Portal de Periódicos da CAPES

ER arrival sites for COPI vesicles localize to hotspots of membrane trafficking

2016; Springer Nature; Volume: 35; Issue: 17 Linguagem: Inglês

10.15252/embj.201592873

1460-2075

Saskia Schröter, Sabrina Beckmann, Hans Dieter Schmitt,

Endoplasmic Reticulum Stress and Disease

Article20 July 2016free access Transparent process ER arrival sites for COPI vesicles localize to hotspots of membrane trafficking Saskia Schröter Saskia Schröter Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany Search for more papers by this author Sabrina Beckmann Sabrina Beckmann Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany Search for more papers by this author Hans Dieter Schmitt Corresponding Author Hans Dieter Schmitt orcid.org/0000-0001-6152-2716 Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany Search for more papers by this author Saskia Schröter Saskia Schröter Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany Search for more papers by this author Sabrina Beckmann Sabrina Beckmann Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany Search for more papers by this author Hans Dieter Schmitt Corresponding Author Hans Dieter Schmitt orcid.org/0000-0001-6152-2716 Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany Search for more papers by this author Author Information Saskia Schröter1, Sabrina Beckmann1 and Hans Dieter Schmitt 1 1Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany *Corresponding author. Tel: +49 551 201 1652; E-mail: [email protected] The EMBO Journal (2016)35:1935-1955https://doi.org/10.15252/embj.201592873 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Abstract COPI-coated vesicles mediate retrograde membrane traffic from the cis-Golgi to the endoplasmic reticulum (ER) in all eukaryotic cells. However, it is still unknown whether COPI vesicles fuse everywhere or at specific sites with the ER membrane. Taking advantage of the circumstance that the vesicles still carry their coat when they arrive at the ER, we have visualized active ER arrival sites (ERAS) by monitoring contact between COPI coat components and the ER-resident Dsl tethering complex using bimolecular fluorescence complementation (BiFC). ERAS form punctate structures near Golgi compartments, clearly distinct from ER exit sites. Furthermore, ERAS are highly polarized in an actin and myosin V-dependent manner and are localized near hotspots of plasma membrane expansion. Genetic experiments suggest that the COPI•Dsl BiFC complexes recapitulate the physiological interaction between COPI and the Dsl complex and that COPI vesicles are mistargeted in dsl1 mutants. We conclude that the Dsl complex functions in confining COPI vesicle fusion sites. Synopsis The ER-resident Dsl tethering complex interacts with COPI vesicles at spatially distinct ER arrival sites to prevent their mistargeting to adjacent ER exit sites, thus allowing to confine retrograde membrane transport vesicle fusion sites. BiFC data reveal that coated COPI vesicles interact with the ER-resident Dsl complex at discrete subdomains of the ER (ER arrival sites, ERAS). ERAS and ER exit sites (ERES) share an interface at the ER and are juxtaposed to the Golgi, possibly marking an interaction hub of the early secretory pathway. The Dsl complex regulates trafficking by preventing COPI vesicles from approaching adjacent ERES. Growing cells feature a dominant polarized ERAS along with markers of secretion at the site of outgrowth, pointing towards a polarized focus of early trafficking activity. Non-growing cells show a disperse ERES-like ERAS pattern. Introduction The secretory pathway is a multi-step process with the endoplasmic reticulum (ER) as the starting point, the Golgi apparatus as a sorting hub and lysosome or plasma membrane as two main destinations. Vesicles and tubules are the main carriers that mediate transport between these organelles. The concomitant loss of material from the membrane compartments has to be compensated either by de novo synthesis and fresh supply of material, or by retrograde transport (Pelham, 1996). Since retrograde transport also involves the re-routing of material via additional organelles like the early and recycling