Portal de Periódicos da CAPES

Subcellular distribution of proteasomes implicates a major location of protein degradation in the nuclear envelope-ER network in yeast

1998; Springer Nature; Volume: 17; Issue: 21 Linguagem: Inglês

10.1093/emboj/17.21.6144

1460-2075

Cordula Enenkel, Andrea Lehmann, Peter‐Michael Kloetzel,

Nuclear Structure and Function

Article2 November 1998free access Subcellular distribution of proteasomes implicates a major location of protein degradation in the nuclear envelope–ER network in yeast Cordula Enenkel Corresponding Author Cordula Enenkel Institut für Biochemie, Humboldt Universität, Universitätsklinikum Charité, Monbijoustrasse 2, D-10117 Berlin, Germany Search for more papers by this author Andrea Lehmann Andrea Lehmann Institut für Biochemie, Humboldt Universität, Universitätsklinikum Charité, Monbijoustrasse 2, D-10117 Berlin, Germany Search for more papers by this author Peter-Michael Kloetzel Peter-Michael Kloetzel Institut für Biochemie, Humboldt Universität, Universitätsklinikum Charité, Monbijoustrasse 2, D-10117 Berlin, Germany Search for more papers by this author Cordula Enenkel Corresponding Author Cordula Enenkel Institut für Biochemie, Humboldt Universität, Universitätsklinikum Charité, Monbijoustrasse 2, D-10117 Berlin, Germany Search for more papers by this author Andrea Lehmann Andrea Lehmann Institut für Biochemie, Humboldt Universität, Universitätsklinikum Charité, Monbijoustrasse 2, D-10117 Berlin, Germany Search for more papers by this author Peter-Michael Kloetzel Peter-Michael Kloetzel Institut für Biochemie, Humboldt Universität, Universitätsklinikum Charité, Monbijoustrasse 2, D-10117 Berlin, Germany Search for more papers by this author Author Information Cordula Enenkel 1, Andrea Lehmann1 and Peter-Michael Kloetzel1 1Institut für Biochemie, Humboldt Universität, Universitätsklinikum Charité, Monbijoustrasse 2, D-10117 Berlin, Germany *Corresponding author. E-mail: [email protected] The EMBO Journal (1998)17:6144-6154https://doi.org/10.1093/emboj/17.21.6144 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info 26S proteasomes are the key enzyme complexes responsible for selective turnover of short-lived and misfolded proteins. Based on the assumption that they are dispersed over the nucleoplasm and cytoplasm in all eukaryotic cells, we wanted to determine the subcellular distribution of 26S proteasomes in living yeast cells. For this purpose, we generated yeast strains that express functional green fluorescent protein (GFP) fusions of proteasomal subunits. An α subunit of the proteolytically active 20S core complex of the 26S proteasome, Pre6/YOL038w, as well as an ATPase-type subunit of the regulatory 19S cap complex, Cim5/YOL145w, were tagged with GFP. Both chimeras were shown to be incorporated completely into active 26S proteasomes. Microscopic analysis revealed that GFP-labelled 20S as well as 19S subunits are accumulated mainly in the nuclear envelope (NE)–endoplasmic reticulum (ER) network in yeast. These findings were supported by the co-localization and co-enrichment of 26S proteasomes with NE–ER marker proteins. A major location of proteasomal peptide cleavage activity was visualized in the NE–ER network, indicating that proteasomal degradation takes place mainly in this subcellular compartment in yeast. Introduction Metabolic and transcriptional control mechanisms, signal transduction and cell cycle progression depend on specific regulatory mechanisms including protein breakdown. The central enzyme complexes responsible for selective proteolysis of short-lived proteins are 26S proteasomes. Proteasomal substrates include metabolic key enzymes, transcription factors and cyclins, that are found to be targeted to proteasomal breakdown by conjugation to ubiquitin (Hershko and Ciechanover, 1992; Hochstrasser, 1995; Coux et al., 1996; Hilt and Wolf, 1996). The 26S proteasome consists of a proteolytically active 20S core complex and a regulatory 19S cap complex. In the yeast Saccharomyces cerevisiae, the subunit arrangement of the 20S core complex has been resolved by X-ray crystallography (Groll et al., 1997). Fourteen different subunits, classified into seven α- and seven β-type subunits, are arranged in four heptameric rings stacked