF125 proteins: the VIT1 subfamily represented by AtVIT1 (Kim et al., 2006), the SEN1/NODULIN21 subfamily represented by LjSEN1 (Hakoyama et al., 2012), and the MEB1/MEB2 subfamily. In yeast, CCC1 sequesters excess cytosolic iron to the vacuole, and a defect of CCC1 increases iron sensitivity by disrupting the regulation of cytosolic iron concentration (Li et al., 2001;Lin et al., 2011). Interestingly, overexpression of MEB1 or MEB2 in the ccc1 mutant weakly suppressed iron toxicity. The yeast ccc1 mutant could not grow on synthetic Gal medium supplemented with 2 mM iron, whereas the wild type grew on medium with and without iron. The ccc1 mutant transformed with MEB1 or MEB2 could grow on medium containing 2 mM iron, although growth was reduced (Fig. 6B). This indicates that MEB1 or MEB2 partially reduces the iron toxicity in the yeast ccc1 mutant. CCC1 transports manganese, and overexpression of CCC1 suppresses the toxicity resulting from high manganese concentrations in wildtype yeast (Lapinskas et al., 1996). Overproduction of MEB1 or MEB2 also suppresses the growth retardation caused by high concentrations of manganese in yeast (Fig. 6C). These results show that MEB1 and MEB2 serve as functional homologs of CCC1, which is an iron/manganese transporter. In yeast, MEB1 and shows the Rubisco large subunit as a loading control. B, Fluorescence images of cotyledon epidermal cells from 7-d-old seedlings transiently expressing tdTOM-MEB1 or tdTOM-MEB2. A nai2-2 mutant constitutively expressing ER-targeted GFP (GFP-HDEL) was bombarded with gold particles coated with the appropriate plasmids and germinated for 1 d. Bars = 10 mm. C, MEB1 and MEB2 proteins were immunoprecipitated (IP) from Arabidopsis seedlings with anti-MEB1 and anti-MEB2 antibodies. Complexes containing NAI2 protein were detected with biotinylated anti-NAI2 antibody. Sera from preimmunized rabbits were used as negative controls. Input shows samples used for immunoprecipitation (1:100 dilution).
MEB2 may localize in the ER and sequester excess cytosolic iron and manganese into the ER to reduce metal ion toxicity.

