The N-terminal region of Jaw1 has a role to inhibit the formation of organized smooth endoplasmic reticulum as an intrinsically disordered region

Jaw1/LRMP is a type II integral membrane protein that is localized at the endoplasmic reticulum (ER) and outer nuclear membrane. We previously reported that a function of Jaw1 is to maintain the nuclear shape as a KASH protein via its carboxyl terminal region, a component of linker of nucleoskeleton and cytoskeleton complex in the oligomeric state. Although the oligomerization of some KASH proteins via the cytosolic regions serves to stabilize protein-protein interactions, the issue of how the oligomerization of Jaw1 is regulated is not completely understood. Therefore, we focused on three distinct regions on the cytosolic face of Jaw1: the N-terminal region, the coiled-coil domain and the stem region, in terms of oligomerization. A co-immunoprecipitation assay showed that its coiled-coil domain is a candidate for the oligomerization site. Furthermore, our data indicated that the N-terminal region prevents the aberrant oligomerization of Jaw1 as an intrinsically disordered region (IDR). Importantly, the ectopic expression of an N-terminal region deleted mutant caused the formation of organized smooth ER (OSER), structures such as nuclear karmellae and whorls, in B16F10 cells. Furthermore, this OSER interfered with the localization of the oligomer and interactors such as the type III inositol 1,4,5-triphosphate receptor (IP3R3) and SUN2. In summary, the N-terminal region of Jaw1 inhibits the formation of OSER as an IDR to maintain the homeostatic localization of interactors on the ER membrane.


Results
Jaw1 coiled-coil domain is a candidate for the oligomerization site. As we previously reported, Jaw1 plays a role in maintaining the nuclear shape as a KASH protein in the state of oligomer 28 . However, the oligomerization site has not been identified. Furthermore, the possibility that the oligomerization is due to the interaction between the KASH domain and the SUN domain of the PNS has not been excluded. Therefore, we first investigated the issue of whether the cytosolic region is involved in the oligomerization or not. To accomplish this, we focused on three distinct cytosolic regions: the N-terminal region of the coiled-coil domain (N-terminal region), the coiled-coil domain and the C-terminal region of the coiled-coil domain (stem region). For a coimmunoprecipitation (Co-IP) assay, we prepared plasmids encoding GFP-fused mouse Jaw1 long form (GFP Ms LJaw1) and GFP-fused mutants encoding entire cytosolic regions (GFP Ms LJaw1 ΔTM) and single regions (GFP Ms LJaw1 N, GFP Ms Jaw1 Coil or GFP Ms Jaw1 Stem) (Fig. 1A). These GFP-fused proteins were co-expressed with HA FLAG tandem tagged Ms LJaw1 (FLAG Ms LJaw1) in HEK293 cells and co-immunoprecipitated by treatment with anti-FLAG beads. The results showed that GFP Ms LJaw1 was co-immunoprecipitated by the FLAG Ms LJaw1, consistent with previous data 28 . Importantly, GFP Ms LJaw1 ΔTM and GFP Ms Jaw1 Coil were also co-immunoprecipitated by FLAG Ms LJaw1, indicating that Jaw1 forms an oligomer on the cytosolic face via the coiled-coil domain (Fig. 1B, see Supplementary Fig. S1 online).
In order to verify the oligomerization site in terms of localization, these GFP fused mutants were expressed alone or co-expressed with FLAG Ms LJaw1 in B16F10 cells derived from mouse melanoma cells followed by immunostaining. The confocal images showed that GFP Ms LJaw1, GFP Ms LJaw1 ΔTM and GFP Ms Jaw1 Coil were localized in the cytosol, whereas GFP Ms LJaw1 N and GFP Ms Jaw1 Stem were localized in the cytosol but also in the nucleus (Fig. 1C). Furthermore, the GFP Ms LJaw1 N and GFP Ms Jaw1 Stem that were localized in the nucleus were not translocated to the cytosol by the co-expression of the FLAG LJaw1, which implies that the N-terminal region and the stem region of Jaw1 do not function as oligomerization sites, consistent with www.nature.com/scientificreports/ Co-IP data. These data suggest that the coiled-coil domain of Jaw1 probably functions as an interaction site for the oligomerization.
