Analysis of COPII vesicles indicates a role for the Emp47-Ssp120 complex in transport of cell surface glycoproteins

Abstract

Coat protein complex II (COPII) vesicle formation at the endoplasmic reticulum (ER) transports nascent secretory proteins forward to the Golgi complex. To further define the machinery that packages secretory cargo and targets vesicles to Golgi membranes, we performed a comprehensive proteomic analysis of purified COPII vesicles. In addition to previously known proteins, we identified new vesicle proteins including Coy1, Sly41 and Ssp120, which were efficiently packaged into COPII vesicles for trafficking between the ER and Golgi compartments. Further characterization of the putative calcium-binding Ssp120 protein revealed a tight association with Emp47 and in emp47Δ cells Ssp120 was mislocalized and secreted. Genetic analyses demonstrated that EMP47 and SSP120 display identical synthetic positive interactions with IRE1 and synthetic negative interactions with genes involved in cell wall assembly. Our findings support a model in which the Emp47-Ssp120 complex functions in transport of plasma membrane glycoproteins through the early secretory pathway.

The biogenesis of secretory proteins is initiated at the endoplasmic reticulum (ER) where nascent polypeptides are translocated across the ER membrane and folded into transport competent forms. For anterograde transport from the ER, fully folded secretory cargos are then packaged into COPII (coat protein complex II) coated transport intermediates for delivery to the Golgi complex (1). Selective export of cargo from the ER is directed by the inner layer Sec23/24 adaptor subunits of COPII, which form ternary cargo complexes with the Sar1 GTPase and segregate anterograde cargo from ER resident proteins (2, 3) at specific ER exit sites (4–6). Polymerization of outer layer Sec13/31 subunits at ER exit sites produces cage-like structures that drive membrane deformation and produce transport carriers (7, 8). In addition to nascent secretory cargo, the COPII machinery selectively incorporates vesicle components that are required for targeted membrane fusion and bi-directional transport between the ER and Golgi compartments (9).

To efficiently export a diverse array of secretory proteins from the ER, additional transmembrane receptor proteins act in concert with the COPII machinery to chaperone and/or link cargo to COPII adaptor subunits (10). A well-characterized example is the mammalian lectin-binding ERGIC-53/LMAN1 (henceforth ERGIC53) cargo receptor that is required for efficient transport of coagulation factors V and VIII, specific cathepsins and additional glycoprotein cargo (11–13). Other examples include p24 protein packaging of GPI-anchored cargo (14), Erv29 incorporation of soluble secretory cargo (15) and Erv14-dependent export of specific transmembrane cargo (16, 17). However, the mechanisms by which receptors act to capture the full range of diverse cargo, how cargo binding is regulated and how export is coordinated with the ER quality control machinery are open questions in the trafficking field.

Proteomic analyses of isolated intermediates in the early secretory pathway have proven valuable in defining cargo sorting and vesicle fusion machineries (18–20). In this study we apply a comprehensive proteomic approach to identify additional factors that are efficiently packaged into yeast COPII vesicles and appear to cycle between the ER and Golgi compartments. The putative Ca²⁺ binding protein Ssp120 was detected as an abundant COPII vesicle cargo and further analyses indicated this protein forms a tight complex with Emp47, a putative cargo receptor in yeast that shares amino acid identity and structural similarity to the mammalian carbohydrate-binding cargo receptor ERGIC53 (21–23). Recent large-scale surveys also identified and partially characterized the Emp47-Ssp120 interaction (17, 24). In accord with these findings we observed that Ssp120 depends on Emp47 for efficient incorporation into COPII vesicles and a normal distribution between ER and Golgi compartments. Our results indicate that the Emp47-Ssp120 complex acts together to manage secretory cargo transport through the early secretory pathway.

Results

Identification of COPII vesicle proteins through en bloc mass spectrometry analysis

A protocol to generate ER-derived vesicles in vitro using washed microsomes and purified COPII components was scaled up for purification and detection of vesicle polypeptides on protein-stained SDS-PAGE gels (19, 25, 26). As shown in Figure 1A, silver staining revealed increased levels of a number of protein species coincident with the addition of COPII components. To analyze vesicle polypeptide content, samples were concentrated and run partially into an SDS-PAGE gel. After staining with colloidal Coomassie, each lane was cut into three sections as depicted in Figure 1B, which were then submitted for protein analyses by microcapillary LC/MS/MS. Gel sections were cut unevenly to compensate for the higher concentration of proteins near the dye front.

Figure 1. Preparation of COPII vesicles for analysis by mass spectrometry.

A) Nycodenz gradient isolated COPII vesicles visualized by silver stain. B) Vesicles as in (A) run partially (~1.7 cm) into an SDS-PAGE gel and visualized with colloidal Coomassie. Rectangles denote the partitions of the gel submitted for peptide identification by mass spectrometry.

Analysis by mass spectrometry yielded peptide identification of 290 unique proteins across the six panels (Table S1). Of these identified proteins, 152 had peptide representatives in both mock (no added COPII components) and plus COPII-component samples. Importantly, the number of unique peptides identified for known ER vesicle proteins was greater in the plus COPII-component sample compared to the mock control. We used this comparison to qualitatively assess COPII-dependent protein enrichment. Table 1 reports the peptide representation of previously documented ER vesicle proteins, representative secretory cargo proteins, and candidate vesicle proteins, which displayed ER vesicle-like peptide representation. The list of known vesicle proteins identified here includes most we had identified previously (19) as well as several other notable Erv proteins described elsewhere (27–31). This indicates the en bloc method is robust and should assist in identification of new ER vesicle proteins.

Table 1.

Peptide representation of various classes of ER vesicle proteins

ORF	Protein	Unique peptides		Description
		−COPII	+ COPII
Erv proteins reported in Otte et al., 2001 (19)
YIL004C	Bet1	0	2	ER and Golgi Qc-SNARE protein
YGL200C	Emp24	0	5	Member of the p24 family involved in ER to Golgi transport
YFL048C	Emp47	5	9	Type I membrane protein of ER-derived COPII-coated vesicles
YAR002C–A	Erp1	3	5	Member of the p24 family involved in ER to Golgi transport
YAL007C	Erp2	1	4	Member of the p24 family involved in ER to Golgi transport
YGL054C	Erv14	1	3	Protein localized to COPII-coated vesicles, cargo receptor
YML012W	Erv25	1	7	Member of the p24 family involved in ER to Golgi transport
YGR284C	Erv29	2	11	Protein localized to COPII-coated vesicles, cargo receptor
YML067C	Erv41	0	13	Protein localized to COPII-coated vesicles, involved in retention
YAL042W	Erv46	0	5	Protein localized to COPII-coated vesicles, involved in retention
YCL001W	Rer1	1	6	Protein involved in retention of membrane proteins
YNL263C	Yif1	0	4	Integral membrane protein, Yip1 interacting factor
YNL044W	Yip3	0	0	Protein localized to COPII vesicles
ER vesicle proteins in other reports (19, 27–29, 31)
YEL036C	Anp1	1	1	Subunit of Golgi mannosyltransferase complex
YLR078C	Bos1	0	4	ER and Golgi Qb-SNARE protein
YBL040C	Erd2	0	5	HDEL receptor, retrieves ER resident proteins
YHR181W	Erv26	1	7	Integral membrane protein of the early Golgi apparatus and ER
YBR205W	Ktr3	0	6	Golgi-localized mannosyltransferase
YPL050C	Mnn9	3	5	Subunit of Golgi mannosyltransfease complex
YGL038C	Och1	0	3	Golgi-localized outer chain mannosyltransferase
YLR268W	Sec22	0	2	ER and Golgi R-SNARE protein
YLR026C	Sed5	1	5	cis-Golgi Qa-SNARE protein, syntaxin family
YGR172C	Yip1	0	1	Integral membrane protein
New proteins with ER vesicle-like peptide representation
YKL179C	Coy1	0	7	Golgi membrane protein with similarity to mammalian CASP
YOR307C	Sly41	0	4	Protein involved in ER-to-Golgi transport
YLR250W	Ssp120	4	8	Protein of unknown function
Secretory cargo detected in ER vesicles
YDR058W	Fet3	0	3	Plasma membrane ferro-oxidoreductase, involved in iron uptake
YMR307W	Gas1	1	5	Plasma membrane GPI-anchored β-glucanosyltransferase
YDR345C	Hxt3	4	7	Plasma membrane glucose transporter
YDR245W	Mnn10	0	3	Subunit of a Golgi mannosyltransferase complex
YOR153W	Pdr5	0	10	Plasma membrane ATP-binding cassette (ABC) transporter
YGL008C	Pma1	21	30	Plasma membrane H+-ATPase
YOR270C	Vph1	7	22	Subunit a of vacuolar-ATPase V0 domain
YJL012C	Vtc4	5	20	Vacuolar membrane polyphosphate polymerase
YDR135C	Ycf1	0	11	Vacuolar glutathione S-conjugate transporter

