Protein export from the endoplasmic reticulum (ER) to the Golgi apparatus occurs at specialized regions known as the ER exit sites (ERES). In Saccharomyces cerevisiae, ERES appear as numerous scattered puncta throughout the ER. We examined ERES within the peripheral ER, finding that the proteins comprising the ERES localize on high-curvature ER domains where curvature-stabilizing protein Rtn1 is present. Δrtn1 Δrtn2 Δyop1 cells have fewer high-curvature ER domains, but ERES accumulate at the remaining high-curvature ER domains on the edge of expanded ER sheets. We propose that membrane curvature is a key geometric feature for the regulation of ERES localization. We also investigated a spatial relationship between ERES and Golgi cisternae. Golgi cisternae in S. cerevisiae are unstacked, dispersed, and moving in the cytoplasm with cis-cisternae positioned adjacent to ERES, whereas trans-cisternae are not. Morphological changes in the ER of Δrtn1 Δrtn2 Δyop1 cells resulted in aberrant Golgi structures, including cis- and trans-markers, and there was reduced movement at ERES between expanded ER sheets and the plasma membrane.
The secretory pathway conveys a large number and wide variety of proteins as cargo to their final destinations, such as the extracellular space and the plasma membrane. The ER is the starting organelle of the pathway, and the Golgi apparatus acts as its pivotal sorting station. Recent studies on the budding yeast Saccharomyces cerevisiae have revealed that the Golgi apparatus is very dynamic in its structure, and the compartments of the Golgi change from cis-cisternae to trans-cisternae over time (Losev et al., 2006; Matsuura-Tokita et al., 2006). Though these observations provide strong support for the cisternal maturation model, they also raise new questions as to how new Golgi are generated and the cargo molecules transported into cis-Golgi (Emr et al., 2009; Glick and Nakano, 2009; Nakano and Luini, 2010).
ER-to-Golgi transport is mediated by coat protein complex II (COPII) vesicles. Components responsible for COPII vesicle formation are well conserved between yeast and mammals (Kuge et al., 1994; Swaroop et al., 1994; Paccaud et al., 1996; Tang et al., 1999; Tang et al., 2000; Weissman et al., 2001; Bhattacharyya and Glick, 2007) and include COPII coat protein subunits Sec23, Sec24, Sec13, and Sec31, a small GTPase Sar1, and its specific guanine nucleotide exchange factor (GEF) Sec12 (Nakano et al., 1988; Nakano and Muramatsu, 1989; Barlowe and Schekman, 1993; Barlowe et al., 1994). When activated by Sec12, Sar1–GTP initiates COPII vesicle formation on the ER by sequentially recruiting Sec23/24 heterodimers and Sec13/31 heterodimers (Lee et al., 2004). Cell-free experiments using synthetic liposomes, proteoliposomes, and a planar lipid bilayer have shown that Sec23/24, Sec13/31, and Sar1–GTP are sufficient for formation of the COPII vesicle, but multiple rounds of the Sar1 GDP/GTP cycle stimulated by Sec12 are pivotal for efficient cargo selection (Matsuoka et al., 1998). Another key component of COPII vesicle formation is a peripheral membrane protein, Sec16, which directly interacts with Sec23, Sec24, and Sec31 (Espenshade et al., 1995; Gimeno et al., 1996; Shaywitz et al., 1997). Sec16 facilitates the recruitment of COPII coat proteins on liposomes and stabilizes the coat to prevent premature disassembly (Supek et al., 2002). Thus, Sec16 is considered to function as a scaffold for assembling COPII coats (Supek et al., 2002; Connerly et al., 2005).