The Jaw1 N-terminal region inhibits the formation of organized smooth ER. Co-IP assays showed that coiled-coil domain of Jaw1 functions as a candidate for the oligomerization site and that other cytosolic regions have no affinity for oligomerization. In order to investigate the intracellular function of each region, we prepared for several new deletion mutants of Jaw1 that lacked each region: FLAG Ms Jaw1 ΔN, FLAG Ms LJaw1 ΔCoil and FLAG Ms LJaw1 ΔStem ( Fig. 2A). These mutants were expressed in B16F10 cells and the intracellular localization was then observed. Interestingly, aggregate-like structures in the cytosol and along the perinuclear membranous region were observed in nearly all of the cells that were expressing Jaw1 ΔN, a mutant lacking the N-terminal region, whereas LJaw1 and SJaw1 were localized at the conventional ER (Fig. 2B,C). These abnormal structures were observed in more detail by transmission electron microscopy. As shown in electron micrographs, these membranous structures were composed of highly dense stacked membranes, which are referred to as organized smooth ER (OSER) such as whorls (Fig. 2D, box 1) and nuclear karmellae (Fig. 2D, box 2). While whorls structure was observed in all analyzed OSER positive cells (N = 8), nuclear karmellae was observed in six out of eight cells. OSER has been reported to be caused by the ectopic overexpression of some ER resident proteins that undergo abnormal oligomerization on the ER membrane. Taking Jaw1 oligomeriza- Identification of the oligomerization site by co-immunoprecipitation followed by western blotting. FLAG Ms LJaw1 was co-expressed with GFP, GFP Ms LJaw1 or mutants in HEK293 cells by transfection. After incubation for 24 h, the lysates were subjected to co-immunoprecipitation. For western blotting, an anti-FLAG rabbit antibody and an anti-GFP rabbit antibody as primary antibodies were used. (C) GFP, GFP Ms LJaw1 or mutants was expressed alone or co-expressed with FLAG Ms LJaw1 in B16F10 cells by transfection. After incubation for 24 h, the cells were subjected to immunostaining using an anti-FLAG rabbit antibody as a primary antibody. The images were acquired by confocal microscopy. Scale bar; 20 μm.  (Fig. 2C). This result indicates that the structural exposure of the coiled-coil domain by the deletion of the N-terminal region caused the OSER formation. In order to exclude the possibility that this OSER formation is due to the effects of the N-terminal tag, Ms Jaw1 ΔN without the tag was expressed in B16F10 cells. As a result, the cells expressing Ms Jaw1 ΔN formed significant amounts of the OSER, indicating that this OSER formation can be attributed to the deletion of the N-terminal region (see Supplementary Fig. S2A,B online). Furthermore, this OSER formation was also detected in cells that were expressing the FLAG tagged human Jaw1 ΔN (see Supplementary Fig. S3A-C online). In summary, the N-terminal region has a role in intrinsically inhibiting OSER formation in both of human and mouse Jaw1.  www.nature.com/scientificreports/ The entire Jaw1 N-terminal region inhibits OSER formation. Although at least two splicing variants of Jaw1 depending on the translation initiation site, i.e. LJaw1 and SJaw1 26 , are currently known, there was no significant differences in OSER formation between them in both mouse and human Jaw1 (Fig. 2B,C, see Supplementary Figs. S2 and S3 online). These results imply that OSER formation in cells expressing Jaw1 ΔN is due to the loss of the N-terminal region . Here, in order to further investigate the issue of whether a more specific region within the N-terminal region is involved in the inhibition of OSER formation, we prepared some deletion mutants that lacked the limited N-terminal region; Ms LJaw1 ΔsN1, Ms LJaw1 ΔsN2 and Ms LJaw1 ΔsN3 and compared the extent of OSER formation with cells that were expressing Ms LJaw1, Ms SJaw1 and Ms Jaw1 ΔN (Fig. 3A). In the cells expressing Ms LJaw1 ΔsN1 and Ms LJaw1 ΔsN2, a slightly higher percentage of the OSER formation was observed compared to Ms LJaw1 and Ms SJaw1 (Fig. 3B,C). However, in the cells expressing these mutants, including Ms LJaw1 ΔsN3, the percentage of OSER formation was not as high as Ms Jaw1 ΔN. This result prompted us to consider a new hypothesis, namely, that the entire N-terminal region  of Jaw1 has a role in inhibiting OSER formation. To verify this, we prepared some new deletion mutants in which the N-terminal region had different lengths; Ms Jaw1 ΔN1 and Ms Jaw1 ΔN2 and compared the OSER formation with LJaw1, SJaw1 and Jaw1 ΔN (Fig. 3A). As expected, the percentage of OSER positive cells was high in the order; Jaw1 ΔN, Jaw1 ΔN2, Jaw1 ΔN1 and LJaw1 or SJaw1, indicating that the shorter the N-terminal region is, the higher percentage contains OSER formation (Fig. 3D, E). Thus, these data indicate that the entire N-terminal region is involved in the inhibition of OSER formation.