The three proteins with ER vesicle-like peptide representation (Coy1, Sly41, and Ssp120) are non-essential and thought to function in the early secretory pathway, but have not been localized to ER-derived vesicles before. Peptides from Coy1 (CASP Of Yeast) correspond to the ORF YKL179c that encodes an ≈80 kDa single pass type II transmembrane protein containing extensive cytoplasmic coiled-coil regions, and shares structural and sequence similarity with the mammalian family of Golgin proteins. Coy1 localizes to the early Golgi compartments at steady-state and deletion of COY1 was shown to alleviate growth defects displayed by cells harboring deletions of the genes encoding the SNARE proteins Gos1 and Sec22, which are involved in ER-Golgi transport (32). Peptides from Sly41 (Suppressor of Loss of YPT1) correspond to the ORF YOR307c that encodes a ≈51 kDa multipass membrane protein with sequence similarity to solute transporters (33–35). Sly41 localizes to the ER at steady-state, and multi-copy expression of SLY41 was shown to suppress the loss of YPT1, an essential Rab GTPase required for ER-Golgi transport. Peptides from Ssp120 (S. cerevisiae Secretory Protein) correspond to the ORF YLR250w that encodes a ≈27 kDa soluble protein containing an N-terminal signal sequence and dual calcium-binding EF-hand motifs (35–37). Ssp120 localizes to Golgi compartments and shares structural and sequence similarity with the Golgi-localized mammalian nucleobindins (38, 39).

Efficient packaging of newly identified ER vesicle proteins into COPII vesicles in vitro

The initial requirement of an authentic ER vesicle protein is selective export from ER membranes under conditions that reconstitute COPII vesicle budding (16, 19). To confirm the selective packaging of ER vesicle proteins identified by mass spectrometry, we first created strains with a C-terminal triple HA-epitope tag at the chromosomal locus of the genes encoding each of these proteins (40). Washed semi-intact cells prepared from untagged wild-type, COY1-HA, SLY41-HA, and SSP120-HA strains were used to perform reconstituted in vitro vesicle budding assays (Figure 2A). Relative packaging efficiencies were assessed in immunoblots by comparing the level of specific proteins packaged into vesicles synthesized in the presence or absence of purified COPII proteins with total lanes containing 10% of the total reactions (19). Coy1-HA (17%), Sly41-HA (33%), and Ssp120-HA (13%) were efficiently incorporated into COPII vesicles at levels comparable to the positive control Erv29 (25%, 30% and 21%, respectively), whereas the ER resident Sec61 (negative control) was not efficiently packaged. Therefore, we observed that Coy1, Sly41, and Ssp120 fulfill the initial requirement for authentic ER vesicle proteins.

Figure 2. Efficient packaging of ER-vesicle proteins into COPII-coated vesicles.

(A–C) In vitro budding assays with semi-intact cells prepared from wild-type (BY4742) cells and strains expressing Coy1-HA (A), Sly41-HA (B), or Ssp120-HA (C). One-tenth of a total reaction (T) was compared to budded vesicles produced in the absence (−) or presence (+) of COPII proteins. Tagged proteins were visualized by immunoblot with anti-HA antibody, and Sec61 (ER resident) as a negative control and Erv29 (COPII vesicle protein) as a positive control were detected using polyclonal antisera. Samples from (C) were resolved on a parallel SDS-PAGE gel and immunoblotted with anti-Ssp120 polyclonal antibody (blot labeled α-Ssp120). The HA-tagged and native forms of Ssp120 are indicated. D) In vitro budding assay with microsomes prepared from wild-type (BY4742) and ssp120Δ strains performed as in (A–C). Proteins were detected by immunoblot with polyclonal antisera against Sec61, Ssp120, and Erv29.

For the remainder of this report, we chose to further characterize the role of Ssp120 in the early secretory pathway. Further analysis of Coy1 and Sly41 will be described elsewhere. We next generated a polyclonal antibody against Ssp120 to determine the packaging efficiency of the endogenous form. Immunoblot of the wild-type and SSP120-HA budding assay samples above with the polyclonal anti-Ssp120 antibody showed native Ssp120 and Ssp120-HA were incorporated into COPII vesicles at similar levels (14% and 13%, respectively), indicating that the epitope tag does not affect ER packaging of Ssp120 (Figure 2C, blot labeled α-Ssp120). Next, we performed in vitro budding assays on microsomes prepared from wild-type and ssp120Δ strains (Figure 2D). Ssp120 was efficiently packaged (15%) in the wild-type strain as in Figure 2C, while the absence of an immunoreactive band in the ssp120Δ strain demonstrates specificity of the anti-Ssp120 antibody. Both strains displayed similar levels of Erv29 incorporation into COPII vesicles (18% for wild-type and 19% for ssp120Δ) and exclusion of Sec61, indicating that the ssp120Δ deletion does not generally affect the level of COPII vesicle budding or protein sorting at the ER. We confirmed that ER to Golgi transport is normal in ssp120Δ strains using a more sensitive in vitro transport assay that monitors the vesicle budding, tethering, and fusion stages (Figure S1) (41, 42). Consistent with our observations, previous work found the ssp120Δ mutation did not interfere with secretion, glycosylation, or enzymatic activity of the secreted acid phosphatase Pho5 (36). These results suggest that Ssp120 traffics between ER and Golgi compartments, but is not required for ER to Golgi transport in general.

Emp47 co-immunoprecipitates with Ssp120

Many ER vesicle proteins contain cytosolic sorting signals that bind directly to cytosolic COPI and COPII coat complexes for efficient vesicle packaging (10). However, Ssp120 cannot access these coat complexes because it is confined to the lumen and lacks known vesicle trafficking signals (e.g., the HDEL ER retrieval motif). Thus, we reasoned that selective COPII-vesicle packaging and steady-state Golgi localization of Ssp120 could be explained by association with a transmembrane ER vesicle protein. To test this, we performed native immunoprecipitation experiments to identify Ssp120-interacting proteins. Budding competent microsomes prepared from wild-type and SSP120-HA strains were solubilized in Triton X-100 and immunoprecipitated with monoclonal anti-HA antibodies. A protein of ~47 kDa was co-precipitated at near-stoichiometric levels with Ssp120-HA, but was not detected in the untagged wild-type strain (Figure 3A). This band was excised and identified as Emp47 by mass spectrometry. The co-immunoprecipitation of Emp47 with Ssp120-HA was confirmed by immunoblot with polyclonal anti-Emp47 antibodies (Figure 3B). As a control, the soluble protein CPY was not immunoprecipitated, indicating specificity of the Ssp120-Emp47 interaction. To rule out any potential influences of the HA-epitope tag, the immunoprecipitation was repeated using polyclonal anti-Ssp120 antibodies on solubilized microsomes from wild-type and ssp120Δ strains. As shown in Figure 3C, similar levels of Emp47 co-precipitated with endogenous Ssp120 as with Ssp120-HA.