endosomes, the cells are crowded with a large number of carriers with different origins and destinations. Thus, the arrival and exit of material at a particular organelle has to be carefully organized. One way to manage the simultaneous arrival and departure of material is the spatial separation of exit and entry sites within one organelle. Remarkably, the initial and the final steps of secretory pathway are best characterized in respect to such a spatial organization. At the endoplasmic reticulum, COPII vesicles always form within defined regions called ER exit sites (ERES), also termed transitional ER (Bannykh et al, 1996). In highly polarized cells like yeast cells (Saccharomyces cerevisiae) or nerve cells, the secretory vesicles fuse at the plasma membrane within areas marked by the polarisome (Pruyne et al, 1998; Sheu et al, 1998; McCusker et al, 2012) or the scaffold of the presynaptic active zone in nerve cells, respectively (Jahn & Fasshauer, 2012; Chua, 2014). At the rim of these exocytic regions, one finds areas of active internalization (McCusker et al, 2012). Clearly, the actin cytoskeleton plays an important role in the arrangement of spots of highest exocytic and endocytic activities in yeast and nerve terminals (Li & Gundersen, 2008; Nelson et al, 2013). Marker proteins for exit sites can be scaffolding factors, proteins required for the vesicle formation, or coat proteins themselves. In case of the ERES, the scaffolding protein Sec16p or the COPII coat subunit Sec24p are often used in yeast and mammalian cells (Shaywitz et al, 1997; Shindiapina & Barlowe, 2010; Bharucha et al, 2013). Likewise, suitable markers for the site of active clathrin-mediated endocytosis are the clathrin coat complex itself or dynamin 2, a protein involved in vesicle formation (Rappoport & Simon, 2003). Similarly, the site of vesicle fusion at the plasma membrane in yeast can be identified by staining proteins present on secretory vesicles like the small GTPase Sec4p or proteins that reside at the plasma membrane like Sec3p, a subunit of the exocyst tethering complex present at the plasma membrane (Finger et al, 1998). For Golgi membranes and endosomes, the sites of vesicle formation and fusion seem to be less clearly segregated. At the Golgi, the formation of COPI-coated vesicles seems to be restricted to the edges of cisternae (Ladinsky et al, 1999; Klumperman, 2011). On the surface of endosomes, a separation of large domains of similar size has been observed based on the presence of different GTPases of the Rab/Ypt superfamily, as well as different phosphoinositides (Miaczynska & Zerial, 2002; Jean & Kiger, 2012). In contrast to the well-characterized ERES, however, we know next to nothing about the sites where the incoming COPI vesicles fuse with the ER in Saccharomyces cerevisiae or mammalian cells. In fact, to date it has been impossible to visualize postulated ER arrival sites (ERAS) (Spang, 2009, 2012). The alternative term "ER import sites" (ERIS) was later introduced by plant researchers (Lerich et al, 2012). The situation is different in the yeast Pichia pastoris and in plant cells, where COPI and COPII vesicles are confined to a narrow interface between closely apposed ER and the stacked Golgi membranes (Rossanese et al, 1999; Lerich et al, 2012). In these cells, closely confined ER arrival sites may be the result of a tight apposition of Golgi and ER membranes. To identify ER arrival sites in budding yeast or mammalian cells, tethering complexes may be suitable marker proteins. They are thought to make the first contact between incoming vesicles and the target membrane. Many of them interact with SNARE proteins, the actual catalysts of membrane fusion (Yu & Hughson, 2010). This is also true for the Dsl complex in yeast. This complex is responsible for the tethering of Golgi-derived COPI vesicles to the ER. One of its three subunits, Dsl1p, carries a lasso domain, which can bind two of the seven COPI subunits, α-COP and δ-COP in vitro (Andag & Schmitt, 2003; Ren et al, 2009; Zink et al, 2009). The other two Dsl subunits, Dsl3(Sec39)p and Tip20p, are in a tight complex with the SNAREs Use1p and Sec20p at the ER (Kraynack et al, 2005; Ren et al, 2009). In