Jaw1 N-terminal region inhibits its irregular oligomerization via coiled-coil domain. As shown in Fig. 2B and C, in the cells expressing the FLAG Ms Jaw1 ΔN Coil, the percentage of OSER formation was significantly lower than FLAG Ms Jaw1 ΔN, compatible with FLAG Ms LJaw1 and FLAG Ms SJaw1. This result prompted us to hypothesize that the Jaw1 N-terminal region intrinsically inhibits irregular oligomerization, thus preventing the structural exposure of the coiled-coil domain. To verify this, we first investigated whether or not the addition of a GFP tag as a larger structural tag to the N-terminal of Ms Jaw1 ΔN inhibited OSER formation.
The results showed that OSER formation in the cells expressing GFP Ms Jaw1 ΔN was still detected, with significant differences from GFP Ms LJaw1 and GFP Ms SJaw1 (Fig. 4A,B). These data indicate that the Jaw1 N-terminal region itself is essential for the inhibition of OSER formation. Furthermore, the amino acids sequence of mouse LJaw1 and human LJaw1 were subjected to an automated protein structure prediction: SWISS-MODEL, but the structural prediction of the regions, except for the coiled-coil domain, were not available. In order to examine the amino acids sequence further, the amino acid sequence of mouse LJaw1 and human LJaw1 were subjected to a DICHOT analysis, an established system for exploring IDR within a protein 35 . Intriguingly, the computational analysis showed, in both of human and mouse Jaw1, that the N-terminal region, the stem region and the luminal region are IDRs, whereas the coiled-coil domain and the transmembrane domain are structural regions (Fig. 4C). Furthermore, another computational analysis, D 2 P 2 platform, also predicted that both the N-terminal region, stem region and the luminal region are IDRs (see Supplementary Fig. S4 online). To verify that N-terminal region is an IDR, the N-terminal region (Ms LJaw1 N) was produced in E. coli as an N-terminal maltose binding protein (MBP) tagged protein (MBP Ms LJaw1 N). The MBP Ms LJaw1 N was purified by affinity chromatography using an amylose resin and the MBP tag was digested with TEV protease. After the digestion, MBP and Ms LJaw1 N were separated by anion exchange chromatography and affinity chromatography using amylose resin (see Supplementary Fig. S5A online). Each recombinant protein was then subjected to circular dichroism (CD) spectroscopy. The CD spectra of Ms LJaw1 N showed the typical curve of random coil region characterized with a negative value around 200-210 nm and no distinguished change around 210-250 nm compared to that of MBP as a control of structured protein, which suggests that the N-terminal region exists as an IDR (see Supplementary Fig. S5B online) 36,37 . Next, to investigate how the N-terminal region functions as an IDR on the oligomerization, Native-PAGE was carried out using a Ms LJaw1 N Coil bearing both the N-terminal region and the coiled-coil domain and Ms Jaw1 Coil with the coiled-coil domain alone. Here, these recombinant proteins were produced in E. coli as N-terminal MBP tagged proteins and were purified by affinity chromatography using an amylose resin (Fig. 4D). The Coomassie Brilliant Blue (CBB) staining images following SDS-PAGE showed that both the purified MBP tagged proteins were obtained in high purity ( online). However, both the MBP Ms Jaw1 Coil and MBP cut Ms Jaw1 Coil showed multiple bands. This result indicates that the loss of the N-terminal region changed the oligomeric states of Jaw1. In summary, these data suggest that the N-terminal region has a role in maintaining the ordered oligomerization of Jaw1 as an IDR, thus preventing the structural exposure of the coiled-coil domain, an oligomerization site. We therefore speculate that the loss of the Jaw1 N-terminal region causes the irregular oligomerization on the membrane, resulting in OSER formation in the cell. Furthermore, the result of multiple bands in the Jaw1 Coil ( Fig. 4F) also implies that the coiled-coil domain is one of the direct interaction sites among Jaw1 molecules. This is consistent with the result of the co-immunoprecipitation assay in Fig. 1B. The OSER derived from Jaw1 ΔN separated from conventional ER network. The effects of OSER on the cell would depend on the origins of its formation. For example, the OSER derived from the overexpression of Stbd1 recruits the interactor and changes the morphology of the ER-mitochondria network 18 . Furthermore, the OSER derived from the variant of TorsinB, an ATPases associated with diverse cellular activities, also recruits several ER-NE resident proteins such as sec61, calnexin and emerin but not protein-disulfide isomerase www.nature.com/scientificreports/ www.nature.com/scientificreports/ (PDI), LaminB1 and SUN2 24 . In order to characterize the OSER derived from Jaw1 ΔN, the effects on the localization of ER resident proteins were first evaluated. Here, in cells expressing Ms LJaw1, Ms SJaw1 or Ms Jaw1 ΔN, the localization of the ER markers; calnexin or calreticulin were observed. The confocal images and line plot profiles showed that both Ms LJaw1 and Ms SJaw1 are co-localized with calnexin nearly and calreticulin strongly within the conventional ER network (Fig. 5A,B). Furthermore, Ms Jaw1 ΔN outside the OSER was also co-localized with calnexin and calreticulin, similar to Ms LJaw1 and Ms SJaw1. However, in the compartment of OSER derived from Ms Jaw1 ΔN, calnexin and calreticulin were not recruited to the OSER at the same high level as the Ms Jaw1 ΔN. These data indicate that the OSER derived from Ms Jaw1 ΔN is likely to be a compartment that is separated from the conventional ER network and does not affect the conventional ER morphology. Furthermore, the morphology of the Golgi apparatus and mitochondria were evaluated by staining using an anti-GM130 antibody and Mito Tracker, respectively. As shown in Supplementary Fig. S6 online, the morphology and positioning were normal in the case of the OSER positive cells expressing Ms Jaw1 ΔN, the same as the cells expressing Ms LJaw1 and Ms SJaw1. Thus, the existence of OSER derived from Ms Jaw1 ΔN does not affect the morphology of these organelles.