Figure 3. Emp47 co-immunoprecipitates with Ssp120-HA.

A) Budding-competent microsomes prepared from untagged (BY4742) and SSP120-HA strains were solubilized in Triton X-100 and subjected to immunoprecipitation with anti-HA monoclonal antibodies. After washing, immunoprecipitated complexes were denatured in SDS-containing sample buffer, resolved on an SDS-PAGE gel, and visualized by silver stain. Arrows indicate bands that were excised (from a parallel colloidal Coomassie-stained gel, not shown) and identified by mass spectrometry. Asterisks indicate the position of anti-HA antibody heavy and light chains. B) Immunoprecipitations performed as in (A) comparing total input (T), unbound (UB), and IP samples. Proteins were detected by immunoblot with anti-HA antibodies (for Ssp120-HA) or polyclonal antisera against Emp47 and CPY (negative control). C) Triton X-100 solubilized microsomes from wild-type (BY4742) and ssp120Δ strains were immunoprecipitated with anti-Ssp120 polyclonal antiserum and analyzed as in (B). Gas1, Och1, and Yet3 were included as negative controls. Asterisk indicates anti-Ssp120 heavy chain.

Emp47 is a non-essential, single pass type I transmembrane protein with a larger N-terminal lumenal portion that contains the carbohydrate recognition domain (21, 23). The C-terminal cytoplasmic tail sequence of Emp47 contains COPI and COPII sorting signals that cycle the protein between ER and Golgi compartments with a predominant localization to the early Golgi under steady-state conditions (19, 21, 22). Emp47 is thought to function in sorting a subset of glycoproteins into COPII vesicles for export from the ER (21, 22) and the protein assembles into homo-oligomers as well as heter-oligomers with its paralog Emp46 (43). Additional Ssp120-HA immunoprecipitation experiments revealed co-precipitation of Emp46, but the interaction between Emp47 and Ssp120-HA was not affected in emp46Δ cells (Figure S2). Moreover both Emp47 and Ssp120 are efficiently packaged into COPII vesicles in the absence of Emp46 (Figure S2). We next focused our analyses on the functional relationships between Emp47 and Ssp120.

Ssp120 requires Emp47 for proper localization

Emp47 is homologous to the mammalian cargo receptor ERGIC53, which is required for the ER export of a subset of glycoproteins including the secreted blood coagulation factors V (FV) and VIII (FVIII) (11, 12, 44, 45). ERGIC53 requires a soluble co-factor with two calcium-binding EF-hand domains, MCFD2, for the efficient transport of FV/FVIII from the ER to the ERGIC (46). The ERGIC53-MCFD2 complex is localized to the ERGIC at steady-state in a calcium-dependent manner, and ERGIC53 is required for the proper localization of MCFD2 (46, 47). While MCFD2 and Ssp120 share little sequence similarity, they share a nearly identical domain organization and lack a C-terminal KDEL/HDEL retrieval signal. The parallels between Emp47-Ssp120 and ERGIC53-MCFD2 have also been noted by Babu et al., (24). These observations combined with previous results suggesting Emp47 is required for secretion of a subset of glycoproteins led us to investigate if the Emp47-Ssp120 complex serves as a functional analog of the ERGIC53-MCFD2 complex (22).

Initial immunoblot analysis comparing wild-type and emp47Δ strains revealed that steady-state levels of Ssp120 were sharply reduced in emp47Δ cells (~16% of wild-type) (Figure 4A). To further explore the functional relationship between Emp47 and Ssp120, we performed in vitro COPII vesicle budding assays using microsomes prepared from wild-type, ssp120Δ, and emp47Δ cells to determine subunit interdependence (Figure 4B). In comparison to wild-type microsomes, the level of Ssp120 was greatly reduced in emp47Δ microsomes, which required an extended exposure to detect (Figure 4B, lanes labeled T). Importantly, the low level of Ssp120 remaining in emp47Δ microsomes was not efficiently packaged into COPII vesicles. In contrast, both overall levels and vesicle packaging efficiency of Emp47 were similar in wild-type and ssp120Δ cells. As expected, the ER resident Sec61 was not packaged whereas the ER vesicle protein Erv29 was efficiently packaged for all strains. These results indicate that ER-Golgi cycling and maintenance of steady-state levels of Ssp120 require Emp47, but loss of Ssp120 does not affect localization or trafficking of Emp47.

Figure 4. Ssp120 requires Emp47 for proper localization.

A) Immunoblot analysis of the steady-state levels of Ssp120 and Emp47 in wild-type (BY4742), ssp120Δ, emp47Δ, and emp46Δ strains. Whole-cell lysates were prepared as described in Materials and Methods, and equal amounts resolved on an SDS-PAGE gel. Proteins were detected with polyclonal anti-sera against Cdc48 (control), Och1 (control), Emp47, and Ssp120. B) In vitro budding assay with microsomes prepared from wild-type (BY4742), ssp120Δ, and emp47Δ strains performed as described in the legend of Figure 2A. Proteins were detected with polyclonal anti-sera against Emp47, Sec61 (negative control), Ssp120, and Erv29 (positive control). The panel labeled ‘(dark)’ shows a saturated exposure of the anti-Ssp120 blot to visualize the low levels of protein in the emp47Δ background. C) Amount of Ssp120-HA secreted after 6 hr in wild-type (BY4742), SSP120-HA, and emp47Δ SSP120-HA strains was detected by immunoblotting whole cell lysates (“Intracellular”) and TCA-precipitated culture medium (“Extracellular”) with anti-HA antibody for Ssp120-HA and polyclonal antisera against Kar2 as a positive control and Sar1 as a negative control. D) Immunoblot analysis of the influence of a PEP4 deletion on steady-state levels of Ssp120 and Emp47 in wild-type (BY4742), emp47Δ, and ssp120Δ strains performed as in (A). Proteins were detected with polyclonal anti-sera against Cdc48 (loading control), CPY (control for pep4Δ), Emp47, and Ssp120. The ER (p1), Golgi (p2), and mature (m) forms of CPY are indicated.

MCFD2 is secreted in ERGIC53-depleted mammalian cells, though a fraction may be degraded in the lysosome (equivalent to the yeast vacuole) (47). Thus, we sought to determine the fate of Ssp120 in emp47Δ cells. To assess secretion from cells, log-phase wild-type (untagged) and SSP120-HA cells with or without emp47Δ deletion were washed and back-diluted into fresh YPD. After 6 hr of growth at 30°C, cells were pelleted and the culture supernatant subjected to TCA-precipitation. Whole-cell lysates and TCA-precipitated culture medium were analyzed by SDS-PAGE and immunoblot. In emp47Δ cells, the reduction of intracellular Ssp120-HA corresponded to a proportional increase of extracellular secreted Ssp120-HA as compared to the EMP47 strain (Figure 4C). Interestingly, a low level of Ssp120-HA was detected in the extracellular fraction of EMP47 cells, and may reflect a normal low level of secretion from wild type strains. As controls, the ER lumenal protein Kar2, which is also secreted at low levels (48), was detected at similar levels for all strains, and the cytoplasmic protein Sar1 was not detected in the extracellular fraction to exclude cell lysis.