into a hollow cylinder. Both outer rings contain the α-type subunits, both inner rings the β-type subunits. The interior of the 20S particle harbours the catalytic centres of the protease. N-terminal threonine residues of certain β-type subunits form the central active sites that are liberated by precursor processing during proteasome assembly (Chen and Hochstrasser, 1996; Schmidtke et al., 1996; Seemueller et al., 1996; Heinemeyer et al., 1997). Different peptide cleavage preferences could be assigned to distinct β-type subunits by using yeast genetics (Hilt et al., 1994; Arendt and Hochstrasser, 1997). The regulatory 19S cap complex of the 26S proteasome confers the specificity towards ubiquitinated substrates and an ATP dependence on proteolysis (Dubiel et al., 1995; Tanaka and Tsurumi, 1997). In yeast, the 19S cap complex comprises six AAA-type ATPases and 11 non-ATPases (Glickman et al., 1998). The ATPase subunits were proposed to function as reverse chaperones in substrate unfolding and translocation into the proteolytic cavity of the proteasome (Confalonieri and Duguet, 1995). Ubiquitin binding and isopeptidase activities were assigned to non-ATPase subunits, responsible for editing and rescuing ubiquitin moieties from the substrate prior to degradation (Deveraux et al., 1994; Lam et al., 1997). Different genetic approaches using yeast mutants deficient in degradation of short-lived proteins or defective in cell cycle progression identified several 26S components and provided considerable insight into the in vivo functions of 26S proteasomes. In particular, the timely progression of the cell cycle turned out to be tightly regulated by selective proteolysis through proteasomes (Ghislain et al., 1993; Gordon et al., 1993; Friedman and Snyder, 1994; Kominami et al., 1995, 1997; Hilt and Wolf, 1996; McDonald and Byers, 1997; Wilkinson et al., 1997). Genetic studies in yeast further revealed that the ubiquitin–proteasome machinery is also involved in endoplasmic reticulum (ER) degradation (Hampton et al., 1996; Hiller et al., 1996). A retrograde transport mediated by ER translocons and the ER-resident chaperone Kar2/Bip is required to transport abnormal proteins from the ER lumen to the cytoplasmic face for final proteolysis (Plemper et al., 1997). In higher eukaryotic cells, immunolocalization and biochemical fractionation studies showed a cytoplasmic and nucleoplasmic localization of proteasomes (Peters et al., 1994). The amounts of nuclear proteasomes were found to be increased in mitotic cells (Amsterdam et al., 1993; Palmer et al., 1996), particularly in highly proliferating cells during early stages of Drosophila embryogenesis, during Xenopus oocyte maturation and in tumorigenic cells (Klein et al., 1990). In human fibrosarcoma cells, a subpopulation of green fluorescent protein (GFP)-tagged proteasomes was shown to be taken up mainly into the nucleus upon the reassembly of the nuclear membrane during mitosis, but was also reported to be transported unidirectionally over the nuclear membrane (Reits et al., 1997). In order to study in vivo the subcellular distribution of yeast 26S proteasomes, we tagged the Pre6/YOL038w α-type subunit of the 20S core complex (Heinemeyer et al., 1994) as well as the Cim5/YKL145w ATPase-type subunit of the 19S regulatory complex (Ghislain et al., 1993) with GFP fused to a double hemagglutinin (HA) epitope. The GFP-labelled proteasome subunits were shown to be incorporated into active 26S proteasomes, thus providing suitable reporters of proteasomal degradation in living yeast cells. Using different techniques, proteasomes were observed to be enriched in the nuclear envelope (NE)–ER network independently of the cell and life cycle phases. Furthermore, the analysis of NE–ER fractions showed the majority of proteasomes to be bound to membrane structures. Proteasomal peptide cleavage occurred mainly in the NE–ER network, suggesting a major degradation location in this subcellular compartment in yeast. Results Generation of yeast strains expressing functional GFP-tagged proteasomes In higher eukaryotes, 26S proteasomes are dispersed over the cytoplasm and nucleoplasm (Peters et al., 1994; Reits et al., 1997). In this work, we set out to study in detail the in vivo