The OSER derived from Jaw1 ΔN caused the localization of interactors to be altered. According to previous reports, OSER often recruits interactors to this structure 18,24 . Jaw1 has been reported to interact with IP 3 R3 localized at the ER via coiled-coil domain 30 . Furthermore, we previously reported that Jaw1 interacts with SUN proteins that are localized at the INM 28 . Therefore, we investigated the issue of whether or not these proteins are recruited to the OSER derived from Ms Jaw1 ΔN. Here, FLAG tagged IP 3 R3 (FLAG IP 3 R3) or SUN2 (FLAG SUN2) were co-expressed with Ms LJaw1, Ms SJaw1 or Ms Jaw1 ΔN in B16F10 cells and co-immunostaining using an anti-FLAG antibody and an anti-Jaw1 antibody was carried out. The confocal images and line plot profiles showed that FLAG IP 3 R3 was nearly co-localized with Jaw1 in the cells expressing Ms LJaw1 and Ms SJaw1 (Fig. 6A). Importantly, in cells expressing Ms Jaw1 ΔN, FLAG IP 3 R3 is recruited to the OSER. The counting of cells with Jaw1 + IP 3 R3 + OSER out of Jaw1 + cells showed that the cells expressing Ms Jaw1 ΔN formed significant amounts of OSER compared to that for Ms LJaw1 and Ms SJaw1 (see Supplementary Fig. S7 online). Furthermore, FLAG SUN2 were localized at the INM along with NE in cells expressing Ms LJaw1 and Ms SJaw1, but it was recruited to the OSER and the localization was disrupted due to the formation of nuclear karmellae in the cells expressing Ms Jaw1 ΔN (Fig. 6B). In summary, the OSER derived from Ms Jaw1 ΔN altered the localization of interactors such as IP 3 R3 and SUN2.
The OSER derived from Jaw1 ΔN altered the localization of the oligomer. As shown in Fig. 6, the interactors are recruited to the OSER and the localization is disrupted. This led us to the hypothesis that the expression of Jaw1 ΔN also disrupts the formation and localization of the oligomer. Therefore, FLAG Ms LJaw1 or FLAG Ms SJaw1 was co-expressed with PA tagged Ms LJaw1 (PA Ms LJaw1) or Ms Jaw1 ΔN (PA Ms Jaw1 ΔN) and co-immunostaining using an anti-FLAG antibody and an anti-PA antibody was carried out. The confocal images and line plot profiles showed that FLAG Ms LJaw1 or FLAG Ms SJaw1 was almost co-localized with PA Ms LJaw1 in cells expressing these proteins (Fig. 7A). As expected, in the cells co-expressing PA Ms Jaw1 ΔN with FLAG Ms LJaw1 or FLAG Ms SJaw1, FLAG Ms LJaw1 or FLAG Ms SJaw1 were recruited to the OSER (Fig. 7B). Thus, the OSER derived from Ms Jaw1 ΔN altered the localization of the oligomer.

Discussion
In this study, we performed functional analyses of Jaw1 cytosolic regions in terms of its oligomerization and characterized the role of the N-terminal region and coiled-coil domain. Co-immunoprecipitation experiments showed that the coiled-coil domain probably functions as an oligomerization site. On the other hand, the loss of the Jaw1 N-terminal region leads to OSER formation. According to the computational analysis using the DICHOT system and D 2 P 2 platform and CD spectroscopy, the N-terminal region of Jaw1 exists as an IDR, and the loss of the Jaw1 N-terminal region causes irregular oligomerization in vitro, which probably results in OSER formation. Furthermore, this OSER recruits the interactors, IP 3 R3, SUN2 and full-length Jaw, but does not affect the localization of the conventional ER resident proteins, calnexin and calreticulin. In summary, these data suggest that the inhibition of OSER formation by the N-terminal region of Jaw1 as an IDR is essential for maintaining the homeostatic localization of Jaw1 oligomer and interactors on the ER membrane.