During the course of our study, large-scale protein mapping approaches identified and partially characterized the Emp47-Ssp120 interaction (17, 24). Ssp120-GFP was observed to mislocalize to vacuoles by fluorescence microscopy in emp47Δ cells and Ssp120-HA was secreted from emp47Δ vps10Δ double mutant strains. To assess the level of Ssp120 degradation in vacuoles, we examined steady-state levels of Ssp120 in wild-type, ssp120Δ, and emp47Δ strains alone or in combination with a PEP4 deletion. Pep4 is required for activation of most vacuolar proteases, therefore vacuolar degradation is deficient in pep4Δ cells (49). As shown in Figure 4D, the low Ssp120 levels in emp47Δ cells were increased by combining with the pep4Δ mutation but were not restored to wild-type levels, indicating that only a fraction of mislocalized Ssp120 traffics to the vacuole while most is secreted from cells. In contrast, the levels of Emp47 were unaffected by the ssp120Δ and pep4Δ mutations, singly or in combination. Because conditions that perturb Emp47 cycling between the ER and Golgi result in its mis-localization to the vacuole (21, 50), this provides additional evidence that Ssp120 does not influence Emp47 localization or ER-Golgi cycling. Based on these results we conclude that in the absence of Emp47, Ssp120 exits the ER via a bulk flow process and accesses later Golgi compartments where a major fraction is secreted and the remainder traffics to the vacuole for degradation. The strict requirement of Emp47 for Ssp120 localization, and the independence of Emp47 on Ssp120 for proper localization and ER-Golgi cycling, mirrors phenotypes observed in mammalian cells for the ERGIC53-MCFD2 complex and supports a model in which the Emp47-Ssp120 complex is functionally analogous (46, 47).

Phenotypic analysis of ssp120Δ and emp47Δ strains

To understand the functional relevance of the Emp47-Ssp120 interaction, we analyzed the phenotypes of ssp120Δ, emp47Δ, and double deletion mutant strains. The emp47Δ deletion strain was previously shown to have temperature and calcium sensitivities in certain genetic backgrounds (21, 22). We could reproduce these phenotypes in the W303a as well as the YPH500/499 backgrounds used previously (22), but did not observe these sensitivities in the BY4741/4742 or the FY833/834 backgrounds (Figure S3). Therefore, we used the W303a background for further phenotypic analysis.

The ssp120Δ, emp47Δ, and double deletion strains grew comparable to wild-type on YPD at 30°C, but displayed severe growth defects at 37°C (Figure 5A). Notably, combining the ssp120Δ and emp47Δ deletions did not exacerbate the growth defects of the single deletion mutations. The absence of synthetic negative effects in the ssp120Δ emp47Δ double deletion supports a model in which Ssp120 and Emp47 function as a complex. It is possible that overexpression of either subunit could partially complement for function and suppress the temperature sensitive phenotype. To address this, we tested whether multi-copy expression of EMP47 could rescue the ssp120Δ growth defect. As shown in Figure 5B, only multi-copy expression of SSP120 suppressed the ssp120Δ growth defect at 37°C, while ssp120Δ temperature sensitive growth was not suppressed in strains transformed with empty vector or expressing multi-copy EMP47. This result is consistent with the observation that the ssp120Δ emp47Δ double deletion strain has no additive growth defect and indicates that both subunits of the Emp47-Ssp120 complex are required for function. We also noted that shifting the ssp120Δ and emp47Δ deletion strains to a restrictive temperature produced no detectable block in CPY maturation or alterations in ER-Golgi morphology (Figure S4). These observations indicate the Emp47-Ssp120 complex is not required for overall function of the secretory pathway but are more consistent with a role in biogenesis and delivery of specific cargo to their proper cellular location. Deficiencies in cargo delivery are likely the cause of thermosensitivity in ssp120Δ and emp47Δ cells.

Figure 5. Growth comparison of ssp120Δ and emp47Δ strains.

A) Temperature sensitivity of the emp47Δ and ssp120Δ strains. Wild-type (W303a), emp47Δ, ssp120Δ, and emp47Δ ssp120Δ strains were grown to saturation in YPD at 30°C and adjusted to an OD₆₀₀ of 1.0. A 10-fold dilution series was spotted onto YPD and incubated at the indicated temperatures for 2 days. B) Overexpression of Emp47 does not suppress the temperature sensitivity of the ssp120Δ strain. Wild-type (W303a) and ssp120Δ strains carrying empty vector (pRS425) or multicopy EMP47 or SSP120 (pNM23 or pNM17) expression plasmids were grown to saturation in SC-Leu at 30°C and adjusted to an OD₆₀₀ of 1.0. A 10-fold dilution series was spotted onto SC-Leu and incubated at the indicated temperatures for 3 days. C) Calcium sensitivity of the emp47Δ and ssp120Δ strains. Wild-type (W303a), emp47Δ, ssp120Δ, and emp47Δ ssp120Δ strains were prepared as in (A) and spotted onto YPD or YPD containing 0.25 M or 0.5 M CaCl₂ and incubated for 2 days at 30°C. Incubation of the 0.5M CaCl₂ plate was extended to 7 days to rule out masking effects from slow growth (bottom right panel).

We next asked if the ssp120Δ deletion influenced the Ca²⁺ sensitivity of an emp47Δ strain. To do this, wild-type, ssp120Δ, emp47Δ, and double deletion strains were grown on YPD supplemented with 0.25 M or 0.5 M CaCl₂ (Figure 5C). As with growth at 37°C, growth on 0.5 M CaCl₂ caused severe growth defects in ssp120Δ and emp47Δ deletion strains, but no observable synthetic effect in the double deletion strain as previously reported (24). These results are again consistent with the hypothesis that both subunits are required for a functional complex. The strong calcium phenotype also provided a means of assessing whether the HA-epitope tagged version of Ssp120 was fully functional. Growth of the SSP120-HA strain was indistinguishable from wild-type on 0.5 M CaCl₂, while ssp120Δ showed the expected growth defect (Figure S5).

Deletion of SSP120 and EMP47 suppresses the DTT sensitivity of an ire1Δ mutant

Emp47 is generally thought to function as a glycoprotein cargo receptor in the anterograde pathway based on its requirement for the ER-Golgi transport of Emp46, though attempts to identify Emp47-dependent secretory cargo have been unsuccessful (17, 22, 43). In searching the BioGRID database, we found multiple interactions that suggested this complex may influence glycoprotein quality control (51). Emp47 co-purified in complex with the ER quality control lectin Yos9 along with other members of the core ER-associated degradation (ERAD) machinery (52). Emp47 and Ssp120 are not glycosylated and thus not likely to be Yos9 substrates (21, 36). Deletion of EMP47 was shown to alleviate the increased activation of the unfolded protein response (UPR) in cells with a hypomorphic allele of KAR2 (BiP) (53). The UPR is a program of transcriptional up-regulation of ~400 genes induced by conditions that cause severe ER stress (54). In addition, a genome-wide synthetic genetic array (SGA) identified positive interactions between SSP120 and the glycoprotein-specific Kar2 co-chaperone SCJ1 and the UPR transducer IRE1 (55, 56). These interactions are not due to the influence of SSP120 and EMP47 on the UPR because neither gene is a UPR target, nor do their deletion mutations induce the UPR, in contrast to most other known ER-Golgi anterograde cargo transporters (29, 31, 54, 57). The collective results indicate that loss of SSP120 and EMP47 can alleviate ER stress in a UPR-independent manner.