Previous studies have shown that COPII vesicles are formed at the specialized sites within the ER, termed the ER exit sites (ERES) or the transitional ER. ERES are the sites where cargo and COPII coat proteins are concentrated, and are morphologically distinct from the surrounding ER (Palade, 1975; Orci et al., 1991; Bannykh et al., 1996). Although COPII vesicle formation has been characterized in detail, the structure and organization of ERES remain to be elucidated. In mammalian cells, COPII components are concentrated in hundreds of punctate structures along the ER, enriched at the juxtanuclear region where stacked Golgi exists (Orci et al., 1991; Bannykh et al., 1996; Stephens, 2003). Punctate ERES structures are found adjacent to the ER-to-Golgi intermediate compartment (ERGIC). Localization of Sec23/24, Sec13/31, and Sec16 is limited at ERES, whereas Sar1 is localized throughout the ER with some accumulation at ERES (Watson et al., 2006). Sec12 is localized uniformly within the entire ER (Weissman et al., 2001). For budding yeast species, the Golgi cisternae of Pichia pastoris form a stacked structure, whereas those of S. cerevisiae are unstacked and dispersed in the cytoplasm (Orci et al., 1991; Bannykh et al., 1996; Rossanese et al., 1999; Bevis et al., 2002; Stephens, 2003). P. pastoris has a small number of ERES (two to six per cell), each juxtaposed to the stacked Golgi (Rossanese et al., 1999; Bevis et al., 2002). Sec12 and Sar1 accumulate at ERES with COPII coats and Sec16 (Soderholm et al., 2004). The structural properties of S. cerevisiae ERES were unclear until recently. Live cell imaging demonstrated that S. cerevisiae also has organized ERES that consist of numerous punctate structures marked by COPII coat proteins (Castillon et al., 2009; Levi et al., 2010; Shindiapina and Barlowe, 2010). Sar1 and Sec12 are reported to localize throughout the ER (Nishikawa and Nakano, 1991; Rossanese et al., 1999). The structural differences between the ERES of P. pastoris and S. cerevisiae might be reflected in the discrepant features of their Golgi cisternae.
In mammalian cells and plant cells, punctate ERES are present on the surface of cortical ER tubules (Hammond and Glick, 2000; Yang et al., 2005). The ERES of budding yeast are found on the nuclear envelope and the peripheral ER (Bevis et al., 2002; Shindiapina and Barlowe, 2010), but their precise distribution on the ER membrane has not been revealed. Because the ER has an elaborate shape consisting of a network of membrane-enclosed tubules and sheets, the distribution of ERES might depend on the ER morphology. Although little is known about how the ER structures are generated and maintained, recent studies identified two protein families intimately involved in shaping the ER. Reticulons (Rtns) and DP1/Yop1 are curvature-stabilizing proteins that localize in the ER tubules and at the edge of ER sheets (De Craene et al., 2006; Voeltz et al., 2006; Shibata et al., 2010). Overexpression of Rtns generates longer unbranched tubules and fewer sheets, whereas depletion of Rtns results in the proliferation of sheet regions at the expense of tubules (Voeltz et al., 2006; Anderson and Hetzer, 2008). Atlastin family of proteins, including S. cerevisiae Sey1, have also been implicated in organization of tubular network of the ER (Farhan and Hauri, 2009; Hu et al., 2009).
We conducted high-resolution live imaging of ERES in S. cerevisiae for the seamless understanding of the ER-to-Golgi system. In the present study we report that S. cerevisiae ERES consist of COPII coat proteins and Sec16 and are preferentially distributed on high-curvature domains of the ER membrane, ER tubules and the edge of ER sheets. Morphological changes of the ER affected the distribution of ERES, but they were still present on the high-curvature regions of the ER. Correlation analysis revealed that cis-Golgi cisternae, but not trans-Golgi cisternae, were juxtaposed to ERES. Remarkably, ectopic localization of the proteins comprising the ERES resulted in the organization of some aberrant Golgi cisternae, which were also formed in the vicinity of ERES, but where cis and trans cisternal markers acted together. These findings provide new information about the relationship between the ERES localization and ER morphology and the organization of the ER-to-Golgi system.