To date, it has been reported that OSER formation is frequently due to the overexpression of ER resident proteins [16][17][18][19][20][21][22][23][24][25] . Some of them are due to the structural effects by the addition of the oligomerized tags, others are derived from mutations such as TorsinB E178Q and VAPB P56S 24,25 . The OSER formation reported in this study is due to the expression of Jaw1 lacking its N-terminal region, but not the overexpression of the full-length form of Jaw1, which suggests that the function of N-terminal region is to inhibit OSER formation as an intrinsic regulator.
Furthermore, our data show that the loss of the N-terminal region caused OSER formation in both of human and mouse Jaw1. Intriguingly, a comparison of amino acid sequences between them showed that the percentage identity of the N-terminal region is relatively low (43.6%) compared to remainder of the region (71.9-92.9%) (see Supplementary Fig. S8, Table S1 online). Furthermore, a computational analysis indicated that both the N-terminal regions are IDR. The IDR is characterized by the higher polarity and lower hydrophobicity, which causes the regions to be in the unstable conformational state [38][39][40] . Consistently, both N-terminal regions have a lower hydrophobicity compared to structured regions; coiled-coil domain and transmembrane domain (see Supplementary Table S2 online). In addition, they also have a bigger difference between the percentage of negatively and positively charged amino acids which means there is a higher polarity compared to other regions (see Supplementary Table S2 online). Thus, these data suggest that the disorder with above features in the amino acid sequence which does not create a stable structure, in contrast to the conserved amino acids, provides the www.nature.com/scientificreports/ common functionality for both N-terminal regions. Furthermore, in the experiment using mutants lacking the limited N-terminal region, the percentage of OSER formation in the cells expressing Jaw1 ΔsN3 is less than Jaw1 ΔsN1 and Jaw1 ΔsN2, although all these mutants have a much lower percentage than Jaw1 ΔN. In the comparison of the amino acid composition in each N-terminal region (N1-N3), N3 region showed a relatively lower difference between the percentage of negatively and positively charged amino acids compared to N1 and N2 (see Supplementary Table S3 online). Although further wet-experiment is required for this verification, this lower polarity in N3 region would weaken the functionality as an IDR to inhibit the OSER formation compared to N1 and N2 regions. However, the most important factor for the function of the N-terminal region to inhibit the OSER formation seems to be the N-terminal length as shown in Fig. 3E. Previous investigations reported that IDR regulates the protein-protein interaction and the intramolecular binding 38,41 . It has been also reported that IDR is involved in the regulation of precise oligomerization 42 . It is noteworthy that an in vitro assay using Jaw1 with/without the N-terminal region showed that the N-terminal region regulates the oligomeric states, which means that the irregular oligomeric state due to the lack of N-terminal region results in OSER formation. Furthermore, interestingly, the expression of Jaw1 ΔN Coil lacking N-terminal region and the coiled-coil domain repressed OSER formation. We therefore suggest that the N-terminal region of this protein has a role in inhibiting OSER formation as an IDR, thus preventing the structural exposure of coiled-coil domain, an oligomerization site. In this point, this study also supports the novel functional concept of IDR in the intrinsic regulation of oligomerization, as shown in the previous report 42 . In addition, Jaw1 ΔN, which probably causes the aberrant oligomerization, dominantly recruits the full-length Jaw1 into the OSER as shown in Fig. 7. This result perhaps indicates that the N-terminal region of full-length Jaw1 interferes with each other in its oligomer to inhibit the aberrant oligomerization at intramolecular level.
In this study, we uncovered that the coiled-coil domain is a candidate for the oligomerization site. However, we could not directly exclude the possibility that the regions except coiled-coil domain are oligomerization sites. Furthermore, a truncation mutant lacking transmembrane domain only appeared as a weak band (Fig. 1B, in the lane of co-immunoprecipitated Jaw1 ΔTM), which might mean there is a low affinity for oligomerization. We speculate the reason for this is that the oligomer consisting of Jaw1 without transmembrane domain might be unstable. Thus, we hypothesized that the insertion into the membrane is also an important factor for the precise oligomerization. On the other hand, a truncation mutant coding for only coiled-coil domain showed a strong band compatible with the full-length Jaw1 although this mutant also lacks transmembrane domain (Fig. 1B, in the lane of co-immunoprecipitated Jaw1 Coil). This would be due to the aberrant oligomerization by the loss of N-terminal region. These data indicate the possibility of additional oligomerization sites or factors for precise oligomerization, although we need further quantitative analysis.
