Role of Erv29p in Collecting Soluble Secretory Proteins into ER-Derived Transport Vesicles

See allHide authors and affiliations

Science  16 Nov 2001:
Vol. 294, Issue 5546, pp. 1528-1531
DOI: 10.1126/science.1065224


Proteins are transported from the endoplasmic reticulum (ER) in vesicles formed by coat protein complex II (COPII). Soluble secretory proteins are thought to leave the ER in these vesicles by “bulk flow” or through recognition by hypothetical shuttling receptors. We found that Erv29p, a conserved transmembrane protein, was directly required for packaging glycosylated pro-α-factor (gpαf) into COPII vesicles in Saccharomyces cerevisiae. Further, an Erv29p-gpαf complex was isolated from ER-derived transport vesicles. In vivo, export of gpαf from the ER was saturable and depended on the expression level of Erv29p. These results indicate that membrane receptors can link soluble cargo proteins to the COPII coat.

In eukaryotic cells, secretory proteins are packaged into COPII-coated vesicles at the ER for transport through the early secretory pathway. The mechanisms by which secretory proteins are segregated away from ER resident proteins during vesicle formation are still a matter of debate (1). Certain portions of integral membrane cargo appear to bind directly to subunits of the COPII coat (2,3), allowing for their concentration into ER-derived vesicles (4, 5). It is less clear how soluble secretory cargos are exported from the ER, and evidence supporting bulk flow (5) and receptor-mediated export mechanisms (2,6) exists. One difficulty with the receptor model has been the failure to identify integral membrane proteins that could fulfill this function.

In S. cerevisiae, COPII-coated vesicle formation has been reconstituted in cell-free reactions using ER membranes and purified COPII components: Sar1p, Sec23p complex, and Sec13p complex (7). COPII vesicles have been isolated and several of the abundant integral membrane constituents have been characterized in an effort to identify proteins involved in sorting during vesicle formation (8). One such ER-vesicle protein of 29 kD (hence Erv29p) is conserved across species (9), is selectively packaged into COPII vesicles (8), and contains multiple membrane-spanning domains with a terminal dilysine sorting signal (10).

Haploid erv29Δ strains are viable and display no observable growth defects (8, 10). To test whether ERV29 deletion influenced protein transport between the ER and Golgi, we performed a reconstituted cell-free assay that measures transport of [35S]gpαf to the Golgi complex (11). Surprisingly, no transport of gpαf was detected in membranes lacking Erv29p, although translocation of [35S]pre-pαf into ER membranes was unaffected. Specifically, the defect in gpαf transport occurred at the COPII-dependent budding step. Budding of gpαf in wild-type membranes was efficient (33% of total), whereas only minor amounts (4% of total) were budded from erv29Δ membranes (Fig. 1A). To distinguish whether this result was due to a general decrease in COPII budding or a failure to package gpαf into COPII vesicles, we monitored budding of other integral membrane proteins contained on COPII vesicles. Bos1p, an ER/Golgi SNARE (soluble N-ethylmaleimide–sensitive factor attachment protein receptor), and Erv25p, a member of the p24 family, were budded with equal efficiency from both wild-type anderv29Δ membranes in the presence of COPII proteins (Fig. 1B). Sec61p, an ER resident protein, served as a negative control and indicated selectivity in the budding reaction. Thus, in the absence of Erv29p, COPII vesicles formed normally, but gpαf was not packaged into these vesicles.