Organization of ERES in S. cerevisiae
ERES is the ER membrane subdomain where COPII vesicles are assembled (Orci et al., 1991; Bannykh et al., 1996; Stephens, 2003; Connerly et al., 2005; Watson et al., 2006). Recently, Shindiapina and Barlowe reported that S. cerevisiae Sec13–GFP and Sec23–GFP localize to small punctate spots, which were long-lived and exhibited restricted movement within the ER, representing ERES in this organism (Shindiapina and Barlowe, 2010). We confirmed that the COPII coat proteins (Sec13, Sec31, and Sec24) colocalized with each other and with Sec16, another well established ERES marker. GFP- or mRFP-fused COPII coat proteins Sec24, Sec13, and Sec31, and Sec16–GFP all yielded punctate patterns of fluorescence. Simultaneous observations of two of these proteins indicated their precise colocalization, indicating that COPII coat proteins and Sec16 accumulate at ERES in S. cerevisiae (Fig. 1A). Sec16 is a peripheral membrane protein predicted to be a scaffold for COPII assembly at ERES (Supek et al., 2002; Connerly et al., 2005). Consistently, Sec16–GFP puncta appear to be unaffected by the sec12-4 mutation, which induced a couple of coalescence of COPII coat puncta upon shift to the restrictive temperature (Fig. 1B). These results suggest that accumulation of the COPII coats at ERES depends on Sar1 GTPase activation, but that of Sec16 does not.
Next, we examined whether Sec12 also accumulate at ERES. Confocal imaging near the center of the cell indicated that fluorescent signals of mRFP–Sec12 were observed throughout the ER, and scattered patterns of ERES fluorescence were adjacent to the Sec12 signals (Fig. 2A, upper panels). Confocal images of the cell periphery revealed that Sec13 puncta were positioned around Sec12, but there was little overlap between the two (Fig. 2A, lower panels). To confirm this result, two-dimensional confocal sections of mRFP–Sec12 and Sec13–GFP fluorescence were reconstructed into three-dimensional images, which are shown in Fig. 2B. Sec13 puncta were localized on the edge of ER membrane labeled by mRFP–Sec12. These results indicate that Sec12 does not accumulate at S. cerevisiae ERES.
ERES localize on high-curvature domains of the peripheral ER network
The appearance of mRFP–Sec12 fluorescence was similar to that of the ER sheet structure. We explored whether the localization of ERES correlates with the shape of the ER. The ER consists of a network of branching tubules and flat sheets. The morphology of the peripheral ER was visualized by GFP targeting to the ER lumen (GFP–HDEL). The peripheral ER consisted of both the tubules (Fig. 3A, arrowhead) and fenestrated sheets (Fig. 2A,B; Fig. 3A, arrows). Rtns and DP1/Yop1 are curvature-stabilizing proteins and partition to high-curvature regions of the ER, ER tubules, and the edge of ER sheets (De Craene et al., 2006; Voeltz et al., 2006; Shibata et al., 2010). Rtn1–GFP signals were found on the tubules and at the rim of the fenestrated sheets (Fig. 3B, lower panels). Sec12 was predominantly enriched in the ER sheets because Rtn1–GFP fluorescent signals circumscribed those of mRFP–Sec12 (Fig. 3B, lower panels). Dual-color observations of Sec13–GFP and Rtn1–mRFP revealed that ERES puncta marked by Sec13–GFP always localized at the high-curvature ER domain labeled by Rtn1–mRFP fluorescence (Fig. 3C, lower panels). The ERES of budding yeast are also found on the nuclear envelope which has much less curvature than the peripheral ER (Bevis et al., 2002; Shindiapina and Barlowe, 2010). Dual color observation of mRFP–Sec12 and Rtn1–GFP showed that high-curvature domains of ER membrane labeled by Rtn1–GFP distributed not only at the cell periphery but also on the nuclear envelope (Fig. 3B, upper panels). We also found that these punctate signals of Rtn1–GFP on the nuclear envelope localized at the bases of the ER tubules and central cisternal ER (Fig. 3B, upper panels, arrowhead) (West et al., 2011). Furthermore, Sec13 colocalized with Rtn1 on the nuclear envelope (Fig. 3C, lower panels). These results indicate that the ERES are preferentially distributed on the high-curvature domains of the ER: ER tubules and the edge of ER sheets. We also examined ERES localization in relation to the surface geometry of the ER membrane. Dual-color confocal images of Sec13–GFP and mRFP–Sec12 were reconstructed into three-dimensional data and visualized by the isosurface mode of Volocity software (Fig. 3D). Their quantitative analysis showed that ERES preferentially faced the saddle-shape surfaces of the high-curvature ER membrane, saddle shape meaning convex (positive curvature) toward the edge and concave (negative curvature) along the edge (Fig. 3E) (Zimmerberg and Kozlov, 2006). Collectively, our results suggest that the distribution of ERES is governed by the geometric features of the ER.
