Golgi-mediated Glycosylation Determines Residency of the T2 RNase Rny1p in Saccharomyces cerevisiae

Authors


Abstract

The role of glycosylation in the function of the T2 family of RNases is not well understood. In this work, we examined how glycosylation affects the progression of the T2 RNase Rny1p through the secretory pathway in Saccharomyces cerevisiae. We found that Rny1p requires entering into the ER first to become active and uses the adaptor protein Erv29p for packaging into COPII vesicles and transport to the Golgi apparatus. While inside the ER, Rny1p undergoes initial N-linked core glycosylation at four sites, N37, N70, N103 and N123. Rny1p transport to the Golgi results in the further attachment of high-glycans. Whereas modifications with glycans are dispensable for the nucleolytic activity of Rny1p, Golgi-mediated modifications are critical for its extracellular secretion. Failure of Golgi-specific glycosylation appears to direct Rny1p to the vacuole as an alternative destination and/or site of terminal degradation. These data reveal a previously unknown function of Golgi glycosylation in a T2 RNase as a sorting and secretion signal.

image

The T2 family of RNases represents an ancient group of enzymes identified in every domain of life; T2 RNases exist in bacteria, viruses, fungi, plants and metazoans, and are found in the genomes of almost all eukaryotic organisms [1, 2]. T2 RNases belong to a class of secretory nucleases that also includes the RNase A and RNase T1 families. T2 RNases are united into one family based on their structural and biochemical characteristics (reviewed in [3]). However, these enzymes exhibit various intra- or extracellular localization patterns in different species and perform different functions. For example, RNase LE from Solanum lycopersicum and RNases EhNucI and II from the nitrogenous base auxotroph Entamoeba histolytica are secreted into the extracellular environment acting as scavengers that degrade RNA to help these cells procure limited essential nutrients, such as phosphate or nitrogenous bases [4-6]. Other T2 nucleases require secretion in order to enter into other cells where they engage in cytotoxic activities. For instance, plant S-RNases, which are expressed by cells of the pistil, destroy cellular RNAs of incompatible pollen cells to prevent self-fertilization (reviewed in [7]). Similarly, the envelope protein Erns of the CSFV and BVDV viruses targets lymphocytes, thereby, dampening the host immune response [8, 9].

Localization and intracellular functions have been shown for the T2 RNase Rny1p (RNase in Yeast 1) in budding yeast and its orthologs in plants and mammals. Rny1pGFP was detected predominantly in the vacuole of unstressed cells [10]. It was shown that during oxidative stress or stationary phase, Rny1p translocates to the cytoplasm where it can cleave cytoplasmic tRNA and rRNA [10]. These data are in agreement with the intracellular localization and rRNA-degrading activity observed for the Rny1p orthologs RNS2 in Arabidopsis [11] and RNASET2 in zebrafish [12]. Interestingly, initial work on Rny1p showed an increased Rny1p-dependent nucleolytic activity in growth media, suggesting that this nuclease can be secreted outside yeast cells [13]. Furthermore, co-culturing experiments performed using the Rny1pGFP fusion protein did not support the secretion/uptake mechanism of Rny1p delivery into the cytoplasm [10]. Collectively, these studies suggest that (i) Rny1p may be supplied to the cytoplasm from the intracellular secretory compartment; and (ii) secretion of this protein to the extracellular environment represents an alternative destination that might result in a different biological function. Currently, it is unknown what molecular mechanisms may be responsible for determining whether Rny1p is secreted outside or retained inside the cell. An understanding of Rny1p targeting may provide insight into the regulation and function of other T2 RNases, some of which are implicated in human diseases, including neurodegeneration and cancer [12, 14]. Specifically, recently identified mutations in the human RNASET2 gene have been associated with white matter disorders in the human brain [15].

Yeast Rny1p is a 50 kDa protein that contains two catalytically active histidine residues (His87 and His160) [3] and a 140 amino acid long C-terminal extension, which is a specific feature of fungal T2 RNases. A recent study demonstrated that the C-terminal region is dispensable for the catalytic activity of Rny1p [16] but to date its function remains unclear. As a typical secretory protein, Rny1p contains an amino-terminal signal peptide responsible for the protein translocation into the ER lumen [16]. Rny1p is modified with glycans [13, 16], however, neither the glycosylation sites nor the functional role of these modifications have been determined.

In this study, we explored the molecular mechanisms controlling intra- and extracellular localization of Rny1p. We created several mutants of Rny1p that are impaired in certain functional properties and assessed their nucleolytic activity, glycosylation state, intracellular localization and the ability to be secreted outside of cells. Using biochemical and genetic analyses we determined that Rny1p is transported out of the ER in COPII vesicles using Erv29p as an adaptor. We identified four N-linked glycosylation sites and found that glycosylation in the Golgi is necessary for the extracellular secretion of this nuclease. Unglycosylated and underglycosylated forms, and those modified only with the ER core glycans retained their catalytic activity. Our results demonstrate for the first time the role of Golgi-mediated glycosylation in T2 RNase secretion.

Results

Rny1p is modified by glycosylation in the ER and Golgi

To examine Rny1p progression through the secretory pathway, we created several mutants of the nuclease; the mutants generated include a catalytically inactive form (H87R) with mutation of one of the two crucial histidine residues in the active center [3], an N-terminal truncation lacking the ER signal peptide (ΔN), and a mutant that was serendipitously generated during polymerase chain reaction (PCR, W399R) and found to be defective in Golgi-mediated glycosylation as shown below (Figure 1A). All constructs were expressed in BY4741 rny1Δ cells from a 2μ plasmid using the constitutive ADH promoter and carried a C-terminal FLAG tag as a means of detection.

Figure 1.

Rny1p is glycosylated in the ER and Golgi compartments. A) Schematic representation of Rny1p and mutants used in this study. ‘x’ indicates asparagine residues at the position 37, 70, 103 and 123. B) BY4741 rny1Δ cells were transformed with an empty vector (V) or plasmids expressing WT or the indicated Rny1pFLAG mutants. Cells were grown in synthetic medium; total protein lysates were prepared and analyzed by western blotting with anti-FLAG antibodies. Ponceau S staining of blots after transfer revealed equivalent loading of the total protein. Asterisk indicates the position of a background band cross-reacting with the antibody. C) Proteins from cells expressing Rny1pFLAG WT and W399R were immobilized on anti-FLAG M2-agarose beads and treated with endoglycosylase H (+) or remained untreated (−). Proteins were then eluted and analyzed by western blotting. D) WT and och1Δ cells, expressing either WT or W399R mutant of Rny1pFLAG, were grown in synthetic medium, diluted with YPDA and treated with Tm for 2 h at 30°C (+) or left untreated (−). Total protein extracts were prepared and analyzed by western blotting. Two exposures are shown for samples derived from Rny1pFLAG WT-expressing cells. E) BY4741 rny1Δ cells expressing the indicated mutants of Rny1pFLAG were analyzed by western blotting. Tm and EndoH treatments of Rny1pFLAG WT were performed as described in (C) and (D). Two exposures are shown for samples derived from Rny1pFLAG-N37Q, -N70Q, -N103Q and -N123Q expressing mutants.

First, we analyzed expression of the generated mutants by western blotting. Figure 1B shows that all proteins, except ΔN, migrated slower than the expected molecular size (50 kDa). The WT protein and H87R, but not the W399R mutant, also exhibited heterogenous high molecular weight (HMW) forms, which is typical of glycosylated proteins. We used several approaches to confirm that the slower migrating bands correspond to different glycosylated forms of Rny1p. First, we treated WT and W399R proteins immobilized on beads with endoglycosidase H (EndoH) and found that this shifted the HMW products down to a single band of the expected 50 kDa molecular weight size (Figure 1C). Second, we inhibited core N-glycosylation in the ER with tunicamycin (Tm), which also resulted in unmodified WT and W399R proteins migrating at 50 kDa (Figure 1D). Third, we examined expression of Rny1pFLAG in a strain lacking Och1p (och1Δ), an α-1,6 mannosyl-transferase which adds a single mannose residue to core N-linked oligosaccharides upon their arrival to the Golgi, thereby initiating formation of polymannan chains (reviewed in [17]). Consistent with impaired Golgi glycosylation, Rny1p lost its HMW modification in och1Δ cells but retained the ER-mediated glycan modification that was sensitive to Tm (Figure 1D). Collectively, these data indicate that modification of Rny1p with N-linked glycans starts in the ER resulting in a ˜60 kDa protein and continues in the Golgi generating heterogeneous HMW forms. The W399R mutant is defective in the Golgi-, but not ER-specific glycosylation.