distribution of proteasomes in yeast. Our experimental approach was based on GFP technology, since GFP fusion proteins have become the tool of choice for visualizing molecular events in vivo by direct fluorescence microscopy (Cubitt et al., 1995). To provide suitable reporters of the subcellular localization of 26S proteasomes in living cells, we generated yeast strains that express functional GFP-tagged proteasomal subunits. The α-type subunit Pre6/YOL038w of the outer rings (Heinemeyer et al., 1994) was chosen to mark the proteolytically active 20S core complex of the proteasome. The ATPase-type subunit Cim5/YOL145w (Ghislain et al., 1993) was selected to monitor the subcellular distribution of the regulatory 19S cap complex. Both Pre6 and Cim5 are encoded by single essential genes (Ghislain et al., 1993; Heinemeyer et al., 1994). To allow the expression of each GFP chimera under the control of its own promoter, the chromosomal copies of PRE6 and CIM5 were disrupted 5′ to their stop codons by in-frame insertion of the GFP-coding region fused with a double HA epitope sequence. The chromosomal replacement of the wild-type genes by the corresponding GFP-HA-tagged versions was achieved by homologous recombination using HIS3 and URA3 genes as selection markers (Figure 1A). The PRE6–GFP-HA::URA3::HIS3 and CIM5–GFP-HA::URA3::HIS3 constructs were directed into the desired sites of the PRE6 and CIM5 locus upon transformation of wild-type strain WCGa, generating the yeast strains GCE6 and GAL5, respectively. Southern analysis of the chromosomal DNA of these strains confirmed that the original chromosomal copy was replaced by the respective GFP-HA-tagged version of the proteasomal subunit, including the flanking URA3 and HIS3 genes (data not shown). Figure 1.(A) Pre6 and Cim5 were tagged by an in-frame insertion of GFP-HA into the genomic PRE6 and CIM5 genes, generating C-terminal fusions. Adjacent HIS3 and URA3 genes were used as selection markers upon transformation of wild-type strain WCGa. Desired sites for homologous recombination of PRE6–GFP-HA::HIS3::URA3 and CIM5–GFP-HA::HIS3::URA3 into the original chromosomal loci, respectively, are indicated by dotted crossovers. ATG, start; TAA, stop codon. (B) Wild-type Cim5 is replaced by functionally expressed Cim5–GFP-HA in GAL5 (CIM5–GFP-HA::HIS3::URA3) and Cim5-HA in HAL5 (CIM5–HA::HIS3::URA3) cells, respectively. Total cell extracts of strain WCGa (lane 1), GAL5 (lane 2) and HAL5 (lane 3), respectively, were subjected to SDS–PAGE and analysed by immunoblotting using anti-Cim5 (left panel), anti-HA mAb12CA5 (middle panel) and anti-GFP antibodies (right panel). Download figure Download PowerPoint Since single essential genes (Ghislain et al., 1993; Heinemeyer et al., 1994) were replaced by GFP-HA-tagged versions, Pre6–GFP-HA and Cim5–GFP-HA were expected to be functionally expressed. Both strains GCE6 and GAL5 grew at rates comparable with wild-type strain WCGa at 30 and 37°C (data not shown). In addition, a strain named HAL5 was constructed which was obtained upon transformation of wild-type strain WCGa with CIM5-HA::HIS3::URA3 (see Materials and methods). The replacement of endogenous Cim5 by Cim5–GFP-HA and Cim5-HA was analysed by comparative SDS–PAGE and immunoblot analysis of cell lysates of strain GAL5 habouring CIM5–GFP-HA::HIS3::URA3, of strain HAL5 habouring CIM5–HA::HIS3::URA3 and of their parental strain WCGa by using anti-Cim5, anti-GFP and anti-HA antibodies. Cim5 was absent in total cell extracts of GAL5, but was replaced by the Cim5–GFP-HA chimera (expected molecular mass of 84 kDa) (Figure 1B, left panel, compare lanes 1 and 2). Likewise, in the HAL5 cell lysate, no band corresponding to endogenous Cim5 was detectable. Instead a protein band of slightly higher molecular mass (∼3 kDa) as expected for double HA-tagged Cim5 was visible (Figure 1B, left panel, lane 3). The C-terminal HA epitopes of Cim5–GFP-HA and Cim5-HA, respectively, were recognized by mAb12CA5, while no protein was detected in the parental wild-type strain (Figure 1B, middle panel). The GFP moiety within Cim5–GFP-HA was identified by probing the same blot with anti-GFP antibodies (Figure 1B, right panel). Taken together, the data demonstrate that wild-type Cim5 is indeed replaced by HA- and GFP-HA-tagged versions in HAL5 and GAL5 cells, respectively. GCE6 cells habouring chromosomal PRE6–GFP-HA::HIS3::URA3 (Figure 1A) were found to express a GFP-HA chimera of the expected size corresponding to Pre6–GFP-HA (59 kDa; refer to Figure 2). Figure 2.