A morphological characteristic of OSER structures derived from Jaw1 ΔN is the presence of a nuclear karmellae, where the membranes are stacked along the nuclear envelope. The reason for the expression of the Jaw1 ΔN forms this structure is probably due to the localization of Jaw1 at the ONM, as we previously reported 28 . Furthermore, this OSER recruits SUN2 and disrupts the localization at nuclear membrane in B16F10 cells, but has no dominant negative effects on the nuclear shape. In the OSER derived from Jaw1 ΔN, IP 3 R3, another interactor of Jaw1, is also recruited. IP 3 R3 releases calcium ions into the cytosol in response to an extracellular signal 43 . Furthermore, the transfer of calcium ions to mitochondria via an ER-mitochondria contact site is also involved with the regulation of apoptosis 43 . Although it has been reported that Jaw1 binds to IP 3 R3 via its coiled-coil domain, whether Jaw1 regulates the activity of IP 3 R3 remains uncovered 30 . Therefore, an evaluation of the dominant negative effects on the activity of IP 3 R3 in OSER derived from Jaw1 ΔN will be a future issue in elucidating the functional relation between Jaw1 and IP 3 R3.
In this study, we mainly focused on the cytosolic region and uncovered the physical properties of the Jaw1 N-terminal region to inhibit OSER formation. On the other hand, a few types of proteins that form OSER have been reported to be involved with certain diseases. For example, the expression of mutants such as TorsinA and VAPB are related to dystonia and amyotrophic lateral sclerosis, respectively 23,25 . According to a previous report, a mutational sequence analysis between a non-obese diabetes model (NOD/Lt) and a non-diabetic model (C57BL/6, C3H/HeJ and BALB/c) provided some candidate mutations of mouse Jaw1 that are related to type I diabetes such as G6D, L145P, duplication of three amino acids [STL (452-455)] 44 . Intriguingly, all of the mutated positions are located in the N-terminal region or stem region. Furthermore, some of the residues are www.nature.com/scientificreports/ also conserved in humans (see Supplementary Fig. S8 online). Taking the existence of the N-terminal region and stem region as IDRs in both of human and mouse Jaw1 as indicated by the computational analysis into consideration, these IDRs of Jaw1 would regulate the functions intrinsically. Thus, future work regarding to the issue of whether these mutations of Jaw1 affect the functions such as the ability to form oligomers and protein-protein interactions will promote the understanding how Jaw1 is involved in the pathogenesis. In contrast to the abnormal effects leading to OSER formation, it is not clear whether the OSER is also formed under normal physiological conditions. Here, no significant difference in OSER formation between two variants (LJaw1 and SJaw1) was observed. However, interestingly, the national center for biotechnology information (NCBI) database shows the existence of a transcript variant (NM_001361633) coding for the amino acid sequence corresponding to Ms Jaw1 ΔN1, which would be a variant with a shorter N-terminal region and would form the OSER at slightly higher percentage than LJaw1 and SJaw1, as shown in Fig. 3E. Furthermore, it has been reported that the variant corresponding to Ms Jaw1 ΔN in addition to LJaw1 and SJaw1 is expressed in vitro translational system using the cDNA encoding full-length Jaw1 45 . These data suggest the possibility of OSER formation derived from an alternative splicing variant from the Jaw1 gene under normal physiological conditions. Therefore, to find tissues and cells that express this variant specifically will also provide us with an understanding of the biological significance of the OSER.