Disruption of the peripheral ER network changes ERES distribution
The restricted localization of ERES on the high-curvature domains of the ER prompted us to examine whether morphological changes of the ER influence the distribution of ERES. We examined ERES distribution in Δrtn1 Δrtn2 Δyop1 cells and Δsey1 Δyop1 cells (Voeltz et al., 2006; Hu et al., 2009). Both mutants exhibit normal proliferation and secretory properties, but they are defective in the formation of ER tubules and accumulate longer peripheral ER sheets (Voeltz et al., 2006). Confocal imaging of mRFP–Sec12 or HDEL–GFP at the cell periphery of these mutants clearly showed that they have continuous unfenestrated peripheral ER sheets lacking a tubular network (Fig. 4A). The number of ERES in these mutant cells was almost the same as the number in wild-type cells (Fig. 4B). Dual-color observations of Sec13–GFP and mRFP–Sec12 at the cell periphery showed that ERES in these mutant cells clustered, in contrast to the scattered pattern in wild-type cells (Fig. 2A); however, they still associated with the remaining edge of the enlarged ER sheets and avoided the flat surface of these sheets (Fig. 4A). These data indicate that ERES distribution is affected by morphological changes in the ER, and that ERES localization is restricted to the high-curvature domains of the ER.
cis-Golgi cisternae are in the vicinity of ERES
According to the cisternal maturation model of Golgi, ERES are birth places for new Golgi cisternae (Glick and Malhotra, 1998). Consistent with this idea, ERES in cells with stacked Golgi cisternae have been shown to be positioned adjacent to the cis-side of the Golgi apparatus, implying that ERES are the origin of the Golgi (Rossanese et al., 1999; Kondylis and Rabouille, 2003; Yang et al., 2005). Golgi cisternae are not stacked in S. cerevisiae. The cisternae are scattered throughout the cytoplasm while they mature. Thus in this organism, the positional relationship between the Golgi and ERES has been less obvious (Rossanese et al., 1999). However, we thought that cis-Golgi could still localize in closer vicinity to ERES than trans-Golgi in S. cerevisiae. We observed ERES, cis-Golgi, and trans-Golgi proteins by dual-color fluorescence confocal microscopy (Fig. 5). The cis-Golgi markers Sed5 and Rer1 sometimes almost overlapped with Sec13. Overlapping Sec7 and Sec13 signals were rarely observed. To quantify the extent of spatial proximity, we calculated Pearson's correlation coefficients between two fluorescent signals of each combination of proteins in a single confocal plane, confirming that cis-cisternae are more closely associated with ERES than trans-cisternae (Fig. 5). This finding demonstrates that S. cerevisiae also exhibits a spatial relationship between ERES and cis-Golgi as described in other cells with stacked Golgi cisternae.
Disruption of the peripheral ER network alters the dynamics of Golgi cisternae
Golgi and ERES show different dynamics in S. cerevisiae. Golgi cisternae are mobile, mature progressively, and dissipate within minutes (Losev et al., 2006; Matsuura-Tokita et al., 2006), whereas ERES are reported immobile and stable (Shindiapina and Barlowe, 2010). In addition, ERES outnumber Golgi cisternae (Rossanese et al., 1999). We were interested in how these compartments maintain their spatial proximity and conducted dual-color time-lapse imaging of ERES and cis-Golgi. In wild-type cells, cis-Golgi labeled with mRFP–Sed5 frequently localized in the vicinity of ERES without being retained at the ERES (Fig. 6A, upper panels). This result suggested that new cis-Golgi cisternae are generated at ERES de novo or pre-existing cis-Golgi approaches dynamically to ERES.