To further explore the connection between the Emp47-Ssp120 complex and ER stress, we examined the effects of ssp120Δ and emp47Δ mutations on cell growth under UPR-inducing conditions in wild-type and ire1Δ backgrounds. Because Ire1 is the sole transducer of the UPR in yeast cells, conditions of severe ER stress that activate the UPR are lethal to ire1Δ cells, thus potential masking effects are removed in this background (58). Growth of wild-type (BY4742) or ire1Δ cells in combination with ssp120Δ or emp47Δ mutations was assessed on YPD supplemented with either the reducing agent DTT or the N-glycosylation inhibitor tunicamycin, both known to induce the UPR (Figure 6A). Under both conditions, growth of the ssp120Δ and emp47Δ strains was comparable to wild-type cells, while the ire1Δ strain showed the expected growth defects. Interestingly, ssp120Δ or emp47Δ rescued the growth defect of ire1Δ cells on DTT, but not on tunicamycin. In contrast, we did not detect suppression of ire1Δ sensitivity to DTT when combined with emp46Δ (not shown). These results suggest that ssp120Δ and emp47Δ mutations alleviate ER stress in the presence of widespread protein misfolding induced by reducing agents in an N-glycosylation dependent manner.

Figure 6. Effects of ssp120Δ and emp47Δ mutations on ire1Δ cell growth under UPR-inducing conditions.

A) SSP120 and EMP47 deletions suppress sensitivity of ire1Δ cells to DTT, but not tunicamycin. Wild-type (BY4742), ssp120Δ, ire1Δ, ssp120Δ ire1Δ, emp47Δ, and emp47Δ ire1Δ strains were prepared as described in Figure 5A and grown at 30°C on YPD, YPD with 2.5 mM DTT for 1 day (left panels) or YPD with 2 µg/mL tunicamycin for 2 days (right panels). The experiments with the ssp120Δ strain set (top panels) and emp47Δ strain set (bottom panels) were conducted on separated occasions. B) SSP120 and EMP47 deletions do not suppress sensitivity of ire1Δ cells to overexpression of CPY*. Wild-type (BY4742), ssp120Δ, ire1Δ, ssp120Δ ire1Δ, emp47Δ, and emp47Δ ire1Δ strains carrying a plasmid with CPY* under control of the GAL1 promoter (pES28) were prepared as described in Materials and Methods and a 10-fold dilution series plated onto selective medium with 2% glucose (no expression), 2% galactose (overexpression), or a mixture (attenuated expression) and grown at 30°C for 4 days.

Why might the loss of the Emp47-Ssp120 complex benefit cells during severe ER stress? Under normal growth conditions, misfolded glycoproteins are proposed to traffic from the ER to the Golgi at a low rate, followed by retrieval to the ER for additional rounds of folding or for degradation (59–64). The mammalian Emp47 homolog and ERGIC53 family member VIP36 is proposed to function in this retrieval pathway, which in combination with the observations above raised the possibility that in yeast the Emp47-Ssp120 complex may also act in a post-ER quality control retrieval pathway (65–67). In this scenario, loss of Emp47-Ssp120 function could alleviate DTT-induced toxicity in an ire1Δ background by permitting clearance of misfolded glycoproteins from the ER for delivery and degradation in the vacuole (63, 68, 69).

To address this question, we first examined if the turnover rate of a model ERAD substrate such as CPY* (70) was altered in the ssp120Δ and emp47Δ strains. We reasoned that if the Emp47-Ssp120 complex acted in retrieval of misfolded glycoproteins, then turnover rates would be delayed because escaped CPY* would slowly transit later branches of the secretory pathway before degradation in the vacuole (68). However, under the conditions of our analysis we did not detect changes in CPY* turnover rates in the ssp120Δ or emp47Δ strains (Figure S6). To rule out cargo specific effects, we examined turnover rates of other model ERAD substrates and did not observe delays in Gas1* (Figure S7) or PrA* (not shown) degradation rates in ssp120Δ or emp47Δ backgrounds. We also considered the possibility that the fraction of ERAD substrate that escaped the ER quality control machinery and depended on post-ER retrieval would be quite small and undetectable by this method. Therefore we used an alternative approach to saturate the ERAD pathway through GAL1-regulated overexpression of CPY*, which has been shown to induce the UPR and produce ER toxicity in ire1Δ cells (68, 70, 71). If the Emp47-Ssp120 complex retrieves misfolded glycoproteins from Golgi compartments we expected their deletion to partially suppress toxicity of overexpressed CPY* in an ire1Δ sensitized background. However, we observed that CPY* overexpression in the ssp120Δ ire1Δ and emp47Δ ire1Δ double deletion strains grew as poorly as the ire1Δ strain, while growth of the ssp120Δ and emp47Δ strains was comparable to wild-type (Figure 6B). These findings indicate that the Emp47-Ssp120 complex does not function in retrieval of misfolded glycoproteins from Golgi compartments as proposed for other ERGIC53 family members.

Synthetic Genetic Array Analysis of emp47Δ

To gain further insight into Emp47 cellular function, we performed an SGA analysis (72) in which the emp47Δ mutation was combined individually with all other non-essential gene deletions (~4,900 yeast strains) and screened for growth phenotypes. After comparing growth of the double deletion strains on YPD medium at 30°C, we identified 29 gene deletions that produced synthetic negative growth phenotypes when combined with emp47Δ in three independent rounds of screening (Table S2). These gene interactions were ranked based on colony size with more negative scores indicating stronger synthetic negative interactions (73).

Top hits from this screen were independently tested by constructing double deletion mutants in the BY4742 strain background. We observed that the emp47Δ gas1Δ and emp47Δ mnn11Δ strains displayed synthetic negative growth phenotypes compared to the individual deletion mutants (Figure 7), whereas combinations with yta12Δ and ste50Δ did not produce detectable growth phenotypes under our standard conditions. Interestingly both Gas1 and Mnn11 influence cell wall assembly in yeast. Gas1 is an abundant 1,3-β-glucanosyltransferase that is GPI anchored at the cell surface and catalyzes formation of the yeast cell wall (74, 75). Mnn11 is a subunit of the major Golgi-localized α1,6-mannosyltransferase complex that elongates core-type N-glycans necessary for biogenesis of cell wall components (76, 77). Importantly, pairwise combination of ssp120Δ with gas1Δ or mnn11Δ also resulted in synthetic negative growth phenotypes that were indistinguishable from the emp47Δ double mutants and provide further evidence that Emp47 and Ssp120 operate as a functional unit. These collective findings support a model in which the Emp47-Ssp120 complex acts in efficient anterograde transport of cell surface glycoproteins necessary for cell wall integrity.

Figure 7. EMP47 and SSP120 display synthetic negative interaction with GAS1 and MNN11.

A) A 10-fold dilution series of wild-type (BY4742), emp47Δ, ssp120Δ, gas1Δ, gas1Δ emp47Δ (2 independent isolates), and gas1Δ ssp120Δ (2 independent isolates) cells were spotted onto YPD plates and grown at 30°C and 37°C for 2 days. B) Similarly, wild-type (BY4742), emp47Δ, ssp120Δ, mnn11Δ, mnn11Δ emp47Δ (2 independent isolates), and mnn11Δ ssp120Δ (2 independent isolates) cells were spotted onto YPD plates and grown at 30°C and 37°C for 2 days.