MEB1 and MEB2 Are Dispensable for ER Body Formation
To examine the effects of MEB1 and MEB2 on ER body formation, knockout mutants of these proteins were examined. The meb1 and meb2 mutants were isolated, and the meb1 meb2 double mutant was generated. These mutants were viable and similar to wildtype plants. No differences in PYK10 and NAI2 levels were detected between wild-type plants and any of the three mutants (Fig. 7A), suggesting that MEB1 and MEB2 are not essential for the accumulation of ER body components in Arabidopsis. Mutant and wildtype plants showed a similar amount of ER bodies, Figure 6. Functional analysis of MEB1 and MEB2 in yeast. A, Phylogenetic tree of DUF125 domains. DUF125 regions were extracted, and the dendrogram was drawn with ClustalW2 and iTOL. Plant DUF125-containing proteins group into three subfamilies. The MEB1/MEB2 subfamily (red branch) contains MEB1, MEB2, At4g27870, grape CAO17728 (Vitis), rice Os06g0103800, and moss XP_001760792 (Physco). The VIT1 subfamily (blue branch) contains Arabidopsis VIT1 (At2g01770), tulip VIT1 (TgVIT1; BAH98154), and rice VIT1 (Os04g0463400) and VIT2 (Os09g0396900). The SEN1/NODULIN21 subfamily (green branch) contains L. japonicus SEN1 (BAL46698), soybean NODULIN21 (P16313), rice Os04g0538400, and Arabidopsis At3g25190, At1g76800, At1g21140, At3g43630, and At3g43660. The dendrogram also includes DUF125 domain-containing proteins from yeast (S. ). MEB1 and MEB2 are highlighted in red. ScCCC1 and AtVIT1, which have iron/manganese transporter activities, are highlighted in blue. The horizontal scale represents the evolutionary distance expressed as the number of substitutions per amino acid. B, MEB1 and MEB2 suppress the toxicity of iron in the ccc1 mutant. Four yeast strains, the wild type harboring empty vector (WT/vector), ccc1 mutant harboring empty vector (ccc1/vector), MEB1 (ccc1/MEB1), and MEB2 (ccc1/MEB2), were grown on synthetic minimal medium with (left) or without (right) 2 mM iron. The same amount of cells were spotted and germinated for 4 d. C, Three yeast strains, the wild type harboring empty vector (vec; black circles), MEB1 (black triangles), and MEB2 (black squares) were germinated in liquid medium containing 15 mM MnCl 2 , and the growth curves were plotted. White circles indicate the growth curve of yeast without manganese.
with no observable differences in ER body morphology (Fig. 7B), suggesting that MEB1 and MEB2 are not essential for ER body formation or PYK10 accumulation in ER bodies.
At4g27870 is a homolog of MEB1 and MEB2 expressed in seedlings (Fig. 3). Functional redundancy between MEB1, MEB2, and At4g27870 might explain the lack of phenotype in the meb1 meb2 double mutant. Since the proximity of the At4g27870 and meb1 (At4g27860) genes precludes the generation of a meb1 at4g27870 double mutant, two independent artificial microRNA (amiR) genes were constructed for the At4g27870 gene and introduced into the meb1 meb2 double mutant. Two independent lines decreased expression of the At4g27870 gene (Fig. 8A). ER body formation was unchanged in these meb1 meb2 at4g27870 triple mutants (Fig. 8B), suggesting that these membrane proteins are dispensable for ER body formation. Similarly, no differences were detected in PYK10 or NAI2 levels between wild-type plants and meb1 meb2 at4g27870 triple mutants (Fig. 8C), suggesting that these membrane proteins are not essential for the accumulation of ER body components in Arabidopsis.
The ER body is believed to be involved in defense against fungal infection (Hara-Nishimura and Matsushima, 2003;Matsushima et al., 2003a;Sherameti et al., 2008;Yamada et al., 2011). We examined whether pathogen sensitivity is altered in the meb1 meb2 double mutant compared with wild-type plants. We examined Alternaria brassicicola, Fusarium oxysporum, Pythium sylvaticum, and Pseudomonas syringae pv glycinea, which are pathogenic to Arabidopsis accession Columbia. These pathogens gradually damaged wild-type plants. The sensitivity of wild-type and mutant plants to A. brassicicola, F. oxysporum, and P. sylvaticum was not significantly different. In addition, the growth of P. syringae pv glycinea was similar in mutant and wild-type plants (data not shown). These results suggest that MEB1 and MEB2 are not involved in defense against these pathogens.

The ER Body Membrane Is Distinct from ER Network Membranes and Contains MEB1 and MEB2
The ER body is a dilated ER subdomain that accumulates PYK10 and NAI2 (Matsushima et al., 2003b;Yamada et al., 2008Yamada et al., , 2009). The specific components of the ER body membrane remain unknown. We found that MEB1 and MEB2 contain transmembrane regions and localize to the ER body membrane. This indicates that the ER body membrane is distinct from membranes in the ER network. In cereal plants, PBs are ERderived structures that accumulate specific seed proteins. In rice, prolamin mRNAs accumulate on the prolamin PB membrane (Choi et al., 2000), suggesting that this membrane is distinct from the ER network. In maize, FLOURY1 (FL1) is a PB membrane protein involved in PB formation (Holding et al., 2007). Although MEB1 and MEB2 are structurally unrelated to FL1, these findings suggest that other ER-derived structures, such as ricinosomes and KDEL-tailed proteinase-accumulating vesicles (Toyooka et al., 2000;Schmid et al., 2001), may contain integral membrane proteins that distinguish them from the ER network.
The lengths of MEB1 and MEB2 are 611 and 550 amino acids, respectively, and their predicted molecular masses are 68.2 and 61.0 kD, respectively. In SDS-PAGE, the observed molecular masses of MEB1 and MEB2 were approximately 84 and 82 kD, respectively. This difference between predicted and observed molecular masses may be attributable to the large number of hydrophobic residues, which affect the behavior of proteins in SDS-PAGE analysis (Rath et al., 2009). . PYK10 accumulation and ER body morphology are unchanged in meb1 and meb2 mutants. A, PYK10, NAI2, and BiP levels in meb1, meb2, and meb1 meb2 mutants. Total protein extract from 7-d-old seedlings was subjected to immunoblot analysis using antibodies against PYK10, NAI2, or BiP. Coomassie Brilliant Blue staining (CBB) shows the Rubisco large subunit as a loading control. Col, Columbia accession. B, Fluorescence images of cotyledon epidermal cells from 7-d-old seedlings in the wild type (GFP-h) and meb1-1, meb2-1, and meb1-1 meb2-1 mutants expressing ER-targeted GFP (GFP-HDEL). Bar = 10 mm.