Rny1p contains four glycosylation sites

Analysis of the Rny1p amino acid sequence revealed four potential N-linked glycosylation sites with the NxS/T consensus sequence in the amino terminal portion of Rny1p: N37KT, N70ET, N103RS and N123DT. Using site-directed mutagenesis, we changed these asparagines to glutamine residues in Rny1pFLAG both individually and in various combinations and analyzed expression of the mutants by western blotting. As shown in Figure 1E, all four sites contributed to ER core glycosylation of Rny1p, since elimination of each individual site caused a visible increase in the protein mobility in the SDS gel (Figure 1E). When all four glycosylation-competent asparagine residues were mutated (N37/70/103/123Q), Rny1p migrated at 50 kDa, similar to the Tm- and EndoH-treated WT protein (Figure 1E), verifying the unglycosylated status of this mutant. Furthermore, heterogeneous HMW forms of Rny1p were evident for all single-asparagine mutants (for N123Q mutant see long exposures in Figure 1E). These data indicate that loss of any single glycosylation site identified on Rny1p does not cause a complete block in Golgi processing.

Rny1pGFP WT and its mutants localize to the vacuole

Interestingly, elimination of only two out of four glycosylation sites completely abolished formation of HMW forms of the protein, as was observed for various double and triple mutants of Rny1p (Figure 1E). One explanation for this result is an alteration of protein conformation caused by two amino acids substitutions. Altered protein conformation might result in Golgi-delivery failure accompanied by protein retention inside the ER where it could be degraded by ER quality control mechanisms. Therefore, we tested whether glycosylation-defective mutants of Rny1p are substrates for ER-associated protein degradation (ERAD). We studied the turnover kinetics of Rny1p WT, W399 and N37/70/103/123Q in WT and an ERAD-deficient strain (npl4-1). Npl4p is a member of ubiquitin-selective chaperone complex Cdc48p-Npl4p-Ufd1p that is essential for the presentation of retrotranslocated ubiquitinated proteins to the proteasome for degradation during ERAD [18]. The well-characterized ERAD substrate CPY*, whose degradation is dependent on Hrd1p ubiquitin-ligase [19], was used as a positive control. Degradation CPY*FLAG was significantly stabilized in npl4-1 cells over time after cyclohexamide (CHX) treatment, similar to the stabilization observed in hrd1Δ cells [20] (Figure S1A,B). These data confirm that ERAD is suppressed in the npl4-1 strain. Unlike CPY*, the degradation kinetics of ER forms of Rny1p WT (migrate on the gel as 60 kDa protein) and all the tested mutants was unaffected by suppression of NPL4 function (Figure S1C–E). Moreover, similar to npl4-1, disruption of other components of the ERAD machinery [18], HRD1, DOA10, and UBC7, also showed lack of stabilization of W399R (Figure S1F). These data further support the idea that glycosylation-defective Rny1p mutants are not destroyed via the ERAD mechanism.

To examine whether conformational alterations of various Rny1p mutants affect protein exit from the ER and the progression along the secretory pathway, we first assessed the localization of Rny1p WT and mutants in live cells by fluorescence microscopy using a C-terminally GFP-tagged form of these proteins. The fluorescent signal was detected largely in the vacuole in WT cells expressing Rny1pGFP WT (Figures 2A and S2A), consistent with previously published results [10].

Figure 2.

Localization and nucleolytic activity of Rny1p mutants. A) Localization of Rny1pGFP mutants. WT cells were transformed with plasmids expressing the indicated Rny1pGFP constructs. alg5Δ and och1Δ mutant strains were transformed with Rny1pGFP WT. Exponentially growing cultures were analyzed by fluorescence microscopy. Prior to analysis, cells were treated with the vacuolar marker dye FM4-64. BF, bright field; scale bar is 5 µm. B) Glycosylation of Rny1p is dispensable for the nucleolytic activity of Rny1p. BY4741 rny1Δ cells expressing the indicated Rny1pFLAG mutants were lysed and analyzed on SDS-free 8% polyacrylamide gel supplemented with RNA from Torula yeast. Following electrophoresis, the gel was stained with the RNA dye toluidine blue O. The ‘WT+Tm’ sample was derived from Rny1pFLAG WT-expressing cells that were treated with Tm for 2 h. C) BY4741 rny1Δ or alg5Δ cells expressing the indicated Rny1pFLAG mutants were cultured in synthetic medium, diluted with YPDA to OD600 ˜ 0.3, and grown at 30°C for 4 h. RNA was extracted and analyzed by Northern hybridizations with probes against 25S and 18S rRNA. WT+Tm lane demonstrates rRNA from Rny1pFLAG WT-expressing cells that were treated with Tm for 2 h prior to RNA extraction. Asterisk indicates unspecific cross-hybridization of y503 probe with 18S rRNA. Methylene blue staining of the membrane prior to hybridization demonstrates equal RNA loading.

Interestingly, most mutants of Rny1pGFP tested (i.e. the catalytically inactive mutant H87R; the Golgi-glycosylation deficient mutant W399R; the glycosylation site deficient mutants N37Q, N70Q, N103Q and N123Q; and the unglycosylated mutant N37/70/103/123Q) demonstrated localization similar to that of the WT protein inside the cell; fluorescent signal was derived strictly from the vacuolar compartment (Figures 2A and S2C–I). These data indicate that all the Rny1p mutants described are able to exit the ER and traverse through the Golgi to reach the vacuole.

Similarly to unglycosylated Rny1pGFP N37/70/103/123Q mutant, Rny1pGFP WT expressed in the dolichol pathway-defective alg5Δ [21] and Golgi-glycosylation deficient och1Δ [17] cells was also detected in the vacuole (Figures 2A and S2J,L). These data confirm that modification of Rny1p with glycans is not required for trafficking of this protein through the ER-Golgi-vacuole route.

Finally, deletion of the ER signal peptide prevented Rny1p entry into the ER, making the protein localize in the cytoplasm in the form of large bright aggregates (Figures 2A and S2B). Of note, cells expressing this mutant did not show any obvious growth abnormalities, suggesting that ΔN-Rny1pGFP-containing foci are not harmful to the cells.

Taking together, these data demonstrate that Golgi-glycosylation deficient mutants of Rny1p are able to exit the ER and traffic through the Golgi to the vacuole, however they lack the ability to be modified in Golgi. This could be explained by the mutants failure to be recognized and/or modified by the Golgi-glycosylation apparatus.

N-Glycosylation does not affect the catalytic activity of Rny1p

We started addressing the functional role of Rny1p glycosylation by testing whether this modification affected the catalytic activity of this RNase. First, we tested the ability of the generated glycosylation mutants to digest RNA by an in-gel-degradation assay. Protein extracts prepared from cells expressing different Rny1p mutants were resolved by SDS-free polyacrylamide gels supplemented with RNA from the Torula yeast. In this assay, an active RNase that retains its ability to digest RNA inside the gel will produce clear traces visible after Toluidine blue O staining of the gel (Figure 2B). To ensure that the mutants of Rny1p analyzed are expressed at comparable levels, protein extracts were also resolved by SDS–PAGE followed by western blotting (Figure S3A).

As expected, no significant RNA degradation was observed in samples derived from rny1Δ cells that carry an empty vector or express a catalytically dead mutant of Rny1p (H87R). Interestingly, Tm treatment of Rny1pFLAG WT-expressing cells, which results in production of an unmodified enzyme (Figure 1D) did not reduce Rny1p activity, indicating that glycosylation is not required for the catalytic function of this nuclease (Figure 2B).

Elimination of a single glycosylation site did not abolish the nucleolytic activity of the enzyme, although it caused a visible activity decrease (Figure 2B). At the same time, substitution of two or more glycosylation-competent asparagines with glutamines resulted either in further activity reduction (N37/123Q, N70/123Q, N103/123Q) or in complete catalytic incompetence (Figure 2B). Impaired nucleolytic activity observed with the glycosylation site mutants is likely due to altered protein conformation caused by amino acid mutation(s).

Several reports in the literature demonstrate that Rny1p cleaves cellular tRNAs and rRNAs when overexpressed [10, 16]. Therefore, we next tested whether our glycosylation mutants would be active in degrading cellular substrates of Rny1p, such as 18S and 25S rRNAs. Indeed, exogenous expression of Rny1pFLAG WT and all tested glycosylation-defective mutants in rny1Δ cells caused increased degradation of cellular rRNA as revealed by Northern hybridization (Figure 2C). Of note, expression of the catalytically inactive N37/70/103/123Q, H87R and ΔN mutants did not results in increased degradation of cellular rRNA. Consistent with results from the in-gel RNA degradation assay, we found that unglycosylated and underglycosylated Rny1p was capable of promoting rRNA fragmentation in Tm-treated cells and in alg5Δ cells, respectively (Figure 2C). Alg5p is a dolichyl-phosphate glycosyltransferase that adds glucose to dolichyl-phosphate [22]. Protein glycosylation is defective in alg5Δ, since slightly fewer oligosaccharides are transferred to the substrate [22]. Consistently, we found that Rny1pFLAG is expressed in alg5Δ cells in underglycosylated forms, resembling single, double and triple glycosylation site mutants (asterisks in lane 3 in Figure S3B).