mammalian cells, the equivalent of the Dsl complex, the NRZ complex, also interacts with COPI vesicles. It additionally requires the UVRAG protein for this interaction (Hirose et al, 2004; He et al, 2013). In contrast to the aforementioned exocyst, the Dsl complex and its associated SNAREs are distributed uniformly over the ER (Kraynack et al, 2005; Meiringer et al, 2011). This may indicate that COPI vesicles fuse uniformly across the ER. Contrasting evidence came from our previous work, showing that COPI vesicles accumulate next to COPII-positive ER domains upon Dsl1p depletion (Zink et al, 2009), suggesting an association of ER arrival sites with ER exit sites. Thus, the conventional approach of visualizing possible ER arrival sites through a single marker protein did not yield conclusive evidence for or against the existence of specifically confined ER arrival sites. We therefore set out to map ER arrival sites in living yeast, using a more elaborate approach. Two previous findings were crucial for that: (i) COPI vesicles that arrive at the ER still carry their coat (Zink et al, 2009) and (ii) the Dsl complex contains several COPI binding sites (Andag & Schmitt, 2003; Zink et al, 2009; Diefenbacher et al, 2011; Suckling et al, 2015). A recent study by Dodonova et al (2015), which presents the full structure of the COPI coat, has confirmed that the Dsl1p binding sites of COPI subunits are well accessible at the surface of the COPI coat, thereby corroborating the plausibility of an interaction. We made use of these findings by employing bimolecular fluorescence complementation (BiFC) as a probe to visualize sites of interaction between COPI coat and Dsl complex at the ER. This "split-GFP technique" relies on the formation of a fluorescent reporter protein by association of its N-terminal and C-terminal fragments, provided that the proteins carrying these fragments as tags come into close proximity. One caveat of this approach is that, under several in vitro conditions, the formation of bimolecular fluorescent complexes was found to be irreversible (Ghosh et al, 2000; Hu et al, 2002; Magliery & Regan, 2005). Magliery et al (2005) suggested, however, that this property would make the BiFC approach suitable for the visualization of otherwise transient interactions. This strategy was successfully applied by Morell et al (2007). Moreover, BiFC-mediated fluorescence signals have been reported to change dynamically in vivo in response to cell stimuli and associated protein–protein interaction states (Schmidt et al, 2003; Guo et al, 2005). This technique has also been used in yeast for the analysis of proteins interactions in membrane traffic (Lipatova et al, 2012; Mao et al, 2013; Weber-Boyvat et al, 2013). Since interactions between Dsl1p and COPI coat are very weak or transient (Zink et al, 2009), the BiFC technique appeared suitable for identifying the sites where they interact in the cells. We introduced segments encoding parts of the YFP variant Venus (vYFP) (Sung & Huh, 2007) at the chromosomal loci of genes whose products are involved in transport between ER and Golgi. This included four COPI subunits, the two Dsl subunits Dsl1p and Dsl3p, a cargo receptor, and proteins involved in COPII vesicle formation. Because of the potential irreversibility of YFP fragment complementation, we performed meticulous controls to ensure continued functionality of the involved transport processes. By analyzing the growth characteristics of a large number of BiFC combinations in mutant background, we found evidence for a quite efficient reconstitution of the normal COPI•Dsl interaction by the BiFC complexes. Taken together, our data suggest that the BiFC spots in fact represent specific domains where Golgi-derived vesicles fuse with the ER. We confirm the notion that COPI vesicles are at least partially coated upon their arrival at the ER. The emerging picture indicates that COPI vesicles do not fuse randomly across the ER. Rather, ERAS are highly ordered and are localized in a polarized manner, thereby showing a different pattern from ER exit sites. With these new insights, we help build a more detailed picture of intracellular transport routes. Results Construction