MEB1 and MEB2 Accumulate ER Bodies Produced by NAI2
NAI2 is an ER body component that regulates ER body formation. MEB1 and MEB2 were diffusely distributed throughout the ER network in the nai2 mutant. Because there is no ER body in the nai2 mutant, MEB1 and MEB2 cannot localize to the ER body, which resulted in a diffuse ER network in the nai2 mutant. This suggests that NAI2 function is to make an ER body and ER body membrane that acts as a scaffold for MEB1 and MEB2. Alternatively, it is possible that NAI2 itself is a scaffold for MEB1 and MEB2, because NAI2 and MEB1 or MEB2 form a protein complex. We propose that ER body components, such as PYK10 and NAI2, induce an ER body and then trap MEB1 and MEB2 during ER body formation.

Possible Functions of MEB1 and MEB2
Our results indicate that MEB1, MEB2, and At4g27870 are not involved in PYK10 accumulation. Furthermore, MEB1, MEB2, and At4g27870 appear dispensable for ER body formation in Arabidopsis. This implies that MEB1 and MEB2 are mainly involved in ER body function rather than ER body formation.
The C-terminal region of MEB1 and MEB2 contains a DUF125 sequence. Proteins containing a DUF125 sequence are assigned to the VIT family, including CCC1 in yeast, AtVIT1 in Arabidopsis, and OsVIT1 and OsVIT2 in rice. CCC1 is a vacuolar iron/manganese transporter that is responsible for the regulation of cytosolic iron homeostasis (Li et al., 2001). The Arabidopsis protein AtVIT1, a close homolog of CCC1, transports both iron and manganese, but only iron homeostasis is altered following mutation of vit1 in seeds (Kim et al., 2006;Roschzttardtz et al., 2009). OsVIT1 and OsVIT2 transport iron and zinc and are responsible for the accumulation of these metals (Zhang et al., 2012). We found that MEB1 and MEB2 partially reduced the iron sensitivity of the ccc1 mutant and enhanced manganese resistance in yeast. These results suggest that the DUF125 sequence of these proteins underlies their ability to transport metal ion. This also suggests that ER bodies containing MEB1 and MEB2 are involved in metal ion homeostasis. The ER body is constitutively observed in roots (Hara-Nishimura and Matsushima, 2003;Matsushima et al., 2003aMatsushima et al., , 2003b. Root tissues may be exposed to high concentrations of metal ions. Therefore, it is reasonable to expect that Figure 8. PYK10 accumulation and ER body morphology are unchanged in meb1 meb2 at4g27870 triple knockdown/knockout mutants. A, At4g27870 mRNA levels in meb1-1 meb2-1 at4g27870-1 (meb1/2 amiR-1) and meb1-1 meb2-1 at4g27870-2 (meb1/2 amiR-2) triple knockdown/knockout mutants. meb1/2 amiR-1 and meb1/2 amiR-2 plants harbor different constructs to knockdown gene At4g27870 (see "Materials and Methods"). Total RNA from 10-d-old seedlings was subjected to qRT-PCR analysis. The data were normalized to UBIQUITIN10 mRNA levels. The relative quantity of each mRNA was calibrated to the signal observed in wild-type plants (accession Columbia [Col]). The relative quantity was calculated by the 2 2ΔΔCt method (Livak and Schmittgen, 2001). Error bars indicate SE of the threshold cycle (Ct) values, calculated using the equation 2 2ΔΔCt 6 SE . The data represent the results of three independent biological replications. B, Fluorescence images of cotyledon epidermal cells from seedlings transiently expressing ER-targeted GFP. Six-day-old wild-type or meb1 meb2 at4g27870 mutant plants were bombarded with gold particles coated with GFP-HDEL plasmids and germinated for 1 d before inspection. Bar = 10 mm. C, Total protein extract from 10-d-old seedlings was subjected to immunoblot analysis using antibodies against PYK10 and NAI2. Coomassie brilliant blue staining (CBB) shows the Rubisco large subunit as a loading control.
roots have detoxifying mechanisms for metals, which in part may be mediated by ER bodies. Another example of a new function of the ER body was recently revealed. Takahashi et al. (2012) showed that the chlorophyllbinding protein BoWSCP accumulates in the ER body of Brussels sprout (Brassica oleracea). When cells are disrupted, BoWSCP is released from the ER body and then immediately scavenges chlorophyll located in the damaged thylakoid membranes to suppress the production of reactive oxygen species (Takahashi et al., 2012). These observations led to the proposal that the ER body is involved in defense against other environmental stresses in addition to pathogens/herbivores.
To produce at4g27870 knockdown plants in meb1 meb2 double mutants, two amiRs specific to At4g27870 were constructed. The method was described previously (Schwab et al., 2006). Two amiR candidate DNA sequences (amiRat4g27870-1 and amiRat4g27870-2) were calculated using the WMD3 program (http://wmd3.weigelworld.org/cgi-bin/webapp.cgi). Specific primers were generated (Supplemental Table S1), and DNA fragments were amplified by PCR using primers amiR oligo A and B (Schwab et al., 2006) and the template amiR vectors in pRS300 (a kind gift from D. Weigel). The DNA fragments were digested with EcoRI/NotI and introduced into the EcoRI/NotI site of the Gateway entry vector pENTR1a (Invitrogen). The entire amiR coding regions were transferred into the binary vector pMDC7 (Curtis and Grossniklaus, 2003) to generate pMDC7/amiRat4g27870-1 and pMDC7/amiRat4g27870-2. These plasmids encode the amiR gene under the control of the estradiol-inducible promoter. The plasmids were introduced into GV3101 (pMP90RK) and then transformed into meb1-1 meb2-1 double mutants. Transgenic homozygote plants were germinated on a plate containing 5 mM 17b-estradiol and analyzed (Zuo et al., 2002).