Figure 1

Erv29p is required for gpαf packaging into COPII vesicles. Reconstituted COPII budding reactions from ER membranes isolated from FY834 (wild type, WT) and CBY966 (erv29Δ) are shown (21). (A) [35S]gpαf budding reactions contained membranes incubated with or without purified COPII proteins (7). The percent budding represents the amount of [35S]gpαf released in freely diffusible vesicles divided by the total amount of [35S]gpαf contained in reactions. (B) Total membranes and budded vesicles, generated as in (A), were collected by centrifugation, resolved on polyacrylamide gels, and immunoblotted for the indicated proteins (22). Lanes labeled T represent one-tenth of the total; minus lanes show vesicles formed in the absence of COPII components; plus lanes show vesicles produced in the presence of COPII proteins. (C and D) As in (A) and (B), respectively, except that membranes from FY834 (WT) and CBY1160 (ERV29-HA) were used. HA mAb (0.07 mg/ml) was added to COPII budding reactions and is indicated by a bracketed plus sign (+) in (D).

To determine whether Erv29p was directly involved in gpαf export, we analyzed COPII vesicle formation and gpαf packaging in wild-type membranes after inhibition of Erv29p function in vitro. We used a hemagglutinin (HA)–tagged version of Erv29p that places three sequential HA epitopes into the NH2-terminal domain of Erv29p (10). Membranes with Erv29-HA as the sole source of Erv29p displayed sensitivity to monoclonal antibody to HA (HA mAb) when gpαf budding was measured (Fig. 1C), whereas strains expressing endogenous Erv29p were insensitive. Furthermore, the packaging of Erv29-HA into COPII vesicles was inhibited when HA mAb was added (Fig. 1D). Presumably, antibody binding to this HA epitope sterically hinders access of coat subunits to Erv29p. The budding efficiencies of other vesicle proteins were not decreased when Erv29-HA function was neutralized; this result indicates that Erv29p function is specifically and directly required for packaging of gpαf into COPII vesicles.

If Erv29p acts in packaging of gpαf into COPII-coated vesicles,erv29Δ cells should exhibit a reduced rate of gpαf transport in pulse-chase analyses. In wild-type strains, pαf acquires N-linked core oligosaccharide in the ER, generating the 29-kD gpαf form. Transport through the Golgi complex produces a heterogeneously glycosylated species that is cleaved to generate secreted pheromone (12). The ER form of gpαf was exported rapidly in the wild type; however, gpαf accumulated in erv29Δ cells, and after a 20-min chase only 45% had been transported from the ER (Fig. 2A). The sec12-4temperature-sensitive mutant blocked all export from the ER and demonstrated the characteristic accumulation of secretory proteins in the ER.

Figure 2

Cells lacking Erv29p accumulate ER forms of gpαf and CPY, but not Gas1p or invertase. Cell extracts were prepared from FY834 (WT), CBY966 (erv29Δ), and RSY263 (sec12-4) after a 10-min pulse with35S-labeled amino acids and chased for the indicated times (8). (A) gpαf, (B) CPY, and (C) Gas1p were immunoprecipitated from a common extract and resolved on 10% polyacrylamide gels. (D) Invertase pulse-chase analyses were performed with derepressed cells. Proteins were visualized and transport rates quantified using a PhosphorImager (20).

Additional secretory proteins were monitored in this pulse-chase experiment. Transport of carboxypeptidase Y (CPY) from the ER was also delayed in erv29Δ cells (Fig. 2B). The ER form of CPY (p1) migrates as a 67-kD species that is modified in the Golgi complex to generate the 69-kD p2 form and is then cleaved to the 61-kD mature form upon delivery to the vacuole (13). Inerv29Δ cells, 40% of the CPY remained as the p1 form after a 20-min chase, whereas most of the CPY was fully matured after a 10-min chase in wild-type cells. This is consistent with a reported delay in transport of CPY and proteinase A in erv29Δ strains (10). This transport defect was specific for a subset of secretory proteins because the GPI-anchored protein Gas1p (Fig. 2C), the soluble secreted protein invertase (Fig. 2D), and vacuolar targeted alkaline phosphatase (10) appeared to mature at wild-type rates in erv29Δ strains.