(A) Pre6–GFP-HA and Cim5–GFP-HA are incorporated into proteolytically active 26S proteasomes. Extracts of GCE6 and GAL5 cells expressing GFP-HA-tagged Pre6 and Cim5, respectively, were subjected to 10–40% glycerol gradient ultracentrifugation. Fractions were collected from top to bottom of the gradient and assayed for peptide cleavage activity using the substrate Cbz-Leu-Leu-Glu- β-naphthylamide (dotted line). All fractions were analysed for protein contents (solid line). GFP fluorescence intensity was measured (broken line). All values are given as a percentage of the maximum value. Comparable profiles were obtained for GAL5 and GCE6 cell lysates fractionated under equivalent conditions, thus GCE6 stands for GAL5. (B) Protein samples of each fraction were run on SDS–PAGE followed by immunoblot analysis using anti-GFP antibodies (GAL5 fractions, upper panel; GCE6 fractions, lower panel). (C) Pre6–GFP-HA immunoprecipitates with proteasomal 20S subunits. GCE6 cell lysates were incubated with protein A–Sepharose beads. The resin was analysed as a control for unspecific binding by SDS–PAGE followed by immunoblotting (without mAb12CA5; lane 1). The pre-cleared lysate was incubated with protein A–Sepharose beads and mAb12CA5 antibody. The immunoprecipitated proteins (IP) were probed for Pre6–GFP-HA, Scl1, Pre4 and Pre2, respectively (with mAb12CA5; lane 2). Download figure Download PowerPoint The fact that Pre6 and Cim5 were expressed in fusion with GFP was not sufficient to consider them as suitable markers for proteolytically active 26S proteasomes. Two criteria had to be fulfilled further. First, the GFP-HA-tagged versions of both subunits have to be integrated completely into their proteasomal subcomplexes, since maverick GFP-HA-labelled subunits will lead to misinterpretations concerning the localization of proteasome complexes. Secondly, both GFP-tagged subunits will only report on the location of proteasomal degradation if they are fully assembled with matured, proteolytically active core complexes. To examine whether Pre6–GFP-HA and Cim5–GFP-HA were incorporated into proteasome complexes, cell extracts of strain GCE6 and GAL5 were analysed in parallel by density gradient ultracentrifugation. All fractions were assayed for proteasomal peptide cleavage activity and GFP fluorescence intensity. Proteasomal activity was detected in the rapidly sedimenting fractions, peaking in fractions 14–15 (Figure 2A; shown for GCE6). The same profile was found for GAL5 fractions (data not shown). GFP fluorescence intensity was found to peak with proteasomal activity, providing evidence that the fluorescence signals can be traced back to proteolytically active proteasomes. In addition, the proteins of each fraction were resolved by SDS–PAGE and probed with anti-GFP antibodies. The GFP moieties of Pre6–GFP-HA and Cim5–GFP-HA were detected exclusively within those fractions that peaked with proteasomal activity and GFP fluorescence intensity (Figure 2B). No GFP-HA antigens could be detected in the more slowly migrating gradient fractions. Thus, Pre6–GFP-HA as well as Cim5–GFP-HA were mainly part of proteolytically active proteasomes rather than of precursor intermediates that are formed during the processing of several β-type subunit precursors while 20S core complexes assemble (Chen and Hochstrasser, 1996; Schmidtke et al., 1996; Seemueller et al., 1996; Heinemeyer et al., 1997). The incorporation of Pre6–GFP-HA into the mature 20S core complex was investigated further by co-immunoprecipitation experiments. For this purpose, the C-terminal HA epitope following the GFP in-frame moiety within Pre6–GFP-HA turned out to be suitable, since mAb12CA5 monoclonal antibody bound to protein A–Sepharose allowed the easy isolation of the GFP-HA chimera (Figure 2C). The immunoprecipitated proteins were separated by SDS–PAGE, blotted and probed for the GFP-HA chimera and interacting 20S subunits by using available antibodies (Figure 2C, lane 1, control without mAb12CA5; lane 