Next, we examined whether morphological alteration of the ER affects the dynamic features of S. cerevisiae Golgi. In Δrtn1 Δrtn2 Δyop1 cells, we found that some cis-Golgi remained near the aligned ERES for a much longer time (Fig. 6A,C). In wild-type cells, the peripheral ER and the plasma membrane are closely apposed in a way that ribosomes are excluded from the plasma membrane face of the peripheral ER (West et al., 2011). In Δrtn1 Δrtn2 Δyop1 cells, some interstices were found between the expanded peripheral ER and plasma membrane, where ERES were ectopically located on the plasma membrane side of the interstices (Fig. 6B, arrowhead). Dual-color time-lapse images showed that cis-Golgi localized on the plasma membrane side of the ER were constantly positioned in the vicinity of ERES, whereas cis-Golgi on the cytoplasmic side of the ER were not (Fig. 6C). Remarkably, the spatial relationship between cis- and trans-Golgi cisternae also changed. We found some cis- and trans-Golgi cisternae exhibited reduced movement and were localized adjacent to each other with some overlap, which were seldom observed in wild-type cells (Fig. 6D) (Losev et al., 2006; Matsuura-Tokita et al., 2006). These findings suggest that the ectopic localization of ERES caused by the morphological change in the ER influenced the dynamic behavior of the Golgi apparatus.
Because the colocalization of cis- and trans-cisternae might indicate structural changes, we decided to examine the morphology of the Golgi apparatus in Δrtn1 Δrtn2 Δyop1 cells and Δsey1 Δyop1 cells using electron microscopy. As shown by fluorescence microscopy, the ER membrane formed longer continuous structures in these mutant cells (Fig. 7B,C). Strikingly, aberrant membrane structures (rings or concentric circles) were often found in the interstice between the ER and the plasma membrane (Fig. 7B–E). This structure also stained by the PATAg method that detects polysaccharides, indicating that this structure derive from the Golgi apparatus (Fig. 7F,G). To specify whether this structure is the Golgi cisternae stably associated with ERES in fluorescence microscopy, we examined immuno-gold staining for cis- and trans-Golgi marker proteins, Sed5 and Sec7 (Fig. 7H). The staining pattern indicated that these membrane structures have a high oligosaccharide content and contain both Sec7 (10-nm gold particles, white arrow) and Sed5 (6-nm gold particles, black arrow). These results indicate that ring or concentric-circular membranous structures are deformed Golgi apparatus including cis- and trans-cisternae.
In this study, we demonstrated that organized S. cerevisiae ERES structures marked by COPII coat proteins colocalize with Sec16 but not Sec12. A recent study documented that sec12-4 and sec16-2 mutations alter the localization of GFP-tagged COPII coat proteins (Castillon et al., 2009; Shindiapina and Barlowe, 2010), indicating that Sar1–GTP and Sec16 have roles in the maintenance of ERES. Here, we show that, in the sec12-4 mutant at a restrictive temperature, COPII coats are clustered into a large structure, but Sec16–GFP remains, exhibiting punctate structures. These large clusters of COPII coats labeled by Sec13–mRFP puncta still contain Sec16–GFP (supplementary material Fig. S1). These results suggest that Sec16 acts early in the Sar1 GTPase cycle and is a primary determinant of ERES formation in S. cerevisiae (Supek et al., 2002; Connerly et al., 2005; Watson et al., 2006; Bhattacharyya and Glick, 2007; Ivan et al., 2008; Hughes et al., 2009). On the other hand, we found that Sec12 is prominently distributed in the ER sheets and does not accumulate at ERES. In P. pastoris, Sec12 localizes at ERES, but S. cerevisiae–P. pastoris chimeric Sec12 is distributed throughout the ER and does not perturb the localization of ERES and Golgi components (Soderholm et al., 2004). Therefore, the accumulation of Sec12 at ERES is not required for ERES formation.
Previous observations indicated that ERES localize on the surface of ER tubules in mammalian cells and plant cells (Hammond and Glick, 2000; Yang et al., 2005). As expected, the preferential partition of ERES into the high-curvature domains of ER became more apparent in cells lacking Rtns and Yop1 or Sey1 and Yop1. We found that ERES were clustered at the remaining high-curvature domains at the edge of the continuous unfenestrated ER sheets even in these mutants. Therefore, these results suggest that high-curvature domains of the ER membrane are required for the localization of ERES.