In subcellular fractionation experiments, we observed that Erv29p was equally distributed between ER and Golgi membranes (14). If Erv29p cycles as a cargo receptor for gpαf, ER export should be saturable and the expression level of Erv29p should influence the export capacity for these secretory proteins. To test this idea, we overexpressed gpαF (encoded by the MFα1 gene) and monitored the kinetics of ER export in pulse-chase experiments (Fig. 3A). Overexpression ofMFα1 resulted in a delay by a factor of 4 or more in the transport kinetics of gpαf from the ER. We also detected a modest decrease (factor of 1.3) in the transport rate of p1 CPY under this condition. In contrast, the transport rates of Gas1p and alkaline phosphatase were unaffected by MFα1overexpression. Erv29p was limiting under these conditions, because cells with an additional copy of ERV29 expressed twice the level of Erv29p (14) and accelerated the transport kinetics of overproduced gpαf by a factor of 2 (Fig. 3B). These results indicate that transport of gpαf from the ER was saturable, and that increased expression of Erv29p partially alleviated this accumulation. A restoration of CPY transport rate was also observed in this strain (Fig. 3B).

Figure 3

Expression level of Erv29p influences the transport kinetics of gpαf. (A) Cells were pulsed for 5 min with 35S-labeled amino acids, then chased for the indicated times. Proteins were immuno- precipitated from ex- tracts prepared from FY834 (WT), CBY966 (erv29Δ), and CBY1161 (WT,MFα1-2μ), a strain that overproduces gpαf. (B) As in (A), with CBY1162 (MFα1-2μ, ERV29 × 2), a strain that overproduces gpαf and contains two chromosomal copies ofERV29, included. Proteins were visualized and transport rates quantified using a PhosphorImager (20).

The above results suggested that Erv29p binds gpαf in the ER and forms a receptor-cargo complex for capture into COPII-coated vesicles. To look for such a complex, we immunoprecipitated Erv29p after treating ER membranes containing [35S]gpαf with the thiol-cleavable crosslinker dithiobis(succinimidyl propionate) (DSP). Using antiserum to Erv29p, we coimmunoprecipitated [35S]gpαf with Erv29p in a crosslinker- and Erv29p-dependent manner (Fig. 4A). As an independent method, we also immunoprecipitated [35S]gpαf in complex with Erv29-HA using HA mAb in a reaction that depended on crosslinker (Fig. 4A) and on the presence of HA-tagged Erv29p (14). In wild-type microsomes, the recovery of [35S]gpαf in complex with Erv29p was ∼0.5% of total; however, only 2% of the total Erv29p could be immunoprecipitated under these conditions. Recovery of [35S]gpαf linked to Erv29-HA was improved (∼1% of total) using HA mAb compared with the Erv29p polyclonal antibody, although addition of crosslinker reduced the efficiency of Erv29-HA immunoprecipitation (Fig. 4A). Isolation of the cross-linked Erv29p-gpαf complex appeared specific because no Kar2p (mammalian BiP homolog) was detected in these immunoprecipitations.

Figure 4

Isolation and characterization of Erv29p-gpαf complexes. (A) Microsomes (0.2 mg) isolated from FY834 (WT), CBY966 (erv29Δ), and CBY1160 (ERV29-HA) that contained translocated [35S]gpαf were treated with (+) and without (–) DSP (0.5 mM), solubilized in 0.5% SDS, and immunoprecipitated with antiserum to Erv29p (lanes 3, 4, 7, and 8) or HA mAb (lanes 11 and 12) as described (23). Immunoprecipitates were reduced and resolved on polyacrylamide gels. Total lanes represent 1% of initial extracts. [35S]gpαf was detected by PhosphorImager analysis and other proteins detected by immunoblot. (B) Microsomes (0.35 mg) from FY834 (WT) or CBY1160 (Erv29-HA) that contained translocated [35S]gpαf were used to generate vesicles by the addition of purified COPII proteins. Budded vesicles were collected by centrifugation (lanes 3 and 9) or treated with DSG (0.05 mM) and incubated with 0.1 M Na2CO3, pH 11.0 (lanes 4 to 6 and 10 to 12) to extract soluble proteins before collecting membranes. For immunoprecipitations, extracted vesicles were solubilized in 0.5% SDS, diluted, and incubated with protein A–Sepharose beads and preimmune serum (lane 5), antiserum to Erv29p (lane 6), beads alone (lane 11), or HA mAb (lane 12). Total lanes (T) represent 5% of starting microsomal extracts. Arrowheads indicate the primary cross-linked products in immunoprecipitates.