Discussion

A comprehensive proteomic analysis of purified COPII vesicles provided an inventory of transport machinery and secretory cargo packaged into these trafficking intermediates. Several of the identified polypeptides corresponded to previously characterized vesicle proteins that link secretory cargo to coat subunits during the ER budding stage or proteins that operate in vesicle targeting and membrane fusion (10, 19, 31). Three newly identified polypeptides exhibited the properties of other ER vesicle proteins but are of poorly defined function. We confirmed that Coy1, Sly41 and Ssp120 are efficiently packaged into COPII vesicles and given their subcellular distribution we propose these proteins function in transport through the early secretory pathway. We further characterized the Ssp120 protein in this report.

Purification of Ssp120 revealed an apparent stoichiometric complex with Emp47, a membrane protein known to cycle between the ER and Golgi and thought to transport glycoproteins in a manner similar to its mammalian homolog ERGIC53. Emp46 was detected as a minor component of this complex. In emp47Δ cells, most Ssp120 was secreted into the extracellular media indicating a requirement for Emp47 in Ssp120 localization. We observed that deletion of emp47Δ or ssp120Δ produced identical calcium and temperature sensitive phenotypes further supporting function as complex. No detectable trafficking or growth phenotypes were associated with the emp46Δ cells. SGA analysis revealed synthetic negative interactions between emp47Δ and genes that encode abundant cell wall proteins or machinery needed for proper glycosylation of cell wall components. The ssp120Δ deletion displayed the same pattern of synthetic genetic interactions. Based on these findings we conclude that both Emp47 and Ssp120 are required for this complex to function, and that the Emp47-Ssp120 complex acts as a shuttling cargo receptor in the early secretory pathway for efficient biogenesis and transport of cell surface glycoproteins. Emp46 may play a role as a cargo specific adaptor of the Emp47-Ssp120 complex.

During the course of our studies, two high-throughput reports also identified and partially characterized interactions between Emp47 and Ssp120. Herzig and colleagues reported mislocalization of Ssp120-GFP in microscopy screens to pair potential cargo receptors, including Emp47, with specific cargo proteins (17). In addition, Babu et al. identified Ssp120 associated with Emp47 and Emp46 in a global analysis of yeast membrane proteins by affinity purification and mass spectrometry analysis (24). Moreover, the investigators confirmed these interactions through tagging and immunoprecipitation approaches and demonstrated calcium sensitivity of the ssp120Δ strain. Interestingly, Ssp120-GFP was observed to accumulate in the vacuole of emp47Δ cells and Ssp120-HA was secreted from cells when the vacuole sorting receptor Vps10 was deleted in an emp47Δ background (24). Using different assays we found that most Ssp120 was secreted from emp47Δ cells and that a smaller fraction reaches the vacuole where it can be stabilized in a pep4Δ background. Our current work is largely consistent with and extends these published findings.

Our observation that combining the emp47Δ or ssp120Δ deletion with ire1Δ suppressed sensitivity to DTT is intriguing. In contrast, other anterograde cargo receptors display synthetic negative interactions with ire1Δ as buildup of cargo in the ER is thought to be managed by an activated UPR (29, 55, 57). In addition, our results indicate that Emp47-Ssp120 does not function in the clearance of misfolded proteins such as CPY* from the early secretory pathway to alleviate ER stress. However, we note that genome wide analyses turned up both emp47Δ and ssp120Δ as constitutive activators of the yeast cell wall integrity pathway (78). Here, phosphorylation of the yeast mitogen-activated protein kinase (MAPK) Slt2 was increased 4 to 5-fold in emp47Δ and ssp120Δ cells. Activated Slt2 turns on a transcriptional program that increases expression of cell wall proteins and biogenesis machinery (79). Additional studies have shown that activated Slt2 helps manage ER stress in processes that are both Ire1-dependent (80) and independent (81, 82). Cell wall integrity issues in emp47Δ and ssp120Δ strains are also consistent with our SGA screen, which identified synthetic negative interactions with genes involved in cell wall assembly. Based on these collective findings, we propose that cells lacking Emp47-Ssp120 complex have only a low level of nascent secretory proteins accumulating in the ER but that inefficient delivery of cell surface glycoproteins diminishes cell wall integrity. To compensate, these cells constitutively activate Slt2 signaling networks, which partially overlap with targets of the unfolded protein response pathway to alleviate sensitivity to DTT in the absence of Ire1 function. However, we could not detect clear phenotypic consequences from combining slt2Δ with emp47Δ and further studies are needed to support this model.

How does Emp47-Ssp120 complex function compare to the mammalian ERGIC53-MCDF2 complex? Emp47 shares low sequence identity (18.4%) but overall high structural conservation with ERGIC53 (23, 83). These proteins belong to the L-type lectin family, which fold into β-sandwich structures and contain defined carbohydrate recognition domains (83, 84). However, there are distinct differences between Emp47 and ERGIC53 in that Emp47 does not appear to bind Ca²⁺ directly and a conserved histidine residue involved in pH-regulated binding of N-linked glycans to ERGIC53 is absent in Emp47 (23, 83, 85). Moreover, the putative Ca²⁺ binding protein Ssp120 and the known calcium binding protein MCFD2 associated with Emp47 and ERGIC53, respectively, are also distinct. While both proteins contain hydrophobic signal sequences and dual EF-hand motifs, the 234 amino acid Ssp120 protein and 146 amino acid MCFD2 protein share low sequence identity overall. In fact, Ssp120 shares higher sequence identity (31% over 95 residues) with nucleobindin-1, another Golgi-localized calcium-binding protein (39). Therefore it is possible that the Emp47-Ssp120 complex is more analogous to other ERGIC53 family members such as Vip36 and VIPL (86), which could function in complex with associated nucleobindin proteins. A recent chemical proteomic analysis also revealed a role for nucleobindin-1 in lipid metabolism, possibly functioning as a carrier of specific fatty acid amides (87). Thus an alternative model might be that these complexes traffic lipids, which could also explain altered cell wall properties in yeast emp47Δ and ssp120Δ mutants.

In humans, the calcium-binding MCFD2 subunit is thought to act as a cargo specificity factor to bind the blood clotting factors V and VIII to ERGIC53 for their efficient export from the ER (88). Whereas other secretory cargo, including cathespin Z and cathepsin C, appear not to depend on the MCFD2 subunit (47). Biochemical and structural studies have shown that high-mannose oligosaccharides bind directly to the ERGIC53 carbohydrate recognition domain (CRD) in the absence of MCFD2 (86, 89, 90) leading to a model where the CRD of ERGIC53 binds N-linked glycans attached to nascent secretory proteins and the MCFD2 subunit interacts with nearby peptide residues within specific cargo proteins (89). However, structural studies have not confirmed the molecular basis of this dual interaction. Both pH and Ca²⁺ concentrations influence cargo interactions with ERGIC53-MCFD2 and are thought to promote cargo binding in the ER and release in intermediate or early Golgi compartments (85, 91, 92). In yeast, phenotypes of the emp47Δ and ssp120Δ deletion strains are indistinguishable; therefore, we propose that this complex acts as a functional unit for ER export of specific cell surface glycoproteins. Compartmental pH and/or Ca²⁺ levels are also likely to regulate cargo binding to Emp47-Ssp120 in yeast. Studies guided by this simplified model should allow for a full mechanistic dissection of how this family of L-type lectins with their putative calcium-binding subunits can recognize glycoproteins in the ER and then release bound cargo in post-ER compartments.