and validation of BiFC strains We performed bimolecular fluorescence complementation (BiFC) assays to assess whether COPI-coated vesicles or membranes establish contact with the ER in yeast. The system we employed utilizes the N- and C-terminal fragments (VN and VC) of vYFP (Venus version of the yellow fluorescent protein; Nagai et al, 2002; Sung & Huh, 2007). They comprise residues 1–172 (VN) or 155–238 (VC) of the fluorescent protein, respectively, and were introduced at the chromosomal loci of the corresponding genes. Figure 1A lists the genes analyzed here and provides information about the kind of modification that was introduced at each site as well as a graphic representation of all examined BiFC connections. Most single- and double-tagged (expressing a VN- and a VC-tagged protein) strains were fully viable even at 37°C (Figs EV1H and EV2, and Appendix Fig S1). We confirmed the expression of the BiFC-tagged proteins through immunoblots (Fig EV1B–G). Compared to wild-type protein levels, levels of tagged proteins were either reduced or equal to wild type. Importantly, the formation of BiFC complexes in cells expressing VN as well as VC-tagged genes did not affect the expression levels (Fig EV1D–G). Most of the localization experiments involving a COPI subunit were performed using either β'-COP or ε-COP (β'-COPVN or ε-COPVC), two COPI subunits that exhibited either normal or reduced expression levels as determined by Western blotting. To analyze whether the introduction of BiFC tags affects retrograde transport routes, we checked the protein levels of the Golgi proteins Emp47p and Rer1p in BiFC strains. Both proteins cycle between Golgi and ER and require normal Golgi ER transport for their stability (Schröder-Köhne et al, 1998; Sato et al, 2001). Emp47p abundance in COPI•Dsl combinations was reduced to about 60% of the wild-type level, while the Rer1p levels were unaffected. However, the VN tag at a COPI subunit alone was sufficient to induce this reduction (Fig EV1J), suggesting that the reduction in Emp47p levels was due to the tag and not due to the formation of BiFC complexes. HDEL signal-mediated ER retrieval was intact, as the fluorescence of HDEL-tagged RFP yielded normal ER staining (Figs 4A and EV4A). Figure 1. Specificity of BiFC signals and effects of BiFC formation on cell viability Schematic representation of all BiFC combinations studied in this work (lime green cartoons: the two most commonly used BiFC pairs β'-COPVN•Dsl3pVC and Dsl1pVN•ε-COPVC; dashed lines: all other BiFC pairs presented in this work). All BiFC-tagged proteins are listed in the table. Bimolecular fluorescence complementation between subunits of the Dsl complex and the COPI coat. The BiFC pair β'-COPVN•Dsl3pVC yielded fluorescent foci that showed dramatically reduced fluorescence intensity in the COPI binding-deficient dsl1-5xWA mutant. Schematic yeast cell representations depict the typical BiFC foci localization patterns. Scale bar, 10 μm. Quantification of the β'-COPVN•Dsl3pVC BiFC signals in wild-type cells and in COPI binding-deficient dsl1-5xWA and dsl1-Δlasso mutants (as shown exemplarily in B). Mean values + SEM of at least three independent experiments (n = 3–8) are displayed. Statistical analysis was carried out by comparing the number of cells with BiFC signals over integrated thresholds to those without signals (*P < 0.05). dsl1 mutant strains show a significant decrease in brightness, spot number, and spot size in comparison with their corresponding DSL1 wild-type strains (two-sample t-test). The dsl1-5xWA mutation also led to comparable signal reduction in cells carrying the VNI152L variant in the β'-COPVN I152L•Dsl3pVC combination. ***P < 0.001. Subcellular localization of different BiFC-tagged protein pairs. The Dsl1pVN•Dsl3pVC combination involving subunits of the ER-resident Dsl1 complex exhibited typical ER localization. Sec16pVN•Sec16pVC signals in a diploid heterozygous strain displayed a pattern typical for ER exit sites, while diploid cells carrying the α-COPVN•α-COPVC BiFC pair showed a Golgi-like fluorescence pattern. Scale bar, 10 μm. The localization of COPI•Dsl BiFC foci compared to the autophagy marker Ape1p. Double fluorescence