ERES were first identified morphologically as vesiculating ER regions devoid of bound ribosomes (Palade, 1975). Thus, the question becomes whether high-curvature domains of the ER represent the functional ER domains. Recent confocal fluorescence microscopic analysis showed that the components of the translocation complex are enriched in the ER sheets relative to the ER tubules, which indicates that sheets may have more bound ribosomes per surface area than tubules (Shibata et al., 2010). Voeltz and colleagues directly measured the ribosome density of the ER domains in S. cerevisiae, finding that the cytoplasmic side of the peripheral ER sheets has a high ribosome density, but the plasma membrane side rarely has ribosomes (West et al., 2011), indicating that the edge of the peripheral ER sheets is a boundary between domains with high and low ribosome density. Voeltz and colleagues also reported that ER tubules are low ribosome density domains (West et al., 2011). Therefore, our findings validate that the distribution of ERES at the high-curvature domains, the edge of sheets and tubules, is closely related to the functional compartmentalization of the ER.
In vitro experiments will be required to determine the mechanism of ERES formation restricted at the high-curvature domains of the ER surface, whether different membrane curvature affects the efficiency of COPII assembly. However, the structural information for ERES components might provide a clue about this mechanism. The structure of the Sar1/Sec23/24/cargo pre-budding complex is a concave surface associated with its membrane-orientated face (Bi et al., 2002). Balch and colleagues suggested that Sec23–24 first forms an oligomer, coalescing as minimal tetramer clusters of the Sar1/Sec23/24/cargo pre-budding complex to define a site for Sec13–31 recruitment (Stagg et al., 2008). Thus, the concave surface of these tetramer clusters might recognize the high-curvature domains of the ER. It should be noted, however, that vesicle budding requires not only positive curvature but also negative curvature to be constricted at the neck. Our observations may have some interesting implications here, because ERES appear to prefer saddle-like structures which contain both positive and negative curvatures. In mammalian cells, COPII vesicles found different ER cisternae were closely juxtaposed and protrude into a central region containing a collection of vesicles and tubular elements comprising vesicular tubular clusters, suggesting that the bases of these ER tubules surrounding the vesicular tubular clusters might have negative curvatures (Bannykh et al., 1996). Recently, ERES labeled by Sec16A has been reported to localize to concave cup-shaped structures of the ER membrane which have negative curvatures (Budnik and Stephens, 2009; Hughes et al., 2009). These structures might have both positive and negative curvatures as well, because most of ERES in mammalian cells also localize on the surface of ER tubules. Saddle-like ER membrane may be rich in a variety of lipid components including cone-shaped and reverse-cone-shaped lipids. Dual-color imaging and correlation analysis indicated that cis-Golgi cisternae are not randomly dispersed, but present in the vicinity of ERES, whereas no such correlation was found for trans-cisternae. Recent work showed that the enlarged S. cerevisiae ERES are formed as the result of slowed ER export and often seen in close proximity to cis-Golgi cisternae (Levi et al., 2010). Therefore, our findings and previous report provide support for ERES as originating the Golgi.
An unexpected result of our work was the association of cis- and trans-Golgi cisternae after morphological alteration of the ER. In mammalian cells, Golgi stack formations involve GRASP family proteins, which localize to cis- and medial-trans cisternae (Seemann et al., 2000; Xiang and Wang, 2010). Biochemical studies have shown that GRASP65 forms stable homodimers, and homodimers residing on adjacent Golgi membranes form oligomers. These trans-oligomers are capable of holding the cisternal membranes together in stacks (Wang et al., 2003; Wang et al., 2005). S. cerevisiae has Grh1, a homolog of GRASP, but lacks Golgi stacks in wild-type cells and Δgrh1cells (Levi et al., 2010). However, in Δrtn1 Δrtn2 Δyop1 cells, deformed Golgi structures including cis- and trans-cisternae were generated in the space between the plasma membrane and the expanded ER sheets. These cisternae remain in the vicinity of ERES and show reduced motion. Therefore, early and late Golgi cisternae may exhibit associated structures by repressing their movement at ERES because they mature progressively. Important questions remain, including whether the dynamics of Golgi cisternae determine their own structures, which will require further high-resolution imaging.
Materials and Methods
Yeast strains and culture conditions
The S. cerevisiae strains and plasmids used in this study are listed in supplementary material Tables S1, S2. Cells were grown in MCD medium [0.67% yeast nitrogen base without amino acids (Difco Laboratories Inc.), 0.5% casamino acids (Difco Laboratories Inc.), and 2% glucose] with appropriate supplements. For live imaging, cells were grown at 23°C to the early logarithmic phase.