To ensure that detected rRNA degradation does not occur in the lysate during the RNA extraction procedure, we analyzed total cellular RNA from two combined yeast cultures: rny1Δ cells expressing either Rny1pFLAG WT or empty vector-control culture and a tester culture of rny1Δ cells transformed with a plasmid driving expression of functional 25S rRNA tagged with a sequence that allowed it to be distinguished from genome-encoded rRNA during northern blot analysis (scheme in Figure S4A). No effect of Rny1p overexpression was observed on the decay of tagged rRNA in tester cells, whereas endogenous 25S rRNA showed signs of degradation by overexpressed Rny1pFLAGWT (Figure S4B).

Previous work demonstrated that during starvation, autophagosomal machinery imbeds parts of the ribosome-containing ER (rough ER) [23] or specifically captures ribosomes to the vacuole for degradation [24]. Therefore, autophagy might represent an alternative/additional mechanism to deliver ribosomes to the vacuole for degradation and vacuole-resident Rny1p might participate in this process. To address this possibility, we studied Rny1p-mediated degradation of rRNAs in the strain deficient for major autophagic gene ATG1. We observe no differences in rRNA degradation between WT and atg1Δ strains expressing Rny1pFLAG WT or W399R (Figure S4C). These data are in agreement with previously published work [10].

Finally, the lack of detectable RNA-degrading activity of the ΔN mutant (Figure 2B,C) supports the role of these 18 N-terminal amino acids as a signal peptide, responsible for Rny1p entry into the ER lumen. In fact, the inability to be presented to the ER lumen environment results in aggregation of the ΔN protein in the cytoplasm (Figures 2A and S2B), likely due to its misfolding.

Collectively, data presented in this paragraph show that (i) Rny1p-induced RNA degradation is likely to be direct, and (ii) glycosylation of Rny1p is not required for its RNase activity.

Golgi-mediated glycosylation controls Rny1p secretion

As have been discussed in Introduction, some T2 RNases are secreted outside of the cell. Moreover, increased RNase activity was detected in the growth media of RNY1 cells but not in rny1Δ cells [13]. To examine secretion of Rny1pFLAG WT and its mutants we used the colony immunoblotting assay [25, 26]. Briefly, serial dilutions of yeast cultures were spotted on agar plates in duplicate: one plate was covered with a nitrocellulose membrane to capture secreted proteins on the membrane, while the second plate was left intact to monitor culture growth. Secretion of Rny1pFLAG was detected on colony imprints by western hybridization with anti-FLAG antibody. All tested yeast cultures grew to a similar density (Figure 3A, left panel), but secreted Rny1pFLAG was detected only in WT and H87R expressing colonies (Figure 3A, right panel). Secretion failure of W399R suggests that Golgi-mediated glycosylation might be required for the extracellular export of this nuclease. To confirm this, we tested Rny1p secretion under additional settings of suppressed glycosylation. As expected, Rny1p failed to be efficiently secreted when expressed in alg5Δ cells, or when Golgi-mediated glycosylation was inhibited by deletion of OCH1, or when core glycosylation in the ER (an absolute prerequisite for Golgi-mediated glycosylation) was suppressed by Tm treatment (Figure 3B). Next, secretion of our glycosylation site mutants was also tested by the colony immunoblotting assay. As expected, mutation of a single glycosylation-competent asparagine did not abolish protein export outside of the cell (Figure 3C). However, Golgi-glycosylation deficient double and triple mutants, as well as the unglycosylated N37/70/103/123Q mutant, were unable to be secreted (Figure 3D). On the basis of these data, we conclude that in order to be secreted, Rny1p requires glycan modifications that take place inside the Golgi apparatus.

Figure 3.

Golgi-mediated glycosylation of Rny1p is required for extracellular secretion. A) Yeast transformants from Figure 1B were grown in synthetic medium, adjusted to the same OD600, and spotted as a series of six serial dilutions (1:5) on two YPDA plates. One plate was incubated at 30°C for 24 h and demonstrates equal culture growth (left). The second plate was covered with a nitrocellulose membrane and incubated at 30°C for 24–36 h. The membrane was removed, washed and the secreted Rny1pFLAG proteins were visualized by western blotting using anti-FLAG antibody (right). B) WT, alg5Δ and och1Δ cells were transformed with an empty vector (−) or an Rny1pFLAG WT construct (+) and extracellular secretion of Rny1p was examined as in (A). Where indicated, cultures were treated for 2 h with Tm prior to spotting on the plate. C) The indicated yeast transformants were analyzed for secretion as described in (A). D) The indicated yeast transformants were analyzed for secretion as described in (A), however, only three serial culture dilutions were used. E) Exponentially growing BY4741 rny1Δ cells expressing either Rny1pFLAG WT or the W399R mutant were lysed by glass-bead shearing (‘total protein’) or were treated with Zymolyase and both, supernatant (‘external’) and spheroplasts (‘internal’) fractions were analyzed by western blotting using anti-FLAG antibody. Asterisk indicates the position of a background band cross-reacting with the antibody.

In addition to the colony immunoblotting assay, secretion of Rny1pFLAG was analyzed biochemically. Initially, we were unable to detect the presence of secretion-competent Rny1pFLAG in the growth media, indicating that secreted nuclease might remain associated with the cell wall. Therefore, we separated the cell wall-containing fraction (external) from the cellular fraction (internal) by treating Rny1pFLAG-expressing cells with Zymolyase, an enzyme that digests the yeast cell wall [27]. Western blot analysis revealed strong accumulation of HMW Golgi-glycosylated forms of the nuclease exclusively in the external fraction of WT-expressing cells (fraction that contains secreted proteins), while ER-glycosylated Rny1p was present in the internal, intracellular fraction (Figure 3E). This result demonstrates that only the pool of Rny1p that is modified with high-glycans inside the Golgi compartment is capable of exiting the cell. Consistent with this conclusion, the Golgi-glycosylation deficient mutant W399R remained inside the cells, as it was detected in the internal, but not external fraction (Figure 3E).

Secretion of Rny1p allowed us to address whether addition of a FLAG tag on the carboxy terminus of this enzyme influences protein trafficking. We tracked the presence of Rny1pFLAG and Rny1pNO_TAG (untagged version of Rny1p) outside the cells by their ability to digest extracellular RNA in the growth media. Secretion-deficient W399R mutant (FLAG tagged and untagged) was used as negative control. As expected, both FLAG-tagged and untagged Rny1p WT, but not W399R, depicted comparable levels of nucleolytic activity in the growth media, as was revealed by Northern hybridization using a probe against tRNA-Val (Figure S4D). Thus, these data confirm that the FLAG-tagged derivative of Rny1p accurately reflects Rny1p trafficking/secretion.

Rny1p exits the ER via a COPII-mediated Erv29p-dependent mechanism

Most secretory proteins utilize the COPII machinery to travel from the ER to Golgi. COPII is a protein complex that forms ER-derived transport vesicles and selects secretory proteins by direct or indirect interactions [28, 29]. For some soluble substrates, such as the yeast glycopro-α-factor (gpαf) [30] and carboxypeptidase Y (CPY) [31, 32], the COPII vesicle-associated protein Erv29p functions as a transmembrane receptor linking soluble cargo molecules to the COPII coat during vesicle formation [32]. Emp24p and Erv26p are two other cargo receptors that recruit Gas1p and ALP, respectively [31, 33, 34] To determine if Rny1p exit from the ER requires any of these receptors, we studied Rny1pGFP localization in erv29Δ, erv26Δ and emp24Δ cells. As shown in Figure 4B and S2K, deletion of ERV29 was sufficient to prevent transport of Rny1pGFP to the vacuole, resulting in accumulation in the ER, similar to the pattern of the ER marker protein Hmg1GFP [35]. This effect of erv29Δ was specific, as GFP continued to be detected in the vacuole when Rny1pGFP was expressed in erv26Δ and emp24Δ cells (Figure S5).

Figure 4.

Erv29p controls traffic of Rny1p from the ER. A and B) Exponentially growing cultures of WT and erv29Δ cells transformed with an Rny1pGFP WT construct were analyzed by fluorescence microscopy. Prior to analysis, cells were treated with the vacuolar marker dye FM4-64. Hmg1p-GFP was used as a marker protein for the ER [35]. BF, bright field; scale bar is 5 µm. C) WT and erv29Δ cells were transformed with an Rny1pFLAG WT or an empty vector and grown in synthetic medium. Total protein lysates were analyzed by western blotting. D) WT and erv29Δ cells were transformed with an Rny1pFLAG WT (+) or an empty vector (−) and analyzed by the colony hybridization assay as described in Figure 3A. E) Rny1pFLAG WT-expressing erv29Δ cells were incubated in YPDA in the presence (+) or absence (−) of Tm. Cells were harvested, lysed and total protein extracts were analyzed by western blotting.

To confirm the result obtained with the GFP fusion, we expressed Rny1pFLAG WT in WT and erv29Δ cells. Heterogenous HMW forms of Rny1p, indicative of Golgi-mediated glycosylation, were greatly reduced in erv29Δ cells compared to WT cells (Figure 4C), supporting our microscopy data showing the inability of Rny1p to exit the ER in the absence of Erv29p. Moreover, the secretion assay demonstrated a strong inhibition of Rny1p extracellular export in erv29Δ cells (Figure 4D), consistent with the idea that this nuclease is trapped inside the ER in the absence of Erv29p. Finally, lack of Erv29p did not affect ER-mediated core glycosylation of the nuclease, since Tm treatment of erv29Δ cells expressing Rny1pFLAG WT shifted the protein to a 50 kDa form (Figure 4E). Collectively, these data indicate that Rny1p exits the ER largely via a COPII-mediated mechanism in which Erv29p functions as a major and specific receptor for Rny1p recruitment into COPII vesicles.