micrographs of the BiFC pair β'-COPVN•Dsl3pVC and the autophagy marker mRFPApe1p are superimposed with the DIC image. BiFC-YFP signals are pseudocolored green. Fluorescent signals appeared distinct from each other. Also no co-localization was observed when we analyzed COPI•COPII(Sec16p) BiFC spots in cells expressing mRFP tagged APE1 (S. Beckmann, unpublished results). Scale bar, 10 μm. Growth effects of Venus fragment tags in dsl1 mutant cells. Serial tenfold dilutions of liquid cell cultures were spotted on agar plates and incubated at 30°C for 2 days. The images in the second row show that β'-COPVN-producing cells expressing the dsl1-5xWA mutation could not grow. The complementation of β'-COPVN by its cognate interaction partners Dsl1pVC and Dsl3pVC, but not by non-cognate BiFC partners (Sec24pVC, Sec16pVC, or VCRer1p), suppressed these growth defects. Spot assays of exemplary BiFC partners are depicted. Asterisk: Dsl1pVC carrying the dsl1-5xWA defect was expressed from a plasmid. See Figs EV1 and EV2 and Appendix Fig S1 for full data display. BiFC signal quantifications of the strains presented in (F). β'-COPVN yielded BiFC fluorescence signals with all tested complementation partners. This rules out the possibility that the lack of suppression observed in (F) simply reflects the inability of the BiFC partners to form a complex and shows that the successful complementation of the split-YFP fragments does not per se suppress the synthetic lethal effect of the VN tag in dsl1-5xWA cells. Mean values + SEM of at least three independent experiments (n = 3–8) are displayed. BiFC signal intensities greatly depend on which BiFC combination is analyzed. Fluorescence intensities of COPI•Dsl BiFC pairs (left) were much higher than those of COPII•Dsl BiFC pairs (right). Mean values + SEM of at least three independent experiments (n = 3–5) are displayed. Localization of Dsl1pCFP in β'-COPVN•Dsl3pVC cells. An overall ER localization of Dsl1pCFP was retained in β'-COPVN•Dsl3pVC cells, while some of the signal was found in the foci. BiFC-YFP signals are pseudocolored red. Scale bar, 10 μm. Download figure Download PowerPoint Click here to expand this figure. Figure EV1. Expression analysis and growth characteristics of mutants used in this studyAll strains are listed in Appendix Table S1. A. Mutations in the Dsl1p lasso domain and the deletion of the lasso domain do not affect the stability of the protein. Since Dsl1p antibodies preferentially recognize the Dsl1 protein carrying an intact lasso domain, we analyzed myc-tagged proteins. The sequences encoding two myc-tags were introduced 3′ of the DSL1 alleles as described in Materials and Methods. Extracts of two parallel samples were analyzed using the 9B11 antibody (Cell Signaling Technologies, Danvers, MA). δ-COP was used as a loading control. As an additional loading control, a Ponceau S-stained protein band at about 120 kDa is shown. B, C. Immunoblot analysis of strains expressing either VN or VC tags alone. For the upper panels, COPI-specific antibodies were used and untagged COPI subunits were employed as a loading control. For the blots shown in the second row, GFP-specific antibodies were used. As an additional loading control, a Ponceau S-stained protein band at about 120 kDa is shown. The results indicate that expression of α-COP and β'-COP is not affected by the tags, while the tagged versions of genes encoding δ-COP and ε-COP seem to be less well expressed. We do not know why the expression varies so much since all tagged genes except RER1 are under the control of their own promoter, and all 3′-tagged genes are followed by the ADH1 3′-UTR. Our results with VC-RER1 indicate that the codon usage of the tag may be important at least when the tag is at the 5′ end (Lipatova et al, 2015). We obtained very similar results when we used a chromosomally inserted VC-RER1 version under the control of the strong RPL7B promoter or a single-copy vector-encoded version under the endogenous promoter where the codon usage of the VC tag was adapted to yeast. D, E. The formation of BiFC complexes has no effect on the expression of the modified proteins. Blots show extracts from individually tagged strains as well as strains