To further characterize the composition of the Erv29-gpαf complex, we performed cross-linking experiments with a noncleavable agent, disuccinimidyl glutarate (DSG). COPII vesicles containing [35S]gpαf were generated from wild-type and Erv29-HA microsomes and treated with DSG, and Erv29-[35S]gpαf complexes were isolated by immunoprecipitation with antiserum to Erv29p or with HA mAb (Fig. 4B). Immunoprecipitated proteins were resolved on polyacrylamide gels and distinct Erv29-[35S]gpαf complexes were observed by autoradiography. In wild-type vesicles, primary cross-linked products of about 56 and 85 kD were detected. In Erv29-HA vesicles, analogous products of 61 and 90 kD were observed, with a ∼5-kD increase in mass due to the presence of the 3×HA tag. The appearance of a 56-kD radiolabeled species in wild-type vesicles (which is the approximate mass of an Erv29p-[35S]gpαf complex) and a corresponding ∼5-kD shift observed in the Erv29-HA vesicles indicated that Erv29p and [35S]gpαf were in direct contact. This finding is in accord with reports proposing that a 27-kD integral membrane protein binds [35S]gpαf specifically in COPII vesicles (2, 15). The molecular identity of this vesicle protein appears to be Erv29p. The additional higher molecular weight species in these immunoprecipitates (85 kD in the wild type and 90 kD in Erv29-HA) suggests that a gpαf dimer bound to Erv29p or that Erv29p functioned as a dimer. The major cross-linked product of ∼58 kD detected in samples before immunoprecipitation probably represents a gpαf dimer, as previously observed (2).

We propose that Erv29p binds to fully folded gpαf and probably to other soluble secretory cargo in the ER, after which Erv29p-cargo complexes are packaged into COPII vesicles for transport to the Golgi complex. Upon delivery to the Golgi, Erv29p-cargo complexes dissociate and empty receptors are recruited into ER-directed COPI vesicles through a dilysine motif present on Erv29p. Similar mechanisms have been proposed for the Emp24 complex (6) and ERGIC53 (16), integral membrane proteins that also cycle between ER and Golgi compartments. As in these examples, Erv29p has homologs in all higher eukaryotes with sequenced genomes, suggesting a conserved function. In light of recent findings linking ERGIC53 to secretion of blood-clotting factors and to some forms of hemophilia (17), it may be informative to investigate mammalian Erv29p. Furthermore, ERV29 and other conserved genes involved in transport between the ER and Golgi are up-regulated by the unfolded protein response (18) and are required for efficient degradation of soluble misfolded substrates (10). This induction may serve to clear the ER of soluble misfolded proteins when ER-associated degradation becomes saturated (10, 18), although it remains to be determined whether Erv29p acts directly in ER quality control.

Forward transport from the ER is coordinated with a retrograde pathway that recycles vesicle proteins and retrieves escaped ER residents (19). Indeed, selective exclusion of secretory cargo from retrograde vesicles may explain the concentration of certain soluble cargo after export from the ER (5). Our results indicating Erv29p action as a cargo receptor do not exclude other sorting mechanisms; rather, it now seems probable that multiple mechanisms of retention, retrieval, and selective export operate in concert to achieve organization of the early secretory pathway.

  • * To whom correspondence should be addressed. E-mail: barlowe{at}


View Abstract

Navigate This Article