Golgi glycosylation and extracellular secretion of Rny1p are inhibited by C-terminal GFP fusion

When we analyzed Rny1pGFP by western blotting we noticed that, unlike FLAG-tagged forms (Figure 1B), the GFP fusion protein migrated as a single band during SDS–PAGE (Figure 5A). Both EndoH and Tm treatments caused the downshift equal in size to the removal of ER glycosylation (Figure 5B). Furthermore, the migration size of the Rny1pGFP WT protein expressed in either WT cells or in och1Δ cells was similar (Figure 5B). This data suggests that although Rny1pGFP can undergo core glycosylation in the ER, this fusion protein lacks modifications occurring inside the Golgi, resembling the W399R mutation. Moreover, similar decay kinetics of Rny1pGFP were detected in WT and npl4-1 cells (Figure S1G), indicating that GFP fusion of Rny1p is not an ERAD substrate.

Figure 5.

Rny1p-GFP fusion is defective in glycosylation and extracellular secretion. A) BY4741 rny1Δ cells were transformed with the GFP-expressing construct or with plasmids expressing C-terminally tagged GFP fusions of Rny1p WT or the indicated mutants. Total protein lysates were subjected to western blotting with anti-GFP antibody. B) WT and och1Δ cells were transformed with the GFP-expressing plasmid or the Rny1pGFP WT-expressing construct. Cells were grown in synthetic medium, diluted with YPDA and incubated for an additional 2 h at 30°C. Yeast cells were treated with Tm where indicated. Cells were collected and total protein extracts were prepared. Protein extract was treated with EndoH where indicated. Proteins were analyzed by western blotting with anti-GFP antibodies. C) (top) Schematic representation of the FLAG and GFP-FLAG fusion constructs of Rny1p. (bottom) BY4741 rny1Δ cells were transformed with an empty vector (V) or plasmids expressing Rny1p WT fused with FLAG, GFP or GFP-FLAG. Total protein lysates were resolved by SDS–PAGE and analyzed by western blotting with anti-FLAG (right) and anti-GFP (left) antibodies. D) Cells from (C) were analyzed by the colony hybridization assay as described for Figure 3A. E) WT and vps1Δ cells were grown in YPDA and total protein extracts were analyzed by western blotting with anti-CPY antibody. F) WT and vps1Δ cells were transformed with the Rny1pGFP-FLAG or Rny1pFLAG-expressing constructs. Cells were grown in synthetic medium and total protein extracts were analyzed by western blotting with anti-FLAG antibodies.

Given this lack of Golgi-mediated glycosylation, we asked whether the GFP fusion protein could be secreted outside of cells. Placing GFP between the Rny1p coding sequence and the terminal FLAG tag (scheme in Figure 5C) completely abolished both Rny1p modification with HMW glycans (Figure 5C) and protein secretion (Figure 5D).

To ensure that the inability to detect Golgi-glycosylated forms of GFP-fusion proteins was not due to rapid protein degradation inside the vacuole, we examined Rny1pGFP-FLAG expression in vps1Δ cells. In this strain Golgi-to-vacuole trafficking is suppressed [36] and secretory proteins destined to the vacuole accumulate inside the Golgi compartment, as was evident from formation of the predominantly Golgi-specific (p2) form of CPY (Figure 5E). No expression differences were detected for Rny1pFLAG and Rny1pGFP-FLAG fusions when they were expressed in vps1Δ versus WT cells (Figure 5F). Moreover, lack of Golgi-specific modification of Rny1pGFP-FLAG observed in vps1Δ cells indicates that addition of GFP to the carboxy terminus of Rny1p indeed inhibits Golgi-mediated glycosylation (Figure 5F).

The similar outcome between adding the bulky GFP moiety and introducing the W339R mutation points to the important role of the C terminus of Rny1p for the correct Golgi-associated glycosylation and subsequent secretion of the protein. Collectively, our data demonstrate that adding the C-terminal GFP tag alters the normal protein glycosylation pattern and its localization. Without GFP, Rny1p can localize intra- and extracellularly and Golgi-mediated modification of this nuclease provides a secretion signal.

Intracellular Rny1pFLAG co-fractionates with the ER-Golgi compartment

Because GFP fusion clearly affects Rny1p transit along the secretory pathway, we reevaluated intracellular localization of this nuclease using the FLAG-tagged protein. We used a biochemical fractionation approach to investigate localization of Rny1p and its glycosylation status in different parts of the secretory compartment. The method used for purification of membrane organelles from cells expressing Rny1pFLAG WT was based on the methods developed for the separation of the Golgi and the vacuole [37]. Briefly, Rny1pFLAG-expressing cells were first converted to spheroplasts by treatment with Zymolyase, permeabilized and the cytosolic fraction was removed by centrifugation. The remaining cellular material was subjected to differential extraction with isotonic, low ionic strength buffer as previously described [37], resulting in dissociation of permeabilized cells into dispersed membrane organelles. These organelles were further separated by ultracentrifugation in sucrose gradients and individual fractions were analyzed by western blotting. To monitor migration of different organelles in the gradients, we used antibodies against endogenous marker proteins: Kar2p (ER [38]), mCPY (vacuole [39]), and Vps10p (late Golgi-endosome [40]). Although Vps10p cycles between the trans-Golgi-network (TGN) and endosomes, early work [40] demonstrated its co-localization with the late Golgi marker protein Kex2p, suggesting that it is reliable marker for the Golgi compartment. As shown in Figure 6A, the peak of predominantly vacuolar mCPY was present at the top of the 11–51% gradient in fractions 2–3, while distribution of the Golgi marker Vps10p (fractions 7–9) and the ER marker Kar2p (fractions 8–10) overlapped. As expected [37], increasing the percentage of sucrose in the gradient to 21–61% caused a visible shift in the distribution of organelles: vacuolar mCPY migrated in fraction 1, while Golgi marker protein Vps10p was present in fractions 6–8 and ER marker Kar2p accumulated in fractions 7–9 (Figure S6).

Figure 6.

Rny1pFLAG co-fractionates with the ER-Golgi compartment. A) Membrane-containing organelles were purified from BY4741 rny1Δ cells expressing Rny1pFLAG WT and separated by centrifugation in 11–51% (wt/wt) sucrose density gradients. Proteins were isolated from individual gradient fractions, resolved by SDS–PAGE and subjected to western blotting with antibodies against FLAG, Kar2p, CPY and Vps10p. Fraction numbers are indicated. B) Exponentially growing cells transformed with an empty vector or Rny1pFLAG were analyzed by immunofluorescence with the indicated antibodies. BF, bright field; DAPI, 4′,6-diamidino-2-phenylindole, was used to stain nuclei; scale bar is 5 µm. C) Membrane-containing organelles from cells expressing Rny1pGFP WT were analyzed as described in (A). Anti-GFP antibodies were used to detect Rny1pGFP and free GFP. D) The ratio of free GFP to Rny1pGFP for each fraction. Proteins from (C) were visualized by western blotting using anti-GFP primary and HRP-fused secondary antibodies and ECL detection. Chemiluminescence signals that correspond to the free GFP and Rny1pGFP in every fraction were measured using Kodak's Imaging System 400 and plotted as shown.

The gradient distribution of Rny1pFLAG WT revealed a pattern that was distinct from that of the mature vacuolar form of CPY (mCPY); Rny1p co-fractionated largely with the ER and Golgi, and only trace amounts were detectable in vacuolar fractions (Figure 6A). This pattern was reproducible using a different sucrose density in the gradient analysis (Figure S6). Furthermore, defects in glycosylation of Rny1p did not alter its distribution pattern between endomembrane organelles, as both Rny1pFLAG WT after Tm treatment (Figure S7A) and the Golgi-glycosylation deficient W399R mutant (Figure S7B) predominantly co-fractionated with the ER-Golgi compartment. As expected, no HMW forms indicative of Golgi-modified Rny1p were detected in the described biochemical fractionations, since only internal, spheroplast-derived cellular fractions were used in this assay (similar to what is depicted in lane 3, Figure 3E).

In support of the biochemical fractionation data, indirect immunofluorescence showed a predominantly ER localization of Rny1pFLAG WT. As shown in Figure 6B, the fluorescent signal specific for the nuclease co-localized with the signal derived from the ER marker Kar2p.