GFP and mRFP constructs
Strains expressing fluorescent protein-tagged Sec13, Sec16, Sec23 or Sec31 were constructed as described in the yeast GFP database at the University of California, San Francisco (Huh et al., 2003). GFP–Sed5 and mRFP–Sec12 were expressed under the control of the TDH3 promoter on the low-copy plasmid pRS316 or pRS314 (Sato et al., 2001). Sec7–mRFP and Sec7–3HA were expressed similarly except that the ADH1 promoter was used instead of the TDH3 promoter.
Throughout this study, we used the super-resolution confocal live imaging microscope (SCLIM), which we developed by combining a high-speed and high signal-to-noise ratio spinning-disk confocal scanner (Yokogawa Electric, Japan), cooled image intensifies (Hamamatsu Photonics, Japan), and high sensitive HARP cameras (NHK and Hitachi Kokusai Electric, Japan) or EM-CCD cameras (Hamamatsu Photonics, Japan) (Matsuura-Tokita et al., 2006). High space resolution was achieved by oversampling and deconvolution (Nakano and Luini, 2010). Three-dimensional images were reconstructed and deconvoluted by the parameters optimized for the spinning-disk confocal scanner using Volocity software (Perkin Elmer, MA). Temperature-sensitive mutants were observed using a thermocontrol stage (Tokai Hit, Japan) at either a permissive or restrictive temperature. Pearson's correlation coefficients were calculated for two fluorescent signals to estimate spatial proximity using Volocity software (Perkin Elmer, MA).
Cells were rapidly frozen in a high-pressure freezer (HPM010, Bal-Tec Inc., Germany) and transferred to 2% OsO4 in anhydrous acetone pre-cooled in liquid nitrogen. Samples were kept at –80°C for 7 days, –20°C for 2 h, 4°C for 2 h, then at room temperature for 2 h. After a wash with anhydrous acetone, the samples were embedded in Spurr's resin (Nisshin EM, Japan). Ultrathin sections were cut, stained with uranyl acetate and lead citrate, and observed under a transmission electron microscope (JEM1200EX, JEOL, Japan). For oligosaccharide staining, two sections were collected independently, one on a copper grid for conventional structural observation and the other on a gold grid for oligosaccharide staining by periodic acid-thiocarbohydrazide–silver protein (PATAg) (Thiery and Bader, 1967).
For immunoelectron microscopy, samples were fixed as described above and transferred to 0.01% OsO4 in anhydrous acetone. The samples were kept at –80°C for 7 days, –20°C for 2 h, and 4°C for 2 h. After washing with anhydrous ethanol, the samples were embedded in LR-white resin. Polymerization was carried out at –20°C using a UV polymerizer (TUV-200, Dosaka-EM, Japan). Ultrathin sections were cut, immunolabeled, and stained with uranyl acetate. For immunodetection, rabbit anti-GFP polyclonal antibody (1:50; Invitrogen, CA) and 6-nm gold goat anti-rabbit conjugate (1:50 dilution; Jackson ImmunoResearch, CA) were used as primary and secondary antibodies to detect GFP–Sed5, and mouse anti-HA monoclonal antibody 12CA5 (1:25, 16 µg/ml, Roche, Switzerland) and 10-nm gold goat anti-mouse conjugate (1:50 dilution; Zymed, CA) were used as primary and secondary antibodies to detect Sec7–3HA.
We thank T. A. Rapoport and Y. Shibata of Harvard University for the yeast reticulon mutant strains and G. K. Voeltz of the University of Colorado at Boulder for exchange of information prior to publication. We also thank A. Hirata of the University of Tokyo for technical suggestions about EM and Y. Suda, K. Fukaya, Y. Sugisawa, R. Kiuchi and Y. Zenke of the Nakano Laboratory for assistance and helpful suggestions.
This work was supported by a Grant-in-Aid for Specially Promoted Research [grant number 20001009 to A.N.] and Scientific Research (C) [grant number 22570194 to K.K.] from the Ministry of Education, Culture, Sports, Science and Technology of Japan and by the Bioarchitect, the Extreme Photonics, and the Cellular Systems Biology Projects of RIKEN to A.N.
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.100065/-/DC1
- Accepted February 19, 2012.
- © 2012. Published by The Company of Biologists Ltd