Materials and methods
Plasmids. The primer sets used to produce the plasmids below are shown in Supplementary Table S4 online. pcDNA5 FRT/TO HA FLAG Ms LJaw1 was produced from a total cDNA derived from mouse (C57/B6J) spleen as previously described 28 . From the same library, a DNA fragment coding Ms SJaw1 (NM_001368864.1) was amplified by PCR with primer set 1 and the PCR product was subcloned into the pcDNA5 FRT/TO HA FLAG vector (digested with KpnI/BamHI) using an In-Fusion HD cloning Kit (TaKaRa, Z9648), resulting in pcDNA5/ FRT/TO HA FLAG Ms SJaw1. These plasmids were digested with KpnI/BamHI and the fragment was ligated into the pTagGFP2-C vector (Evrogen) (digested with the same enzymes), resulting in pTagGFP2-C Ms LJaw1 and SJaw1, respectively. To produce pTagGFP2-C Ms LJaw1 ΔTM coding for entire cytosolic regions, the PCR product amplified from pcDNA5FRT/TO HA FLAG Ms LJaw1 with primer set 2 was inserted into the pTag-GFP2-C vector (digested with EcoRI/SalI) using an In-Fusion HD cloning Kit. Furthermore, for the generation of the plasmids encoding Jaw1 each cytosolic region, the PCR products amplified from pTagGFP2-C Ms LJaw1 with primer sets 3, 4 and 5 were inserted into the pcDNA3.1(+) vector (Invitrogen) (digested with BamHI/XhoI) using an In-Fusion HD cloning Kit. These plasmids were digested with KpnI/ApaI and each fragment was then ligated into the pTagGFP2-C vector (digested with the same enzymes), resulting in pTagGFP2-C Ms LJaw1 N, Jaw1 Coil and Jaw1 Stem, respectively. For the generation of deletion mutants: pcDNA5 FRT/TO HA FLAG Ms Jaw1 ΔN and ΔN Coil, the PCR products amplified from pTagGFP2-C Ms LJaw1 using primer sets 6 and 7, respectively, were inserted into the pcDNA5/FRT/TO HA FLAG vector (digested with KpnI/BamHI and BamHI, respectively) using an In-Fusion HD cloning Kit. Furthermore, pcDNA5 FRT/TO HA FLAG Ms Jaw1 ΔN was digested with KpnI/BamHI and the fragment was ligated into the pTagGFP2-C vector, resulting in pTagGFP2-C Ms Jaw1 ΔN. To produce pcDNA5 FRT/TO HA FLAG Ms LJaw1 ΔCoil and ΔStem, pcDNA5 FRT/TO HA FLAG Ms LJaw1 was performed by inverse PCR to remove the region coding for the coiled-coil domain or stem region using 5′-phosphorylated primer sets 8 and 9, respectively, and self-ligated. To remove the possible effects of the N-terminal tag, pcDNA5 FRT/TO HA FLAG Ms LJaw1 and pcDNA5 FRT/TO HA FLAG Ms SJaw1 was digested with KpnI/BamHI and each fragment was ligated into the pcDNA3.1(+) vector (digested with the same enzymes), resulting in the pcDNA3.1(+) Ms LJaw1 and SJaw1, respectively. In addition, for the generation of pcDNA3.1(+) Ms Jaw1 ΔN, the PCR product amplified from pTagGFP2-C Ms LJaw1 with primer set 10 was inserted into the pcDNA3.1(+) vector (digested with BamHI/XhoI) using an In-Fusion HD cloning Kit. To produce the limited N-terminal region-truncated mutants, pcDNA3.1(+) Ms LJaw1 was performed by inverse PCR to remove the region coding for each limited N-terminal region using primer sets 11, 12 and 13, resulting into pcDNA3.1(+) Ms LJaw1 ΔsN1, ΔsN2 and ΔsN3. For the generation of pcDNA3.1(+) Ms LJaw1 ΔN1 and ΔN2, the PCR products were first amplified from pTagGFP2-C Ms LJaw1 using primer sets 14 and 15, respectively, and inserted into the pcDNA5 FRT/TO HA FLAG vector (digested with BamHI/XhoI) using an In-Fusion HD cloning Kit. These plasmids were then digested with BamHI/XhoI and each fragment was ligated into the pcDNA3.1(+) vector (digested with the same enzymes). To generate pcDNA3.1(+) PA Ms LJaw1 and Jaw1 ΔN, pcDNA3.1(+) PA vector was first prepared by the insertion of DNA cassette encoding PA tag (primer set 16) into the pcDNA3.1(+) vector (digested with NheI/HindIII). pcDNA3.1(+) Ms LJaw1 and Jaw1 ΔN were then digested with KpnI/BamHI and KpnI/XhoI, respectively, and each fragment was ligated into the pcDNA3.1(+) PA vector. To produce the pMALcRI TEV Ms LJaw1 N, LJaw1 N Coil and the Jaw1 Coil, the PCR products amplified from pTagGFP2-C Ms LJaw1 with primer sets 3, 17 and 4, respectively, were inserted into the pcDNA3.1(+) vector (digested with BamHI/XhoI) using an In-Fusion HD cloning Kit. These plasmids were digested with BamHI/XhoI and each fragment was then ligated into the pMALcRI TEV vector (previously described 28 ) (digested with BamHI/SalI), resulting in pMALcRI TEV Ms LJaw1 N, LJaw1 N Coil, Jaw1 Coil, respectively. For the generation of the pcDNA5 FRT/TO HA FLAG Hu LJaw1, SJaw1 and Jaw1 ΔN, the PCR products were amplified from a total cDNA derived from Burkitt's lymphoma cell line (Raji) with primer sets 18, 19 and 20 and inserted into the pcDNA3.1 (+) vector (digested with BamHI/XhoI) using an In-Fusion HD cloning Kit. pcDNA5 FRT/TO HA FLAG Ms SUN2 was produced from EGFP-C1 SUN2 (kindly provided by Howard, J. Worman 46 ) as previously described 28 . pCMV6 Myc-DDK-tagged Mouse IP 3 R3 (NM_08053) was purchased from Ori Gene (MR225699). The identity of cloned sequences was verified by the DNA sequence contract service (eurofin). After confirmation, each plasmid was prepared using the Plasmid Mini Kit (Qiagen, 12123) and used for following experiments. www.nature.com/scientificreports/ Cells and cell culture. B16F10 and HEK293 cells were incubated in MEM supplemented with 10% fetal bovine serum (FBS) and 5.84 mg/mL of L -glutamine and in DMEM containing 10% FBS, 100 U/mL penicillin, 100 μg/mL streptomycin (Sigma) and 5.84 mg/mL l-glutamine, respectively, and maintained at 37 °C in 5% CO2.