expressing BiFC pairs. The BiFC combinations Dsl1pVN•ε-COPVC and β'-COPVN•Dsl3pVC shown here were those which we used most often in this work for microscopy. They represent BiFC pairs with either a poorly or a well-expressed BiFC-tagged COPI subunit. As a loading control, a Ponceau S-stained protein band at about 120 kDa is shown. F. RER1 carrying the VC tag at its 5′ end was expressed under the control of the strong RPL7B promoter. The full-length fusion protein is present in amounts comparable to the wild-type protein. A smaller protein fragment is very likely the product of cleavage by vacuolar proteases (Sato et al, 2001). The antibody used was raised against the C-terminal 27 residues of Rer1p (Boehm et al, 1997). As a loading control, a Ponceau S-stained protein band at about 120 kDa is shown. G. Dsl1p and Dsl3p levels do not change significantly when they carry BiFC tags or when expressed as BiFC pairs. Antibodies raised against the full-length proteins were used for this analysis. Apart from δ-COP, the area around a 70 kDa band of the Ponceau S-stained blots is shown as a loading control. H. Growth assays with strains expressing different BiFC combinations. The cell density of exponentially growing cells was adjusted to 1 OD600. These suspensions as well as three serial dilutions were spotted on agar plates. Images were taken after 48 h incubation at the indicated temperature. Rer1p marked with one asterisk indicates that the VC-tagged RER1 gene was expressed under the control of the RPL7B promoter from the RER1 locus. The plasmid-encoded version with codon-adapted VC tag under the control of the endogenous promoter is marked by two asterisks. The growth test shown was performed on selective minimal medium to prevent the loss of plasmids. I. The β'-COPVN•Dsl3pVC BiFC pair does not interfere with the binding of the Dsl complex to the SNARE protein Sec20p. Tandem affinity-purified Sec20pTAP containing complexes from radiolabeled BiFC and non-BiFC cells were isolated as described previously (Kraynack et al, 2005). The BiFC pair used was the β'-COPVN•Dsl3pVC combination. Note that no matter whether the TAP-tag was at the C-terminus of Sec20p (first lane) or Dsl1p (third lane), comparable amounts of COPI subunits co-purified with the tagged protein. As a control, SEC20-TAP cells were analyzed which carried no BiFC combination (second lane). Very little Sec20pTAP was visible due to the low amount of Cys and Met residues. We used immunoblot analysis to confirm that the tagged protein is produced. Tandem affinity-purified complexes were analyzed by SDS–PAGE. Gels were dried and analyzed using a phosphorimager (FLA7000, Fujifilm). J. Emp47p levels are reduced to about 60% of the wild-type level in β'-COPVN expressing cells. We observed a reduction in the Emp47p level in extracts from cells with different BiFC combinations (H.D. Schmitt, unpublished results). This was not observed when strains produced β'-COP fused to full-length GFP. Thus, it was not simply the size of the tag that caused the effect. Using extracts from six wild-type, six β'-COPVN, and six β'-COPVN•Dsl3VC strains from older as well as fresh crosses, our immunoblot analysis clearly showed that the reduction in the Emp47p level was not due to the BiFC formation. It can already be observed in β'-COPVN-producing cells. The blots were analyzed using a Lumi-Imager camera system and quantified using the Lumi-Imager software (Roche). Hexokinase was used as a loading control. The stability of the Rer1 protein was not affected. K. The presence of untagged proteins in addition to the VC-tagged Dsl proteins reduces the BiFC signal, while it does not change the polarized appearance of BiFC spots. For the micrographs shown, diploid strains were imaged that express VN-tagged β'-COP from both SEC27 loci. At the DSL3 loci, they express a wild-type and a VC-tagged version (upper panel) or two DSL3VC alleles (lower panel). The same result was obtained with strains that are heterozygous or homozygous for the equivalent DSL1 versions. Moreover, a β'-COPVN/β'-COPVN, DSL1VN/DSL1, DSL3VN/DSL3 strain could not be distinguished from a homozygous β'-COPVN/β'-COPVN DSL3VN/DSL3NV strain (H.D. Schmitt, unpublished