Materials and Methods

Yeast strains and media

Yeast strains used in this study are listed in Table S3. Standard yeast methods were used (93). DNA for plasmid constructions and yeast transformation was generated using the Phusion High-Fidelity PCR system (Thermo Scientific, Rockford, IL). Oligonucleotide primers (IDT DNA, Coralville, IA) are listed in Supplemental Table S4. Yeast cells were transformed by the high-efficiency lithium acetate technique (94).

BY4742-background deletion mutants containing the kanMX4 cassette were purchased from Research Genetics (Invitrogen, Carlsbad, CA). C-terminal 3xHA-epitope tagged strains and deletion mutants containing the HIS3MX6 cassette were created using described methods (40). To generate CBY3296 and CBY3328, the emp47Δ::kanMX4 cassette and flanking regions were PCR amplified from CBY1299 with primers NMp30 and NMp31 and transformed into W303a and SEY6210, respectively. Strains in which the SEC7 and SEC21 genes were replaced with SEC7-GFPx3 and SEC21-GFPx3, respectively, were constructed as previously described (95). The above mutations were confirmed by immunoblot and colony PCR (check primer sequences available upon request). The pep4Δ::URA3 mutation in CBY3564, CBY3565, and CBY3566 was achieved using linearized plasmid pTS15 as previously described (96). Deletion of PEP4 in these strains was confirmed by immunoblot for CPY processing.

Unless otherwise noted, cells were grown at 30°C in rich medium (YPD: 1% yeast extract, 1% peptone, and 2% dextrose) or in synthetic complete medium (SC: 0.67% yeast nitrogen base without amino acids, 2% dextrose, and appropriate amino acid supplement mixture [CSM; MP Biomedicals, Solon, OH]). SR and SG media are SC except with 3% raffinose or 2% galactose, respectively, instead of 2% dextrose. For growth of W303a–background strains, media was supplemented with 40 µg/mL adenine. For overexpression of CPY*HA from the GAL1/10 promoter (pES28) for the growth assay shown in Figure 6B, strains were grown at 30°C to saturation in SR-Ura and adjusted to an OD₆₀₀ of 1.0. A 10-fold dilution series was spotted onto SC-Ura, SG-Ura, or SC-Ura with 2% galactose and 0.2% dextrose and incubated at 30°C for 3 days.

Plasmids

Plasmids used in this study are listed in Supplemental Table S5. The sequences for EMP47 (YFL048c) and SSP120 (YLR250w) were from the Saccharomyces Genome Database (http:/www.yeastgenome.org/). The EMP47 coding sequence and ~440 bp each upstream and downstream regions were amplified from BY4742 genomic DNA with primers NMp28 and NMp29. EcoRI and HindIII sites appended by the primers were used to clone this fragment into the same sites on pRS316 (CEN, URA3) (97) to generate pNM4 as an intermediate step. Plasmid pNM4 was digested with SpeI and HindIII (both cleave in the polylinker of pRS316) and the fragment containing EMP47 subcloned into pRS425 (2µ, LEU2) (98), creating plasmid pNM23. The SSP120 coding sequence and ~450 bp flanking regions were amplified from BY4742 genomic DNA and primers NMp24 and NMp66. NotI and BamHI sites introduced by the primers were used to insert this fragment into the same sites on pRS313 (CEN, HIS3) (97) to create plasmid pNM13 as an intermediate step. Plasmid pNM13 was digested with NotI and SalI (both cleave in the pRS313 polylinker) and the fragment containing SSP120 was subcloned into the same sites on pRS425 (2µ, LEU2) (98).

To construct the 6xHis-tagged N-terminal fusion protein with mature Ssp120 (Ssp120Δss) for bacterial expression, the SSP120 coding sequence after the signal peptide (a.a. 23–234) was amplified from plasmid pNM13 with primers NMp81 and NMp82. These primers introduced a silent T to C mutation in nucleotide 84 to destroy an endogenous AseI site, and appended 5’ AseI and 3’ BamHI sites for insertion into the NdeI and BamHI sites of pET-15b (Novagen, Billerica, MA) to create plasmid pNM18 (AseI and NotI have compatible ends that ligate in-frame). All generated constructs were verified by restriction digest and sequencing. Plasmids pDN436 (71), pES28 (68), and pJK59 (99) were previously described. Briefly, pDN436 encodes HA epitope-tagged CPY* under control of the PRC1 promoter in pRS315, pES28 encodes CPY*HA under control of the GAL1/10 promoter in YCp50, and pJK59 (CEN, URA3) encodes Sec63-GFP under control of the SEC63 promoter.

Antibodies

Polyclonal antibodies were raised against recombinant 6His-Ssp120Δss expressed from pNM18 in E. coli C43(DE3) cells (Lucigen, Middleton, WI). After a 3 h induction with 1 mM isopropyl β-D-thiogalactoside, the fusion protein was purified from the soluble extract on a Ni-NTA column (Qiagen, Valencia, CA) as recommended by the manufacturer. Purified protein was used to immunize rabbits by standard procedures (Covance, Denver, PA). For immunoblotting, anti-Ssp120 antiserum was diluted to 1:2000. Polyclonal antibodies against Sec61 (100), Erv29 (15), carboxypeptidase Y (CPY) (101), Emp47 (21), Gas1 (102), Och1 (19), Yet3 (103), Cdc48 (104), Sar1 (105), and Kar2 (106) were described earlier. Monoclonal anti-HA (HA.11) antibody was obtained from Covance (Princeton, NJ).

Yeast cell lysates and immunoblotting

To assess steady-state protein levels, 2 OD₆₀₀ equivalent of mid-log cells was harvested by low-speed centrifugation and the pellets flash-frozen in liquid nitrogen. Cell pellets were thawed on ice and resuspended in 200 µL of 20 mM NaN₃ with 1 mM phenylmethylsulfonyl fluoride. Glass beads (BioSpec, Bartlesville, OK) and 200 µL 5× SDS-PAGE sample buffer were added before lysis at 4°C using a Mini-Beadbeater-16 (BioSpec). Samples were heated for 5 min at 95°C, cleared by centrifugation for 3 min at 14,000 × g, and resolved by SDS-PAGE. Immunoblots were developed with Supersignal Pico chemiluminescent substrate (Pierce Chemical, Rockford, IL), imaged using a G:BOX Chemi XR5 (Syngene, Frederick, MD), and quantified with GeneTools image analysis software (Syngene). Statistical analyses were performed with Prism (Graphpad, La Jolla, CA) or Excel (Microsoft, Redmond, Wa).

In vitro vesicle budding and transport assays

Preparative scale synthesis of COPII vesicles was performed by incubating microsomes derived from wild-type (BY4742) cells with ATP and GTP in the presence (+COPII) or absence (−COPII) of purified Sar1, Sec23-Sec24 complex and Sec13-Sec31 complex as described previously (19, 25, 26). Budding reactions (300 µL volume) were subjected to centrifugation at 4°C for 3 min at 14,000 × g to separate vesicles from heavier donor membranes. Vesicles contained in this medium-speed supernatant retain a substantial amount of COPII proteins that could interfere with the identification of less-abundant ER vesicle proteins by mass spectrometry. To strip COPII proteins from the vesicles by flotation through a Nycodenz gradient, 270 µL of this medium-speed supernatant was mixed with 400 µL of 60% Nycodenz (in D₂O) and transferred to the bottom of a TLS-55 centrifuge tube (Beckman Coulter, Danvers, MA). Next, 450 µL each of 25% and 20% Nycodenz (in D₂O with 20 mM HEPES pH 7.5, 150 mM potassium acetate, and 5 mM magnesium acetate) were layered on top, followed by 750 µL of buffer B88 (20 mM HEPES pH 7.0, 250 mM sorbitol, 150 mM potassium acetate, and 5 mM magnesium acetate). The gradients were centrifuged at 4°C for 3 h at 166,000 × g in a TLS-55 rotor. To detect the peak vesicle-containing fractions, budding reactions were initially performed in the presence of in vitro-produced ³⁵S-labeled pro-α-factor. The top 250 µL of each gradient was discarded, and 100 µL fractions collected and radioactivity measured to determine the α-factor peak. A COPII-dependent peak of radioactivity was reproducibly observed in fraction 7, with activity in fractions 5 to 8.