Transfection. Plasmids were introduced into B16F10 or HEK293 cells using ScreenFect TM A plus (FUJIFILM Wako Pure Chemical, 299-77103) according to the manufacturer's instructions. After the treatment for 24 h, the cells were used for the following assays.
Immunostaining. B16F10  Transmission electron microscopy. For observing OSER structures using transmission electron microscopy, the samples were prepared as previously described 47  www.nature.com/scientificreports/ www.nature.com/scientificreports/ in washing buffer (10 mM Tris-HCl pH 7.4 and 200 mM NaCl). The suspensions were sonicated on ice and centrifuged at 12,000 rpm for 30 min. The supernatants were performed for purification using an amylose resin (New England Biolabs, E8021). MBP tag was removed by the treatment of TEV proteases (0.58 mg/mL) for 16 h at 4 °C. The purified recombinant proteins were mixed with SDS-PAGE sample buffer, heat blocked at 95 °C for 5 min and subjected to SDS-PAGE. The resulting purified proteins were mixed with Native-PAGE sample buffer and subjected to Native-PAGE. The gels were then subjected to CBB staining. For the preparation of the samples for the circular dichroism spectrometry, BL21(DE3) transformed with pMALcRI TEV Ms LJaw1 N was cultured in 1 L LB medium and the expression of the MBP Ms LJaw1 N was induced by treatment with 1 mM isopropyl β-d-1-thiogalactopyranoside for 4 h at 25 °C. The culture media was centrifuged at 8000 rpm for 10 min and the pellet was resuspended in washing buffer. The suspension was sonicated on ice and centrifuged at 12,000 rpm for 30 min. The supernatant was purified using an amylose resin. Purified MBP Ms LJaw1 N was treated with TEV proteases (0.58 mg/mL) for 16 h at 30 °C to remove the MBP tag, resulting to MBP tag and Ms LJaw1 N. After digestion, the His tagged TEV proteases was removed by the affinity chromatography using Ni-NTA Agarose (Qiagen, 1018244). The solution containing MBP tag and Ms LJaw1 N was concentrated with Amicon Ultra-15 Centrifugal Filter (Millipore, UFC901008), desalted in 20 mM Tris-HCl pH 8.0 and each protein was separated by anion exchange chromatography [HiTrap Q FF (GE Healthcare, 17515601)] on FPLC ÄKTAPurifier system (GE Healthcare). A gradient of 0 to 50% of buffer B (20 mM Tris-HCl pH 8.0 and 1 M NaCl) was applied. The fractions containing Ms LJaw1 N were concentrated with Amicon Ultra-15 Centrifugal Filter, desalted in 20 mM Tris-HCl pH 8.0 and subjected to anion exchange chromatography followed by affinity chromatography using amylose resin to remove the MBP tag or MBP tagged proteins. Each solution containing MBP or Ms LJaw1 N was concentrated with Amicon Ultra-15 Centrifugal Filter and desalted in 20 mM Tris-HCl pH 8.0. After that, the samples were mixed with SDS-PAGE sample buffer, heat blocked at 95 °C for 5 min and subjected to SDS-PAGE followed by CBB staining.
Circular dichroism spectroscopy. CD spectroscopy was performed as previously described 48 . Briefly, purified MBP or Ms LJaw1 N were dissolved in 20 mM Tris-HCl pH7.4, 200 mM NaCl at a concentration of 0.1 mg/mL. CD spectra were recorded on a J-720WI spectrophotometer (Jasco GmbH) using at a 0.1 cm optical path, 0.1 nm interval, 1.0 nm bandwidth and 50 nm/min scanning speed. The spectra were recorded five times followed by averaging and background subtraction. The HT voltage was below 700 V over the entire range of recording (200-250 nm).
Statistical analysis. The collected data were analyzed and graphically presented using GraphPad Prism7 (GraphPad). Statistical significance was determined by one-way ANOVA followed by Tukey Kramer's t-test.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. www.nature.com/scientificreports/