Full-Length cDNAs for MEB1 and MEB2
Complementary DNA (cDNA) fragments containing the MEB1 and MEB2 genes were PCR amplified using gene-specific primer sets (Supplemental Table S1) and EST clone templates (U21078 for MEB1 and U23495 for MEB2). The fragments were introduced into the Gateway entry vector pENTR/SD/ D-TOPO (Invitrogen) to produce pENTR/MEB1 and pENTR/MEB2. The EST clones were Salk/Stanford/Plant Gene Expression Center Consortium fulllength cDNA/open reading frame clones obtained from the Arabidopsis Biological Resource Center.

qRT-PCR
qRT-PCR was performed as reported previously (Yamada et al., 2008). Total RNAs were isolated from 20 7-d-old wild-type or mutant line seedlings using 700 mL of Isogen (Nippon Gene). RNA was dissolved in 175 mL of distilled water, and first-strand cDNA was synthesized from 5 mL of the RNA solution using Ready-to-Go RT-PCR beads (GE Healthcare Life Science) and random oligomers. Real-time PCR was performed using the 7500 Fast Real-Time PCR System (Applied Biosystems) and a TaqMan Gene Expression Assay Kit (Applied Biosystems) according to the manufacturer's instructions. The assay identifiers were At02300790_g1 for MEB1, At02185786_g1 for MEB2, At02300797_g1 for At4g27870, and At02358313_s1 for UBIQUITIN10. The 2 2DDCt method was used for relative quantification (Livak and Schmittgen, 2001).

Microscopy
A confocal laser scanning microscope (LSM510; Carl Zeiss) was used to observe fluorescent proteins. An argon laser (488 nm) and a 505/530-nm bandpass filter were used to observe GFP, whereas a helium-neon laser (543 nm) and a 560/615-nm band-pass filter were used for tdTOM.

Subcellular Fractionation and Extraction of Membrane Protein
The subcellular fractionation method was described previously (Yamada et al., 2008). Briefly, 0.83 g of 8-d-old seedlings was chopped on ice in 2.5 mL of chopping buffer that contained 50 mM HEPES-NaOH, pH 7.5, 5 mM EDTA, 0.4 M Suc, and protease inhibitor cocktail (one tablet per 50 mL; Boehringer-Mannheim). The homogenate was filtered through cheesecloth and then centrifuged at 1,000g at 4°C for 20 min. The pellet was designated the P1 fraction. The supernatant was centrifuged once at 8,000g at 4°C for 20 min and then again at 100,000g at 4°C for 1 h. The pellet after ultracentrifugation was designated the P100 fraction. The P1 and P100 fractions were resuspended in the same volume of chopping buffer and subjected to immunoblot analysis. For the separation of soluble and membrane components, the P1 fractions obtained from 8-d-old seedlings (2 g) were treated with different buffers: a low-salt buffer (10 mM HEPES-NaOH [pH 7.5], 1 mM EDTA, and 50 mM NaCl), a high-salt buffer (10 mM HEPES-NaOH [pH 7.5], 1 mM EDTA, and 500 mM NaCl), an alkaline buffer (0.1 M Na 2 CO 3 [pH 11]), or a detergent buffer (10 mM HEPES-NaOH [pH 7.5] and 2% [v/v] Triton X-100). The mixtures were then centrifuged at 100,000g for 30 min and separated into soluble (supernatant) and precipitate fractions. The precipitate was resuspended in the buffer of equal volume to supernatant and subjected to immunoblot analysis.