For en bloc mass spectrometry, the addition of ³⁵S-labeled pro-α-factor was omitted and the peak α-factor fractions (5 to 8) determined above were collected. These fractions were pooled, mixed with 800 µL of B88, and the vesicles were pelleted by centrifugation at 4°C for 15 min at 152,000 × g in a TLA-100.3 rotor (Beckman Coulter). Pellets were solubilized in 40 µL of 5× sample buffer by heating for 4 min at 75°C, and then resolved 17 mm (measured from bottom of well to dye front) into a pre-cast 12% Novex Tris-glycine gel (Invitrogen). Proteins were stained with a Colloidal Blue Staining Kit according to the manufacturer’s instructions (Invitrogen), and the sample lanes were cut into three sections corresponding to 95–250 kDa, 55–95 kDa, and 55 kDa to the dye front as shown in Figure 1B. These gel sections were submitted to the Taplin Mass Spectrometry Facility (Harvard Medical School, Boston, MA) for peptide sequence analysis by microcapillary LC/MS/MS.

Analytical scale budding reactions to assess COPII-dependent packaging efficiency by immunoblot from semi-intact cells and microsomes incubated in the presence or absence of purified COPII components were performed as described (19, 25). The in vitro assay to measure overall ER-to-Golgi transport efficiency following ³⁵S-glyco-pro-α-factor (Supplementary Figure S1) was performed as previously described (41). The data plotted in this experiment are the average of duplicate determinations and the error bars represent the range.

Immunoprecipitation experiments

For native immunoprecipitations, 200 µL of microsomes (0.25 mg of total membrane protein) were solubilized on ice with 1 mL of IP buffer (15 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride) for 10 min. Triton-insoluble material was removed by centrifugation at 4°C for 5 min at 14,000 × g, and 1 mL of cleared supernatant (total input) was combined with 30 µL of 20% protein A-Sepharose (GE Healthcare, Piscataway, NJ) and 1 µg of anti-HA.11 antibody (Covance) or 1 uL anti-Ssp120 antiserum for each immunoprecipitation. The reactions were incubated for 2 h at 4°C with rotation, and the unbound protein fraction (1 mL) was discarded. The precipitates were washed five times with cold IP buffer and were eluted from the beads by adding 30 µL of 5× sample buffer and heating for 5 min at 75°C. Samples were resolved by SDS-PAGE and proteins visualized by immunoblot or silver stain. For identification of the Ssp120-HA–binding protein, native immunoprecipitations were performed as above except using 0.40 mg of microsomes. Samples were resolved on a pre-cast 10% Novex Tris-glycine gel (Invitrogen) and proteins visualized using a Colloidal Blue Staining Kit according to the manufacturer’s instructions (Invitrogen). The band of interest was excised and submitted for peptide analysis at the Taplin Mass Spectrometry Facility (Harvard Medical School) as described above.

Ssp120-HA secretion assay

The protocol to analyze extracellular Ssp120-HA secretion into the culture media was adapted from a Kar2 secretion assay (29). Saturated cultures were back-diluted into YPD and grown to mid-logarithmic phase. Cells were then harvested, washed twice in sterile water, and resuspended in fresh YPD at equivalent cell densities (OD₆₀₀=0.1). After 6 h growth, 1.5 mL of the cultures was centrifuged for 5 min at 14,000 × g and 1.35 mL of the supernatant was transferred to a new tube. Extracellular proteins were precipitated from the supernatant by adding cold 100% trichloroacetic acid to a final concentration of 10% and incubating on ice for 20 min. The precipitate was collected by centrifugation at 4°C for 15 min at 14,000 × g, washed with 1 mL of cold 100% acetone, and air-dried at room temperature. Precipitated proteins were resuspended in 40 µL of 5× sample buffer supplemented with 50 mM Tris, pH 9.4, heated for 5 min at 75°C, and 10 µL resolved by SDS-PAGE and visualized by immunoblot. The cell pellets from the 1.5 mL culture were resuspended in 200 µL of lysis buffer (0.1 M sorbitol, 20 mM HEPES, pH 7.4, 50 mM KOAc, 2 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) and were lysed with glass beads by vortexing at 4°C for 10 min. Lysates were cleared by centrifugation at 4°C for 3 min at 3,500 × g and 200 µL of the supernatant transferred to a new tube. This supernatant was mixed with 200 µL of 5× sample buffer, heated for 5 min at 75°C, and 10 µL resolved by SDS-PAGE and visualized by immunoblot.

Live-cell microscopy

Strains expressing recombinant GFP-fusion proteins were grown at 30°C in SC-Ura to mid-log. Cultures were then back-diluted into fresh, pre-warmed SC-Ura in duplicate and incubated for 6 h at 30°C (permissive) or 37°C (restrictive). Cells were imaged at room temperature in liquid medium under a coverslip using a Deltavision Imaging System (Applied Precision, Issaquah, WA) composed of a customized Olympus (Center Valley, PA) IX-71 inverted wide-field microscope, a UPlanS Apochromatic 100×/1.40 NA lens, a Photometrics (Tucson, AZ) CoolSNAP HQ2 camera, and an InsightSSI (Applied Precision) solid-state illumination unit. Images were captured as Z-series with 0.2 µm step-size, and processed by iterative deconvolution in SoftWoRx (Applied Precision) and then analyzed and maximum-intensity projections of the z-stacks generated in ImageJ (National Institutes of Health, Bethesda, MD). For the differential interference contrast (DIC) images, a single focal plane in the center of the cells was used.

Cycloheximide decay assays

Cycloheximide decay assays were performed according to (107) with modifications. Strains expressing CPY*HA from pDN436 were grown to saturation in SC-Leu. Cultures were back-diluted into YPD and grown to OD₆₀₀ = 0.8–1.2, then harvested and concentrated to 2 OD₆₀₀ per mL in fresh, pre-warmed YPD. Protein translation was stopped by addition of CHX to a final concentration of 1 mg/mL. At the indicated time points, 1 mL culture was aliquoted into 300 µL of 50 mM NaN₃ on ice, centrifuged at 4°C for 1 min at 14,000 × g, and the pellets flash frozen in liquid N₂. Pellets were processed as described above for whole-cell lysates.

Synthetic genetic array (SGA) analysis

SGA screening was conducted as described (72) with some modifications (73) using a Singer ROTOT HDA (Singer Instruments). The emp47Δ query strain CBY3663 (emp47Δ::NatMX4, constructed in Y7092) was mated to the yeast haploid deletion collection (each gene deleted with KanMX cassette confers resistance to G418), diploids selected and sporulated, followed by selection of double mutant haploids. The automated software HT Colony Grid Analyzer (108) was used to measure colony sizes after growth at 30°C, and raw values were calculated relative to the median colony size of the plate. Three independent rounds of screening were conducted and for synthetic negative interactions that reproduced in all three rounds, the data were averaged and ranked by strength of the effect (the larger the negative number, the stronger the interaction). We used GO Term Finder to identify significant gene ontology (GO) terms enriched as hits in the screen (109).