Similar to Rny1pFLAG, sucrose gradient analysis showed that Rny1pGFP co-sedimented largely with the ER-Golgi markers and only a small amount of the fusion protein co-fractionated with the vacuole (Figure 6C). The latter result contradicts our data obtained by microscopic analysis (Figures 2A, S2A). Strikingly, both vacuolar [1-3] and ER-Golgi [7-11] fractions from Rny1pGFP expressing cells contained a band corresponding in size to free GFP (Figure 6C) and the ratio of free GFP to Rny1pGFP was significantly higher in vacuolar fractions compared with ER-Golgi fractions (Figure 6D). A previous study showed that cleavage of a stable GFP moiety from fusion proteins was observed upon their delivery to the vacuole, leading to GFP accumulation due to its relative resistance to vacuolar degradative enzymes [41]. The same effect might contribute to the increased fluorescence signal in the vacuole in Rny1pGFP-expressing cells observed by microscopy analysis (Figures 2A and S2, also see Discussion). On the basis of these data, we suggest that the vacuole represents an alternative intracellular destination site for those forms of Rny1p that lack Golgi-mediated glycosylation, and that at least a portion of Rny1p undergoes degradation when presented to the vacuolar environment.

Rny1p rapidly traffics through the ER-Golgi and is secreted outside of the cell

In order to trace Rny1p in the above experiments we had to use overexpression of various tagged forms of this protein due to the low expression level of endogenous RNY1 (as previously shown in [13]) and, perhaps, due to rapid secretion of the nuclease. Therefore, we attempted to analyze distribution of the nuclease in the external and internal fractions when it is expressed at physiological levels. RNY1-FLAG WT and W399R mutant were cloned into a centromeric plasmid under the control of the endogenous promoter and expressed in rny1Δ cells. The external fraction, containing secreted proteins, was isolated using Zymolyase treatment as described previously (Figure 3E). Cell wall-free spheroplasts were used to purify membrane-bound secretory organelles and represent the internal fraction of the cell. Western blot analysis demonstrated that when expressed from the CEN plasmid, Rny1pFLAG WT is modified by Golgi-mediated glycosylation (HMW products) and is localized exclusively outside the cell (Figure 7A). In contrast, the Golgi-glycosylation deficient/secretion-incompetent W399R mutant (Figures 1D and 3A) was present only in the internal secretory organelle-derived fraction (Figure 7A). Next, we analyzed CEN-expressed Rny1pFLAG WT in WT and pep4Δ cells. Pep4p is the vacuole-resident protease that is required for the posttranslational maturation and activation of other vacuolar proteases, like CPY (bottom panel in Figure 7B). Thus, proteins that normally undergo vacuolar degradation will accumulate in pep4Δ cells (reviewed in [42]). We were unable to detect Rny1pFLAG WT in the vacuole of pep4Δ cells when the nuclease was expressed from the centromeric plasmid (first two lanes in Figure 7B). However, elevated expression of the nuclease (Rny1pFLAG WT was expressed from the GAL1 promoter during short galactose induction, Figure S8A) resulted in accumulation of Rny1p in the pep4Δ vacuole. Consistent with our model, Rny1p was present in the vacuole of pep4Δ cells in the ER-modified form only, as no HMW Golgi-glycosylated species were observed (third lane in Figure 7B). Taken together, these data suggest that when expressed at physiological level, Rny1p WT rapidly passes through the secretory compartment and gets secreted, avoiding capturing to the vacuole. Nuclease that fails to secrete remains trapped inside the secretory organelles. On the basis of these results, we propose that the large amount of the nuclease detected previously inside the ER-Golgi compartment of WT cells (Figures 6, S6 and S7) is likely a consequence of Rny1p overexpression.

Figure 7.

Rny1pFLAG transiently passes through the ER-Golgi compartment. A) Rny1pFLAG WT and W399R mutant were expressed from CEN plasmid in BY4741 rny1Δ cells and analyzed as described in Figure 3E. B) Rny1pFLAG WT was expressed from CEN plasmid in WT and pep4Δ cells (CEN-WT). Expression of Rny1pFLAG WT from GAL1 promoter in pep4Δ cells was induced for 1 h with galactose, followed by 1-h incubation in the glucose-containing media (GAL-WT). Vacuole was extracted from these cultures by biochemical fractionation like in Figure 6A and analyzed by western blotting with anti-FLAG and anti-CPY antibodies. C and D) BY4741 rny1Δ cells were transformed with GAL1-Rny1pFLAG WT and W399R mutant. Transformants were grown in minimal glucose-containing media and protein synthesis was induced by changing media to galactose-containing for 3.5 h (Gal-pulse), then replaced again with glucose-containing (Glu-chase). Aliquots of cell cultures were collected at the indicated time points. Cells were pelleted, lysed by glass-bead shearing and equal amounts of protein lysate were resolved by SDS–PAGE. The different forms of Rny1pFLAG (WTER, WTGolgi, W399R) were visualized by western blotting using anti-FLAG primary, HRP-fused secondary antibodies and ECL detection. Chemiluminescence signals that correspond to WTER, WTGolgi and W399R were measured using Kodak's Imaging System 400 and plotted as shown (D). E and F) Cells from (C) were induced with galactose for 1 h (Gal pulse), followed by cycloheximide chase (CHX chase) for 1 h. Cells were collected before and after CHX addition and analyzed by western blotting as described in Figure 7C. Chemiluminescence signals that correspond to Rny1pFLAG were measured using Kodak's Imaging System 400 and plotted as shown (F).

Therefore next, we wanted to investigate the dynamics of synthesis, trafficking through the ER-Golgi, and secretion of Rny1p. To control nuclease expression, both Rny1pFLAG WT and W399R were expressed from the GAL1 inducible promoter in galactose-containing media followed by a chase with glucose (Figures 7C and S8B). Three populations of proteins were monitored over time: Golgi-glycosylated Rny1pWT (migrate on the gel as HMW products, WTGolgi), ER-glycosylated Rny1pWT (migrate on the gel as 60 kDa protein, WTER) and Rny1pW399R. Modified only with ER core glycans the W399R mutant (Figure 1C,D) was used as a secretion-deficient protein control (Figures 3A,E and 7A). As seen previously and shown below, HMW forms of Rny1p (WTGolgi) represent an external, secreted nuclease, while the 60 kDa Rny1pWT (WTER) represents the intracellular enzyme (Figures 3E and 7A).

As shown in Figures 7C,D, all three proteins pools demonstrate a similar expression pattern during Gal-pulse/Glu-chase time course, reaching the maximal induction at 3.5 h of galactose treatment. However, W399R was expressed at double the level of the corresponding WTER, reflecting this mutant's inability to be secreted (Figure 7D, compare W399R and WTER graphs). Consistently, the half-life of W399R was increased compared to WTER (W399R T1/2 = 0.93 h; WTER T1/2 = 0.67 h, Figure S8D). Analyzing the decay rates of internal (WTER) and external (WTGolgi) fractions of cells expressing Rny1pWT, we noticed a significant delay in the decay of secreted Rny1p compared to the intracellular form of the enzyme (Figures 7C and S8C). Similar data were obtained by CHX-chase assay (Figure S9A–C). These results suggest continuous secretion of the nuclease after its synthesis is abolished, thus WTER is modified to WTGolgi and is secreted. To further corroborate this model, we induced expression of GAL1-Rny1pFLAG for 1 h only in order to generate predominantly the ER-modified forms of WT (lane 2, Figure 7C), followed by the short CHX chase. As expected, after protein synthesis was arrested, WTER completely converted to WTGolgi (Figure 7E,F), which were detected in the extracellular fraction only (Figure S9D). In agreement with our model, the W399R mutant remains associated with intracellular fraction (Figures 7E,F and S9D). Taken together, the results presented above demonstrate that the traffic of Rny1pWT through the early secretory pathway and its secretion are dynamic processes in which ER-Golgi-modified forms of an enzyme represent transient intermediates.

Discussion

In this work, we studied the molecular mechanisms underlying the intracellular distribution and sorting of Rny1p, the only member of the T2 family of secretory RNases expressed in Saccharomyces cerevisiae. We found that Rny1p undergoes progressive glycosylation that starts in the ER lumen and continues inside the Golgi apparatus. Modifications acquired in the Golgi play a key role in the extracellular secretion of this nuclease.

Glycosylated forms of several RNases from the T2 family have been detected in prior studies and glycosylation sites have been experimentally determined for some of these enzymes. For instance, RNase T2 from Aspergillus oryzae was found to be modified at N15, N76 and N239 [43], whereas nuclease Le37 from Lentinus edodes was shown to contain multiple N- and O-linked glycans [44]. In the recently resolved crystal structure of human RNase T2 (RNASET2), three N-linked glycosylation sites were identified, N76, N106 and N212 [45]. However, the function of these modifications for the most part remains unclear. In one study, the functional role of glycan modifications was investigated for the nuclease Omega-1 from the parasitic blood fluke Schistosoma mansoni. When secreted by the parasite's eggs, Omega-1 is glycosylated at N71 and N176 and these modifications allow it to interact with mannose receptors on dendritic cells. This is followed by internalization and rRNA degradation inside the host cells, which in turn leads to the activation of the immune response [46].