2, mAb12CA5 immunoprecipitate). Pre6–GFP-HA (α4_sc; nomenclature according to Groll et al., 1997) was found to be associated with the α-type subunit Scl1 (α1_sc; molecular mass of 29 kDa) and the matured forms of the β-type subunits Pre2 (β5_sc; 23 kDa) and Pre4 (β7_sc; 26 kDa) (Heinemeyer et al., 1997), respectively. These data substantiated our conclusions from glycerol gradient fractionations that suggested that Pre6–GFP-HA is integrated into the active 20S core complex. Meanwhile, crystallographic structural analysis of 20S proteasomes predicted that the C-terminal region of Pre6 protrudes to the protein surface (Groll et al., 1997), explaining the accessibility of the HA tag for the mAb12CA5 antibody. Immunoprecipitation experiments using GAL5 cell lysates showed the interaction of Cim5–GFP-HA with proteasomal 20S subunits (data not shown) as expected for a 26S subunit (Ghislain et al., 1993). Thus, we concluded that the entire population of 26S proteasomes is confined by both Cim5–GFP-HA and Pre6–GFP-HA fusions. Yeast 26S proteasomes are located mainly in the NE–ER network The subcellular distribution of GFP-tagged 20S and 19S proteasome subcomplexes was monitored in living cells of strain GCE6 and GAL5, respectively (Figure 3). The cells were visualized by Nomarski optics; the 4′,6-diamidino-2-phenylindole (DAPI)-stained nuclear DNA was highlighted by simultaneous UV excitation (right panels). Pre6–GFP-HA of the proteolytically active 20S core complex and Cim5–GFP-HA of the regulatory 19S complex of the 26S proteasome were found to be accumulated predominantly in the nuclear periphery (left panels). The coincident subcellular distribution of both proteasomal subunits indicated that 20S and 19S subcomplexes are associated with each other in vivo, thus demonstrating that 26S holocomplexes are located mainly in the nuclear periphery. An estimation of the GFP signals over the cross-section of a single yeast cell revealed that ∼80% of 26S proteasomes are structurally bound to the nuclear periphery, while 20% of these complexes exist in the cyto- and nucleoplasmic compartments. Figure 3.In vivo localization of GFP-tagged proteasomes in yeast. Cells of strain GCE6 and GAL5 expressing GFP-tagged versions of Pre6 and Cim5, respectively, were monitored by direct fluorescence microscopy (20S Pre6–GFP-HA, upper panels; 19S Cim5–GFP-HA, lower panels). The fluorescent proteasome subcomplexes were detected by using the FITC channel (GFP; left panels). Cells were viewed by Nomarski optics, and the DAPI-stained nuclei were visualized by simultanously opening the UV channel (DIC/DAPI; right panels). Arrows indicate mitotic cells. Bar, 3 μm. Download figure Download PowerPoint The pattern of GFP-labelled proteasomes strongly resembled those patterns that were reported for GFP-labelled nuclear pore ER membrane proteins (Belgareh and Doye, 1997; Bucci and Wente, 1997). These observations gave rise to the question of whether proteasomes co-localize with marker proteins of the NE–ER network such as Kar2/Bip (Rose et al., 1989). For indirect immunolocalization studies, HAL5 cells expressing the HA-tagged version of Cim5 instead of the wild-type subunit were preferred to GCE6 or GAL5 cells, because the intrinsic fluorescence of GFP-labelled proteasomes was drastically reduced during cell fixation. HAL5 cells were processed and probed for Cim5-HA (Figure 4A, upper left panel; visualized by Cy3-conjugated donkey anti-mouse IgG) and Kar2 [Figure 4A, lower left panel; visualized by fluorescein isothiocyanate (FITC)-labelled goat Fab fragments anti-rabbit IgG]. Using confocal laser scanning microscopy, an overlapping signal for the proteasomal 26S subunit and Kar2 was monitored (Figure 4A, lower right panel). In particular, the dotted pattern along the nuclear rim with few peripheral extensions, as already observed for the GFP-HA-tagged variants in living cells (Figure 3, left panels), could be resolved. Since the rough ER (RER) is scarcely distinguishable from the NE (Strambio-de-Castillia et al., 1995), we also performed a double immunolocalization of proteasomes and nuclear pore membrane proteins. Diploid wild-type cells of strain WCGa/α were probed with anti-20S proteasome antibodies (Figure 4B, upper panel; visualized by FITC-labelled goat Fab fragments anti-rabbit IgG) and mAb414 nuclear pore proteins (Figure 4B, middle panel; visualized by Cy3-conjugated donkey anti-mouse IgG) and stained with DAPI (Figure 4B, lower panel). Again, a coincident localization at the nuclear rim was observed, supporting our interpretation that proteasomes are accumulated in the NE–RER. Figure 4.