results). Thus, the BiFC spot formation is not due to a depletion of Dsl proteins. Scale bar, 5 μm. L. PLA detection of COPI/Dsl and COPI/ER-SNARE interactions. Superimposed representative images of fixed yeast cells carrying Dsl3pGFP, Sec20pGFP, or no tag (wild type). Cells were fixed with PFA and immunodecorated with primary antibodies against endogenous COPI, as well as GFP. Subsequently, the PLA reaction was carried out according to the manufacturer's instructions, using Duolink® In Situ red PLA reagents. Proximity of COPI to Dsl3pGFP or Sec20pGFP produced to a fluorescent spot in situ. Gray channel: DIC; blue: DAPI; red: PLA signal. Scale bar, 10 μm. Download figure Download PowerPoint Click here to expand this figure. Figure EV2. COPI•Dsl BiFC interactions can suppress lethal combinations of dsl1 mutations and unpaired VN tagsThe heat maps in this figure represent a summary of the growth assays shown in Appendix Fig S1 (representative data are shown in Fig 1F). Growth in successive dilution spots was scored (green: full growth compared to control strain, red: complete inability to grow). The cells carried either the wild-type DSL1 gene or one of two different dsl1 mutations. Lettering of the datasets is in accordance with Appendix Fig S1. The results indicate that the dsl1 mutations are in fact lethal in combination with all VN-tagged COPI subunits. This is suppressed by the presence of the VC tag at Dsl subunits and, very importantly, only the Dsl proteins. See the Appendix for further discussion of these results. Plasmid-encoded wild-type and mutant versions of DSL1 support growth of DSL1 shutoff cells that carry no other modification. The presence of unpaired VN tags at the C-terminus of α-, β'-, δ-, and ε-COP was lethal when combined with the dsl1-5xWA mutation. Cells producing α-COPVN were temperature-sensitive no matter what DSL1 version was present in the cells. δ-COPVN-producing cells were also temperature-sensitive, but a surplus of Dsl1p suppressed the temperature sensitivity. Compared to the VN tag, the VC fragment at the C-terminus of COPI subunits had a less dramatic effect on growth when combined with the dsl1-5xWA mutation. Cells producing δ-COPVC were an exception. The δ-COPVC cells were temperature-sensitive no matter what DSL1 version was expressed. It has been observed previously that δ-COPGFP cells are also temperature-sensitive (Zink et al, 2009; see also F and G). No synthetic lethality was observed when the dsl1-5xWA mutation was expressed in cells with VN or VC tags at the C-terminus of Dsl proteins or COPII subunits Sec24p and Sec16p. However, cells expressing a VN or VC-tagged dsl1-5xWA protein grew slower than the corresponding wild-type version even at room temperature. The lethal COPI/dsl1-5xWA combinations shown in (B) were viable when DSL3VC was expressed as well, suggesting that the BiFC interaction can reconstitute the normal function of the coat/tether interaction. Note that the β'-COPVN•Dsl3pVC combination, which includes the most efficiently expressed BiFC-tagged COPI subunit, grew well even at 37°C (see also Fig 1F), while the combination that includes the β'-COPVN I152L version could not grow at 37°C. As discussed below, this observation indicates that for suppression to occur, it is optimal if the BiFC formation is very efficient. As shown in (C), cells with VC-tagged δ-COP stopped growing when the dsl1-5xWA mutant allele was the only DSL1 version expressed in the cells. The growth of these cells was improved by the VN tag fused to Dsl1p (compare the last rows of C and F). This effect was limited to 30°C since the cells were temperature-sensitive. Dsl1p carrying the VC tag rescued two of the four COPIVN/dsl1-5xWA combinations (ε-COPVN and β'-COPVN). In contrast to DSL3VC, DSL1VN could not suppress the equivalent combinations with α-COPVN and δ-COPVN. For these tests, a plasmid-encoded VC-tagged version of dsl1-5xWA was used. The failure of this plasmid to support growth of α-COPVN and δ-COPVN producing cells may indicate that DSL1VN is less efficient as suppressor than DSL3VN. However, one must keep in mind that the BiFC tags fused to the dsl1-5xWA protein alone caused a growth defect (D). Therefore, the fac