Modification of the Rny1GFP fusion protein with glycans was demonstrated previously [16]. In our study, we utilized Rny1pFLAG as an alternative to GFP fusion. Importantly, we detected two distinct glycosylated forms of Rny1p: one migrated on a gel 10 kDa higher than the unmodified protein, and the other migrated as a heterogeneous HMW smear (Figure 1B–D). Using different genetic and biochemical approaches, we concluded that the two observed modified forms of Rny1p are formed via glycosylation occurring inside the ER and Golgi, respectively (Figures 1C,D and 4C,E). A total of four sites for N-linked glycosylation were identified (Figure 1A,E). Furthermore, modification of Rny1p with ER core glycans was found to be dispensable for the nucleolytic activity of the enzyme, indicating that it can be properly folded without glycan addition (Figure 2B,C). This result, surprising at first, is in agreement with the recent findings showing that glycosylation of human RNASET2 is not required for the catalytic activity of this enzyme [45].

Sugar moieties on Rny1p also did not appear to influence progression of this protein along the ER-Golgi-vacuole path and capturing in the vacuole. This conclusion is based on several lines of evidence. First, the forms of Rny1p that are normally glycosylated (WT), glycosylated with ER core glycans only (W399R), underglycosylated (in alg5Δ) and unglycosylated (Tm-treated cells) localized inside cells in a similar manner (i.e. inside the vacuole), as revealed by live imaging of GFP fusions (Figures 2A and S2). Second, similar patterns of intracellular distribution were observed by biochemical fractionation of FLAG-tagged proteins (Figures 6A, S6 and S7). In addition, we identified Erv29p as an adaptor required for Rny1p capturing into COPII vesicles during anterograde ER-to-Golgi trafficking (Figures 4 and S2K). Interestingly, Erv29p was shown to recognize specific amino acid sequences on client proteins rather than sugar moieties [30]. In marked contrast to the intracellular distribution data, we found that Golgi-mediated glycosylation is essential for Rny1p extracellular secretion (Figures 3, 7A and S9D).

Because multiple mechanisms may control protein transport from the late Golgi to the cell surface, endosomes and the vacuole/lysosome, sorting signals vary for different proteins (reviewed in [47]). For several mammalian secretory proteins including erythropoietin, clusterin and human corticosteroid binding globulin (hCBG), glycosylation serves as a sorting signal required for transport from the Golgi compartment to the cell surface (discussed in [48]). Moreover, introduction of N-glycosylation sites into non-glycosylated soluble or membrane proteins can target them to the cell surface, indicating that N-glycans can operate as sorting/targeting signals [48, 49]. Our results in this study suggest that Golgi-mediated glycosylation of the yeast T2 RNase Rny1p serves as a signal for protein secretion. The next question that needs to be addressed in future studies is how the oligosaccharide chain(s) on this RNase promotes its export. One possibility may be interaction with a specific lectin-like protein that mediates Rny1p targeting to the cell surface for secretion, as shown for several mammalian secretory proteins (reviewed in [47]).

Surprisingly, we found that the C-terminal GFP fusion Rny1p protein, used in previous studies [10], has an abnormal glycosylation pattern (Figure 5A–C) and is not extracellularly secreted (Figure 5D). The microscopic analysis of GFP fusion Rny1p proteins, however, demonstrated a strong fluorescent signal in the vacuole (Figures 2A and S2A), indicating that Rny1p lacking a secretion signal can travel to the vacuolar compartment.

Biochemical fractionation detected that, similarly to FLAG fusion, Rny1pGFP co-sedimented predominantly with the ER-Golgi compartment, while most of the vacuole-sedimented Rny1pGFP undergoes degradation, resulting in liberation and accumulation of free GFP (Figure 6C,D). Free GFP is known to be resistant to the degradative environment of the vacuole [41] and might simply mask the weaker signal derived from the ER-localized Rny1pGFP fusion. This can explain why Rny1pGFP was not detected in the ER structures by fluorescence microscopy (Figures 2A, S2A, and 4A) whereas Rny1pGFP does display a typical ER localization pattern in the erv29Δ strain (Figures 4B and S2K). Alternatively, the inability to observe Rny1pGFP in the ER could be due to the immature state of the GFP chromophore when Rny1pGFP transits through the ER-Golgi segment of the secretory route. In fact, GFP emits fluorescence after it becomes mature and this is a time consuming event (reviewed in [50]). As shown in Figures 7, S8 and S9, Rny1p rapidly passes through the ER-Golgi compartment; therefore, the GFP portion of Rny1pGFP fusion is likely still immature at this stage. Therefore, it is possible that the GFP fluorescence data does not accurately report on the major fate of Rny1pGFP fusion proteins.

An important question that arises from these observations is whether Rny1p has a dual localization and function (in the vacuole and extracellular environment), or if it is only delivered to the vacuole for degradation when Golgi modification fails. Interestingly, two other yeast secretory proteins, invertase and CPY, are known to localize both inside and outside of cells. The molecular mechanisms underlying their dual localization, however, are different. SUC2 encodes two forms of invertase that diverge from each other by the presence of an N-terminal signal peptide, as they are derived from two distinct mRNAs that differ only in their 5′ ends [51, 52]. The signal peptide directs the extracellular version of invertase into the ER resulting in secretion, while the other form remains in the cytosol. Unlike invertase, CPY progresses through the conventional secretory route to the vacuole. However, mutations in the CPY sorting signal or overproduction of CPY in a cell leads to missorting and secretion of the enzyme [53, 54]. CPY is a vacuolar peptidase, and its secretion outside the cells is thought to be a discard route. Playing an opposite role to CPY, T2 RNases function in the extracellular environment and the vacuole might represent a place for destruction of the nuclease when required. It is possible that lack of a Golgi-generated secretion signal or overexpression of Rny1p from 2μ plasmids could cause missorting and/or saturation of the export branch at the Golgi, resulting in the increased delivery of the nuclease to the vacuole.

Experimental support for this idea comes from our experiments where Rny1p was expressed from its endogenous promoter on the low-copy centromeric plasmid (pRS416). When expressed at physiological levels, Rny1pWT was detected in the external, but not internal fraction of the cell, exclusively as a Golgi-modified form (Figure 7A). Consistent with these data, we were unable to detect CEN-expressed Rny1p in the vacuole of protease-defective pep4Δ cells (CEN-WT, Figure 7B). This result indicates that: (i) RNY1 is expressed at the low levels (as was demonstrated previously [13] and shown in Figure S8A), and (ii) upon synthesis, Rny1p undergoes rapid secretion. In fact, the dynamic mode of synthesis, trafficking through the ER-Golgi followed by secretion is evident from the pulse-chase experiments performed by two alternative techniques: Gal-pulse/Glu-chase and CHX-chase (Figures 7C–F, S8B–D and S9). Elevated expression of Rny1p WT from the GAL1 promoter (Figure S8A) showed accumulation of the ER-, but not Golgi-modified version of the nuclease in the vacuole of pep4Δ cells, supporting that Golgi-glycosylated Rny1p is not destined for this compartment (Figure 7B).

Interestingly, upon secretion, Rny1p remains associated with the cell wall. In fact, we were unable to purify the nuclease from the media of Rny1pFLAG WT-expressing cultures. However, when these cells were treated with Zymolyase and the supernatant was analyzed by western blotting, HMW forms of this protein (that correspond to Golgi-glycosylated forms, Figure 1D) were readily visible (Figures 3E and 7A). As expected, the secretion-deficient W399R mutant remained present within membrane organelles (Figures 3E and 7A). The fact that Rny1p remains associated with the cell wall after secretion indicates that this nuclease might function as an accessory enzyme, required for physiological needs of the cell. Thus, it is possible that like other T2 family RNases [55], Rny1p plays an important role in scavenging nutrients during starvation.

Similar turnover kinetics of glycosylation-deficient Rny1p had been detected in WT and ERAD-deficient npl4-1 cells (Figure S1). These results suggest that mutants, used in this study, are not substrates for ERAD. However, some misfolded proteins had been reported to accumulate in the ER avoiding ERAD [56, 57]. For instance, among mutants of Ste6p that are retained in the ER, some are degraded by ERAD, while others are not [56]. Unlike CPY*, CPYΔ1 (truncated version of CPY) is not destroyed by ERAD due to a missing ERAD-determinant located at the C terminus of the protein. Instead, CPYΔ1 is retained in the ER, causing activation of unfolded protein response (UPR) [57]. In light of these studies, it was reasonable to propound that Rny1p mutants (at least some of them) would avoid ERAD and remain associated with the ER due to misfolding and/or high level of expression. In fact, overexpression might result in a situation where only a small fraction of Rny1p mutants escape ER quality control and end up trafficking to the vacuole, while most of the protein population is retained in the ER. However, a few lines of evidence argue against this possibility: (i) Inability to exit the ER would result in accumulation of Rny1p mutants inside this compartment, resembling Rny1p localization in erv29Δ cells (Figures 4A,B and S2K). However, live fluorescent microscopy demonstrated localization of Rny1pGFP fusions (WT and various mutants) exclusively in the vacuole, while no fluorescent signal had been observed in the ER (Figures 2A and S2); (ii) Stable accumulation of Rny1p mutants in the ER would likely cause ER-stress leading to activation of UPR [58], as had been shown for CPYΔ1 [57]. However, expression of none of the Rny1p mutants resulted in the appearance of the ER-stress marker HAC1i – the spliced form of the HAC1 transcript [59] (NS personal observations); (iii) Finally, retention in the ER and inability to be degraded by ERAD would cause accumulation and stabilization of the protein. However, our kinetic studies performed by several alternative approaches demonstrated rapid trafficking of Rny1p WT and W399R through the secretory pathway (Figures 7, S8 and S9). Therefore, we believe that the strong Rny1p-specific signal detected in the ER-Golgi fractions in biochemical fractionation studies (Figures 6, S6 and S7) could be explained by a combination of high Rny1p expression level, rapid trafficking through the ER-Golgi, and the large capacity of the ER-Golgi sector of the secretory pathway.