(A) Co-localization of Cim5-HA and Kar2/Bip, a reporter protein of the NE–ER network, by confocal laser scanning microscopy. HAL5 cells were processed for indirect immunofluorescence and double-labelled with mAb12CA5 and anti-Kar2 antibodies followed by incubation with Cy3-labelled donkey anti-mouse IgG (Cim5-HA, upper left panel) and FITC-labelled goat Fab fragments anti-rabbit IgG (Kar2, lower left panel). Nuclear DNA was stained with DAPI (upper right panel). All three images were merged, revealing a co-localization within dot-like structures around the nuclear DNA (lower right panel; Bar, 2 μm). (B) Double-labelling of proteasomes and mAb414 nuclear pore proteins in diploid WCGa/α cells. Fixed cells were probed with both anti-20S proteasome antibodies (Seeger et al., 1996) followed by incubation with FITC-labelled goat Fab fragments anti-rabbit IgG (upper panel) and mAb414 antibodies (Rout and Blobel, 1993), and then by Cy3-labelled donkey anti-mouse IgG (middle panel). The DNA was stained with DAPI (lower panel). Download figure Download PowerPoint Our observation that proteasomes remained accumulated in the NE–ER in mitotic cells (marked by arrows in Figure 3) suggested that the majority of proteasomes persist at the nuclear rim throughout the cell cycle. Using α-factor and nocodazole, GCE6 cells were arrested at the G1–S transition and released into M phase, respectively (Seufert et al., 1995). As already shown for non-synchronized cells (Figure 3), proteasomes remained accumulated in the NE–ER independently of the cell cycle stage (data not shown). Furthermore, we were interested in whether the subcellular distribution of GFP-tagged proteasomes is disturbed in proteasomal mutant strains cim3 and cim5 that are defective in cell cycle progression and arrest during mitosis (Ghislain et al., 1993). Both strains were transformed with the PRE6–GFP-HA::HIS3::URA3 construct, resulting in strains expressing Pre6–GFP-HA instead of wild-type Pre6 in a cim3 and cim5 mutant background, respectively. Even at the non-permissive temperature (37°C), no aberrant localization of GFP-labelled proteasomes could be observed (data not shown). This suggested that anaphase promotion in cim3 and cim5 mutants is not delayed due to a mislocalization of proteasomes that might impair the turnover of mitotic cyclins required for proper cell cycle progression. Proteins that are bound peripherally or integrally to the contiguous network of inner–outer NE–ER membranes can be distinguished further by using a membrane fusion assay established by Latterich and Schekman (1994). Their movements together with NE–ER membranes can be followed during zygote formation when the NE–ER bilayer of two conjugating cells joins and their contents intermix as a consequence of nuclear fusion (Latterich and Schekman, 1994; Belgareh and Doye, 1997; Bucci and Wente, 1997). If proteasomes are anchored to NE–ER membranes, they ought to be retained in these structures throughout the mating process. Therefore, GCE6 (MATa) and GCE6 (MATα) cells, both expressing GFP-tagged proteasomes, were permitted to mate. Upon nuclear fusion, the majority of 26S proteasomes were observed to intersperse the intermixing NE–ER network of the premature diploid cell (Figure 5A). Structurally bound proteasomes finally were inherited by the budding-off daughter cell from the diploid mother cell, obviously by membrane fissions (Figure 5B). Figure 5.GFP-tagged proteasomes move along the NE–ER upon zygote formation. Diploid cells were formed by crossing haploid cells of strain GCE6a and strain GCE6α, both expressing fluorescent proteasomes. Cells were viewed by Nomarski optics (DIC, left panels) and the FITC channel (GFP, right panels). (A) While the NE–ER networks of the conjugating cells were joining, proteasomes were spread over the intermixing membranes. (B) The structurally bound proteasomes were inherited by the