Immunoprecipitation
Total protein was extracted from 20 8-d-old Arabidopsis seedlings using 350 mL of Tris-buffered saline (TBS; 50 mM Tris-HCl [pH 7.5] and 150 mM NaCl) and protease inhibitor cocktail (one tablet per 50 mL; Boehringer-Mannheim). The homogenate was centrifuged at 9,000g for 3 min at 4°C to remove cell debris. Protein extract (100 mL) was mixed with 900 mL of TBS and anti-MEB1 serum (2.5 mL) or anti-MEB2 serum (0.5 mL) and then incubated overnight at 4°C. Following the addition of 20 mL of Protein G Sepharose (GE Healthcare Life Science) beads, the mixture was incubated for 3 h.

Yeast Strains and Analysis
Yeast (Saccharomyces cerevisiae) strains BY4741 (wild type; MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) and ccc1 (BY4741 Δccc1::KanMX) were used (these strains were kind gifts from Y. Kamada and Y. Ohsumi, respectively). The entire protein-coding region of pENTR/MEB1 and pENTR/MEB2 was transferred to vector p415 GAL1-GW to generate p415/MEB1 and p415/ MEB2. These plasmids carry a Gal-inducible promoter. The Gateway destination vector p415 GAL1-GW was derived from p415 GAL1 (Mumberg et al., 1994), which had been modified into a Gateway vector by using the Gateway Vector Conversion System (Invitrogen). Cassette B from the Gateway Vector Conversion System was inserted in the SmaI restriction site of p415 GAL1. Yeast strains harboring a control vector (p415 GAL1), p415/MEB1, or p415/MEB2 were grown in synthetic minimal medium containing 2% (w/v) Gal (SG) supplemented with His, Met, and uracil but not Leu. For the iron toxicity assay, 5 mL of cell suspension (optimal density at 600 nm = 0.1) was spotted on an SG plate containing 2% (w/v) agar supplemented with His, Met, and uracil, with or without 2 mM FeSO 4 , and subsequently cultured at 30°C for 4 d. For the manganese toxicity assay, 10 mL of cell suspension (optimal density at 600 nm = 1) was applied to 5 mL of SG containing 2% (w/v) Suc supplemented with His, Met, and uracil, with or without 15 mM MnCl 2 , and subsequently cultured at 30°C and 300 rpm. The optimal density at 600 nm was monitored at various time intervals.

Pathogen Infection Experiment
Alternaria brassicicola (MAFF726527) was obtained from the National Institute of Agrobiological Sciences gene bank. A. brassicicola, Fusarium oxysporum strain NDF9 (a kind gift from M. Nozue), and Pythium sylvaticum strain DSM2322 (a kind gift from P. Schulze-Lefert and P. Bednarek) were maintained in potato dextrose agar (BD Difco). Pseudomonas syringae pv glycinea (a kind gift from D. Takemoto) was maintained in Plusgrow medium (Nacalai Tesque). Conidia of A. brassicicola were suspended in 0.5% (w/v) Triton X-100 and placed on the cotyledons of 8-d-old seedlings, and symptoms were observed for 2 weeks. Mycelia of F. oxysporum or P. sylvaticum were placed close to aseptically grown, 1-week-old plants, and symptoms were observed for 1 week. P. syringae cells were suspended in a solution containing 10 mM MgCl 2 and 0.01% (w/v) Triton X-100, and seedlings were placed in the solution for 1 h. After several days, the bacteria were extracted from cotyledons by homogenization, and the number of colony-forming units was counted.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. Fluorescence intensities of GFP and tdTOM in Figure 4A.
Supplemental Table S1. Nucleotide sequences of oligonucleotide primers used in this study.