A growing body of evidence has demonstrated the ability of Rny1p and related family members to degrade cytoplasmic rRNAs and tRNAs inside cells. To date, a scenario in which secreted Rny1p enters into the cytoplasm of neighboring cells, analogous to T2 RNase Omega-1 and RNase A angiogenin (ANG) [46, 60], has not been experimentally proven. Moreover, we have found that secretion-deficient mutants are as active in rRNA degradation as Rny1p WT (Figure 2B,C), indicating that secretion is not a prerequisite for the enzyme to be active in the cytoplasm. These data were supported by co-culturing experiment (Figure S4A,B). One possibility is that a certain amount of vacuolar Rny1p may remain intact and exit the vacuole during oxidative stress as previously proposed [10]. Alternatively, Rny1p could be released into the cytoplasm from other organelles in the secretory compartment. Clearly, more studies are required to understand the mechanisms leading to Rny1p-mediated tRNA and rRNA degradation.

A C-terminal extension is a unique feature of fungal T2 RNases. Deletion of this C-terminal extension did not affect cell viability or nucleolytic activity of Rny1p [16]. Our results show that manipulation of the C-terminal end of Rny1p, such as the attachment of a bulky GFP tag or mutation of W399, results in proteins that are not glycosylated in the Golgi (Figures 1B,D and 5A,B) and fail to be secreted (Figures 3A and 5D). These data indicate that the C-terminal extension of Rny1p is essential for the proper modification with a secretion signal in the Golgi system. We suggest that the C terminus contributes directly or indirectly to the recognition/binding of Golgi-resident enzymes that promote further protein glycosylation.

Materials and Methods

Yeast strains, media, antibodies and chemicals

All yeast strains used in this study were purchased from Open Biosystems. RNY1, ALG5, OCH1, ERV29, ERV26, EMP24, PEP4, ATG1, HRD1, UBC7, DOA10 and VPS1 were replaced with kanamycin cassette on BY4741 background strain (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0). npl4-1 strain (PSY2340, MATa, ura3-52, leu2Δ1, trp1Δ63 and npl4-1) was a kind gift of Dr. Pamela Silver. We used standard recipes for YPDA (1% yeast extract, 2% peptone, 2% dextrose, 10 mg/L adenine) and synthetic glucose or galactose-containing media. The polyclonal and monoclonal M2 anti-FLAG antibodies were purchased from Sigma; the anti-Kar2p y-115 antibodies were from Santa Cruz Biotechnology; the anti-CPY 10A5B5 and the anti-Vps10p 18C8 antibodies were from Molecular Probes, and the anti-GFP antibody 7.1/13.1 was from Roche Biochemicals. Tm was purchased from Calbiochem and used at a final concentration of 1 µg/mL. Cycloheximide was purchased from Sigma and used at a final concentration of 30 µg/mL.

Plasmids

Full length RNY1 was amplified by PCR from genomic yeast DNA with primers complementary to the 5′ and 3′ ends of the RNY1 coding sequence. To generate an untagged version of Rny1p (Rny1pNO_TAG), the reverse (3′) primer with no additional sequences was used. To generate a C-terminal FLAG-fusion of Rny1p (Rny1pFLAG), the reverse (3′) primer containing a sequence that encoded the FLAG tag was used. To generate the ΔN truncation mutant, the forward (5′) primer was chosen to anneal 54 bp downstream of the start codon of RNY1. PCR products were digested with ClaI and KpnI and cloned into ClaI and KpnI sites of the PUAD vector (a kind gift of Randy Strich). The PUAD plasmid was created by cloning the BamHI/EcoRI 1.5 kb ADH promoter region from pBC102 into BamHI/EcoRI sites of pRS426. Site-directed mutagenesis was performed as previously described [61] to generate H87R, N37Q, N70Q, N103Q and N123Q mutants using the PUAD-Rny1pFLAG WT construct as a template. Sequences of all primers used in this study are available upon request. To place Rny1pFLAG WT and W399R under control of the GAL1 inducible promoter, RNY1 WTFLAG, and W399RFLAG sequences were cut-off from PUAD-WTFLAG and PUAD-W399RFLAG using HindIII and KpnI digestions and inserted into pYes between HindIII and KpnI. To generate the CEN-RNY1FLAGWT construct, the RNY1 coding sequence with 5′-untranslated region was amplified with a forward (5′) primer complementary to the region located 900 bp upstream of the start codon of RNY1 and with a reverse (3′) primer complementary to the 3′ ends of the RNY1 coding sequence. Genomic yeast DNA was used as a template. PCR products were digested with SacII (5′ end) and KpnI (3′ end) and cloned into the SacII/KpnI sites of the pRS416. To generate CEN-RNY1FLAGW399R construct, a 3′ region located 240 bp downstream of the start codon of RNY1-WT was replaced with the similar sequence containing the W399R mutation. To do this, PUAD-W399R was digested with XbaI/KpnI and the 1.1 kb product (containing RNY1-W399R sequence without the first 240 bp) was ligated with CEN-RNY1FLAGWT1-240bp generated by digestion with XbaI and KpnI. pYes-GFP was generated by cloning the EGFP sequence (amplified by PCR from pEGFP-C3, Clontech) into the pYes plasmid (Invitrogen) between the NotI (5′) and XbaI (3′) sites. The region corresponding to the ADH promoter was inserted into pYes-GFP between the BamHI and EcoRI sites, resulting in pYes-ADH-GFP. RNY1 WT and mutants were amplified by PCR from the corresponding PUAD constructs and cloned between the EcoRI and NotI sites of pYes-ADH-GFP in-frame with GFP. To generate PUAD-Rny1pGFP-FLAG double fusion, pYes-ADH-Rny1pGFP WT was used as a template to amplify the RNY1-GFP region, where the reverse (3′) primer contained a sequence corresponding to the FLAG tag. The PCR product was then cloned into PUAD between the ClaI and KpnI sites.

To generate pRS313-CPY*FLAG, the sequence corresponding to CPY* was amplified from pRS316-prc1-1 (ZKb085, kind gift of Dr. Weissman) using a forward primer (5′) that anneals 700 bp upstream of the CPY coding region and reverse (3′) primer that anneals to the 3′ end of CPY. The reverse primer contained a sequence that encoded the FLAG tag. The PCR product was digested with SmaI (5′) and SpeI (3′) and ligated with pRS313 using the same cloning sites.

Live fluorescent microscopy

Cells transformed with pYes-ADH-Rny1pGFP constructs were grown in synthetic dextrose medium (SD-ura) overnight, diluted in YPDA to OD600 ˜ 0.3, and grown for an additional 2–3 h at 30°C. To visualize the vacuole, cells were incubated in medium containing 40 µm FM4-64 (Molecular Probes) for 15 min at 30°C; transferred to fresh medium and chased for 1 h. Cells were plated on glass slides coated with concanavalin A (Sigma) and examined using a Zeiss Apotome microscope.

Immunofluorescence

Immunofluorescent analysis of Rny1pFLAG WT and Kar2p was performed as described previously [62] with a few modifications. Briefly, BY4741 rny1Δ cells transformed with an empty vector or Rny1pFLAG WT were grown in SD-ura overnight, diluted in YPDA to OD600 ˜ 0.3 and grown for an additional 2–3 h. Cells were fixed with 3.7% formaldehyde for 15 min, converted to spheroplasts, and placed on concanavalin A-treated glass slides. Cells were incubated with mouse monoclonal anti-FLAG and rabbit polyclonal anti-Kar2p antibodies for 1 h at room temperature (RT). The corresponding secondary antibodies, TX-conjugate anti-mouse IgG and FITC-conjugate anti-rabbit IgG, (Molecular Probes) were added for 30 min at room temperature. Finally, cells were placed in mounting media supplemented with DAPI (Vector Labs) and visualized using a Zeiss Apotome microscope.