budding-off daughter cell from the new diploid mother cell through NE–ER membrane fissions. Download figure Download PowerPoint Biochemical fractionation reveals co-enrichment of 26S proteasomes and NE–ER membrane proteins Based on our localization studies, we expected 26S proteasomes to be enriched with NE–ER markers in biochemical fractionation experiments. To isolate NE–ER membranes from GCE6 cells, we applied a procedure based on the preparation of highly enriched yeast nuclei (Rout and Blobel, 1993; Strambio-de-Castillia et al., 1995). By lysing these nuclei, NE–ER membranes were released and isolated by flotation in an equilibrium sedimentation gradient. Proteins of the enriched NE–ER fraction of the nuclear lysate and of total spheroplasts were resolved by SDS–PAGE and analysd by immunoblotting. As expected (Strambio-de-Castillia et al., 1995), nucleoporins and the ER lumenal marker Kar2 were co-enriched within the NE–ER membrane fraction (Figure 6A, compare lane 1 with lanes 2 and 3). Nucleolar Nop1 was almost undetectable in the NE–ER fraction, indicating only a small amount of nucleoplasmic remnants within this fraction. Compared with total cell lysates, proteasomal 26S subunits were found to be enriched with NE–ER membrane markers as analysed for Pre6–GFP-HA, Cim5 and Scl1, respectively. Figure 6.Yeast 26S proteasomes are enriched with NE–ER membrane fractions. (A) NE–ER membranes were prepared according to Strambio-de-Castillia et al. (1995). Fractions were analysed by SDS–PAGE followed by Coomassie Blue staining (upper panel) and immunoblotting (lower panel). Equivalent amounts of protein of total spheroplast lysates (T, lane 1), highly enriched nuclear lysates (N, lane 2) and NE–ER fractions (NE, lane 3) were resolved and probed for mAb414 nuclear pore membrane proteins, nucleolar Nop1 (Rout and Blobel, 1993), ER lumenal Kar2/Bip (Rose et al., 1989), the proteasomal subunits Pre6–GFP-HA, Cim5 and Scl1, and phosphofructokinase Pfk1/Pfk2, respectively. (B) The relative amounts of proteasomes either existing in the soluble fraction or bound to membrane structures were estimated. Total cell lysates were obtained by sonication of GCE6 spheroplasts. Membranes were pelleted by ultracentrifugation, separated from the high speed supernatant and resuspended. Cell equivalents of total lysate (T, lane 1), high speed supernatant (S, lane 2) and membrane pellet (P, lane 3) were analysed by SDS–PAGE and immunoblotting using anti-Kar2, GFP and Scl1 antibodies. Each fraction was assayed for relative GFP fluorescence intensity. The mean values given were derived from measurements of three parallel preparations. Download figure Download PowerPoint To estimate the relative amounts of structurally bound proteasomes versus soluble proteasomes, spheroplasts of strain GCE6 were lysed by sonication, yielding >95% cell disintegration. The membranes were pelleted by ultracentrifugation and separated from the high speed supernatant. Cell equivalents of total lysate (Figure 6B, lane 1), high speed supernatant (Figure 6B, lane 2) and membrane resuspension (Figure 6B, lane 3) were probed for Kar2, Pre6–GFP-HA and Scl1, respectively. The lumenal ER marker was almost completely removed from the high speed supernatant, indicating only a small amount of microsomal remnants in the soluble fraction. The proteasomal subunits Pre6–GFP-HA and Scl1 were found to be highly enriched within the membrane resuspension compared with the soluble fraction. To quantify the GFP signals, the fluorescence intensity of each fraction was measured. In the membrane resuspension, we determined 80 ± 5% of the total GFP labels, which is in good agreement with our estimations of structurally bound and soluble proteasomes in intact cells (Figure 3). Proteasomal peptide cleavage occurs mainly in the NE–ER network in yeast To visualize proteasomal activity in the NE–ER network, we employed an in situ test that allowed the selective screening for proteasomal mutants in previous work by taking advantage of the peptide cleavage ability of proteasomes in permeabilized yeast cells (Enenkel et al., 1994). GCE6 cells were transferred into spheroplasts and embedded into a medium containing the proteasome-specific peptide substrate Cbz-Leu-Leu-Glu-β-naphthylam