Membrane organelle extraction and fractionation

Exponentially growing cells were converted to spheroplasts and frozen in liquid nitrogen [63]. Membrane extraction/fractionation was performed as described previously [37]. Briefly, after thawing on ice, spheroplasts were washed with 250 mm sorbitol, followed by washes with 50 mm sorbitol, 250 mm sorbitol/2 m potassium acetate, and 250 mm sorbitol. All sorbitol solutions were prepared in transfer buffer: 20 mm HEPES–KOH pH 6.8, 150 mm potassium acetate, 5 mm magnesium acetate. All washes were performed for 3–5 min on ice and cells were harvested by centrifugation for 45 seconds at 13 000 × g. The resulting pellet (permeabilized cells) was subjected to three consecutive extractions with 0.8 m sorbitol, 10 mm TEA pH 7.6, and 1 mm EDTA buffer (TEA buffer) by passing the solution 5–7 times through a 26 gage syringe needle. After each extraction, material was separated on the pellet and supernatant fractions by centrifugation for 5 min at 13 000 × g. Combined supernatants were centrifuged further at 21 000 × g for 20 min. Pellet was resuspended in 300–500 μL of TEA buffer and loaded on the top of a sucrose gradient. Sucrose solutions were made in 10 mm HEPES–KOH pH 7.6 buffer with ultrapure sucrose (Sigma). The gradients were subjected to ultracentrifugation in a Beckman SW41Ti rotor at 170 000 × g for 19 h and fractionated using a Beckman fraction recovery system. Proteins were precipitated from the sucrose fractions with 10% TCA; protein pellets were washed once with ethanol, once with acetone, dried, and dissolved in 1× SDS–PAGE loading dye.

Protein analysis and deglycosylation assay

Total protein lysates were prepared as described previously [61]. Briefly, yeast cell pellets were resuspended in RIPA buffer (50 mm Tris–HCl pH 8.0, 150 mm NaCl, 0.1% SDS, 1% Nonidet P-40, 0.5% deoxycholic acid, 50 mm NaF, 40 mm Na3P2O7) supplemented with PMSF and protease inhibitors (Pierce). Cells were broken open by vortexing with glass beads (Sigma) at 4°C; cellular lysates were clarified by centrifugation at 10 000 × g for 10 min and protein concentrations were determined by the Bradford assay (BioRad). For western blotting, 7–10 µg of a total protein extract in 1× SDS–PAGE loading dye was boiled prior to SDS–PAGE. Proteins were transferred onto a nitrocellulose membrane (Whatman) and probed with the antibodies indicated in the figures.

For the deglycosylation assay, cells transformed with PUAD-Rny1pFLAG WT or empty vector were grown in SD-ura, pelleted and lysed in RIPA buffer as described above. After clarification, protein concentration was measured by the Bradford assay and 600 µg of total protein extract was incubated with anti-FLAG M2 agarose beads (Sigma) for 4 h at 4°C. Beads were washed once with RIPA buffer and once with 1× EndoH reaction buffer (NEB). During the second wash, beads were split into two tubes. One unit of EndoH (NEB) was added to the beads resuspended in 20 μL of 1× EndoH reaction buffer in the first tube, while empty buffer was added to the beads in the second tube. Reactions were incubated at 37°C for 1 h, beads were washed three times with RIPA buffer, and proteins were eluted from the beads by boiling in SDS–PAGE loading dye.

Colony immunoblotting assay

The colony immunoblotting assay was performed as described previously [25]. Briefly, yeast cultures were adjusted to the same density (initial OD600 of 1.0) and a fivefold dilution series were spotted on a YPDA agar plate in duplicate. One plate was left intact, while the other was overlayed with a piece of nitrocellulose membrane (Whatman). Plates were incubated at 30°C for 24–36 h. Yeast growth on the first plate demonstrated equal growth of the examined cultures. Membranes were removed from the second plate, residual cells were washed off the membranes with a stream of water, and the membranes were subjected to western blotting with anti-FLAG antibody.

Secretion assay

Biochemical secretion assay was conducted as described previously [27]. Briefly, exponentially growing yeast cultures were harvested and incubated 10 min at room temperature in 100 mm Tris–HCl pH 9.4, 10 mm DTT buffer. Cells were collected, resuspended in 1.2 m sorbitol/20 mm potassium phosphate pH 7.4 and treated with Zymolyase 100T (Sigma) at 30°C (0.5 mg of enzyme was added to 1 g of packed cells). Efficiency of spheroplasting was determined by microscopy. The internal (cellular) and external (cell wall) fractions were separated by centrifugation. The supernatant (external fraction) and pellet (internal fraction) were analyzed by western blotting.

Gal-pulse/Glu-chase and CHX-chase assays

Yeast cells, transformed with pYes-Rny1pFLAG WT or W399R, were grown in synthetic glucose-containing medium to the mid-log phase at 30°C; cells were collected, washed with water to remove glucose, and resuspended in synthetic galactose-containing medium. Galactose induction lasted 3.5 h, followed by switch back to glucose-containing medium. Aliquots of yeast cultures were collected every 1–2 h during Gal-pulse and Glu-chase. Cells were pelleted, lysed and equal amounts of protein extracts were resolved by SDS–PAGE. Proteins were transferred to nitrocellulose membrane and analyzed by western blotting. Ponseau S staining of membranes prior to immunoblotting revealed equal amounts of protein material present in each lane.

In the CHX-chase assay, yeast cultures expressing Rny1pFLAG WT or W399R mutant from the PUAD plasmid, were grown in synthetic glucose-containing medium to mid-log phase and 30 µg/mL of CHX was added. Aliquots of cells were collected at 0, 1, 2, 4 and 6 h after CHX addition. Cells were pelleted, lysed and protein extracts were analyzed by western blotting.

Western blotting for both assays was performed with anti-FLAG primary and HRP-fused secondary antibodies, and protein signals were visualized using an ECL detection system (Millipore). Chemiluminescence signals that correspond to different forms of Rny1pFLAG were measured using Kodak's Imaging System 400. Nonlinear regression analysis of the data was performed with Prism 5 (GraphPad Software, Inc), assuming exponential equation and one-phase decay.

RNA analysis

Yeast RNA was isolated by the acid phenol method as described previously [64]. To extract RNA from the media of growing yeast cultures, cells were pelleted by centrifugation, media was collected, centrifuged through the PVDF 0.45 µm filter spin-column (Millipore) to remove unpelleted cells, and subjected to phenol/chloroform extraction, followed by isopropanol precipitation. RNA concentrations were determined and 2 µg of total RNA was resolved on a 1.2% agarose gel containing 1.3% formaldehyde as described in [65]. RNA was transferred to nylon membranes (Hybond N, GE Biosciences) and detected by Northern hybridizations using 32P-labeled oligonucleotide probes as described in [66]. We used y503 probe against 25S rRNA (5′-ACCCACGTCCAACTGCTGT); y500 probe against 18S rRNA (5′-AGAATTTCACCTCTGACAATTG), y600 probe against tag on 25S rRNA (5′-GGGCAGGCTGCAGCTTCCTACCAG), and tRNA-Val probe (5′-TGGTGATTTCGCCCAGGA). Hybridizations were analyzed using a Typhoon 9200 PhosphorImager and ImageQuant software (GE Biosciences).

Nucleolytic activity staining

Protein extracts were prepared as described above, except that NMT buffer (100 mm NaCl, 3 mm MgCl2, 10 mm Tris–HCl at pH 7.4) was used instead of RIPA buffer. Equal amounts of protein lysates (5–7 µg) from yeast cells expressing Rny1pFLAG WT and the different mutants were mixed with loading dye to a final concentration of 5% glycerol and 0.0125% bromphenol blue in 25 mm Tris–HCl buffer pH 6.8. Samples were loaded on an 8% SDS-free polyacrylamide gel supplemented with 2.4 mg/mL of RNA from Torula yeast (Sigma). Gels were run at the constant current of 5 mA for several hours. Gels were stained with toluidine blue O (Sigma) as described previously [67]. De-stained gels were scanned for visualization.

Quantification of mRNA levels

RNA was extracted as described in [64], treated with DNase and 1 µg of RNA was used in the reverse transcription (RT) reaction with RT-MuMLV (1 h, 42°C). 10% of RT reaction was then used in qPCR with Kappa SYBR FAST reagent. 200 nm of forward and reverse RNY1 or ACT1 specific primers were added into reaction. Sequences of all the primers are available upon request. qPCR was performed on Eppendorf Mastercycler Realplex machine; the cycling parameters were as follows: denaturation 95°C for 15 seconds, annealing 55°C for 15 seconds and extension 72°C for 20 seconds (cycle repeated 40 times). Each reaction was done in triplicate. The RNY1 values were normalized to those for ACT1 and presented as the relative fold change compared to the vector-alone control cells. To estimate fold change of RNY1 expressed from GAL1 over CEN-expressed RNY1 in pep4Δ cells the GAL1-WT data were also presented as the relative fold change compared to the CEN-WT expression.

Acknowledgments

I would like to express special gratitude to Dimitri Pestov and Randy Strich for continues stimulating discussions, guidance and help in manuscript preparation. I am grateful to Randy Hampton, Dimitri Pestov, Pamela Silver, Randy Strich and Alan Weissman for providing plasmids and strains. I am very thankful to Alexey Bulychev and Farrah Mansour for technical help. I would like to thank Renee Demarest and Stephen Kim for critical reading of the manuscript. The work was supported by the AHA grant 09SDG2140065 and a UMDNJ Foundation grant to N. S.

Ancillary