Recognition of specific substrates for degradation by the ubiquitin–proteasome pathway is ensured by a cascade of ubiquitin transferases E1, E2 and E3. The mechanism by which the target proteins are transported to the proteasome is not clear, but two yeast E3s and one mammalian E3 ligase seem to be involved in the delivery of targets to the proteasome, by escorting them and by binding to the 19 S regulatory particle of the proteasome. In the present study, we show that SNEV (senescence evasion factor), a protein with in vitro E3 ligase activity, which is also involved in DNA repair and splicing, associates with the proteasome by directly binding to the β7 subunit of the 20 S proteasome. Upon inhibition of proteasome activity, SNEV does not accumulate within the cells although its co-localization with the proteasome increases significantly. Since immunofluorescence microscopy also shows increased co-localization of SNEV with ubiquitin after proteasome inhibition, without SNEV being ubiquitinated by itself, we suggest that SNEV shows E3 ligase activity not only in vitro but also in vivo and escorts its substrate to the proteasome. Since the yeast homologue of SNEV, Prp19, also interacts with the yeast β7 subunit of the proteasome, this mechanism seems to be conserved during evolution. Therefore these results support the hypothesis that E3 ligases might generally be involved in substrate transport to the proteasome. Additionally, our results provide the first evidence for a physical link between components of the ubiquitin–proteasome system and the spliceosome.
- 20 S proteasome
- senescence evasion factor (SNEV)
The proteasome is a large multisubunit protease which is ubiquitous to life and plays a crucial role in intracellular protein degradation . Proteasomes of eukaryotes have a complex structure that has been elucidated at the molecular level recently . They comprise a barrel-shaped catalytic 20 S proteasome core complex, capped at both poles by a 19 S regulatory ATPase complex. This association with the 19 S cap complex  forms the 26 S proteasome, which controls substrate accession to the proteases and enhances the proteolytic activity of the core in an energy-dependent manner.
The 26 S proteasome is a dynamic structure having multiple interactions with transiently associated subunits and cellular factors that are necessary for functions such as cellular localization, presentation of substrates, substrate-specific interactions and generation of various products . Degradation of proteins by the proteasome in the cell depends on their recognition, labelling and transport to it.
A multistep process leads to the transfer of ubiquitin or multiubiquitin chains on to a substrate. This process includes activation of ubiquitin by an E1 enzyme and transfer of this ubiquitin to an E2 ubiquitin-conjugating enzyme that in turn transfers the ubiquitin moiety to an E3 ligase or in concert with an E3 ligase to the substrate. The lysine residue by which the ubiquitin chains are linked together is an important signal that determines the fate of the substrate; Lys48 linkage for example is the major signal that targets proteins for destruction to the proteasome .
We have identified SNEV (senescence evasion factor; hNMP200, Prp19-like, hPso4; NCBI accession no. NP_055317; unknown gene 4 ) in a screening for differentially expressed genes in early passage and replicatively senescent human endothelial cells. SNEV is also involved in the regulation of cellular life span (R. Voglauer, W. M. F. Chang, M. Wieser, K. Baumann, H. Katinger and J. Grillari, unpublished work). Additionally, it is a part of the spliceosome-associated complex , localized in the nucleus  and involved in DNA double-strand break repair .
SNEV displays E3 ubiquitin ligase activity in vitro . This activity is dependent on its U-box domain and on the E2 enzyme UbcH3. Since multiubiquitin chains that are linked by Lys48 are formed in this in vitro assay, it is suggested that SNEV confers the ‘classical’ signal for protein degradation by the proteasome (reviewed in ).
Since yeast E3 enzymes have been reported to transport their substrates to, and to interact with, the proteasome [10,11], we were interested to know whether this is a general and conserved mechanism valid also in mammalian cells. One indication that this might be true for SNEV, is the report that the β7 subunit of the Caenorhabditis elegans proteasome pulled out the homologue of SNEV in a Y2H (yeast two-hybrid) high-throughput screening ( electronic supplement).
In the present study, we show that SNEV indeed interacts with the proteasome. The interaction is direct and evolutionarily conserved from yeast to human and the interacting partner of SNEV is the β7 subunit (PSMB4) of the 20 S proteasome. Although the 19 S cap has been known to bind components of the ubiquitin system, this is the first report that the 20 S core has the same ability. These findings therefore support the idea that interaction of E3 ligase with the proteasome might be a general mechanism to either target the proteasome to the sites of protein degradation or to target and escort the ubiquitinated proteins to the sites of destruction.
MATERIALS AND METHODS
Sequence comparison and structure of β7 20 S subunit
Sequence comparisons of Homo sapiens SNEV (NCBI accession no. NP_055317) and Saccharomyces cerevisiae Prp19 (NCBI accession no. NP_013064), SNEV and C. elegans T10F2.4 (NCBI accession no. AAK21467), H. sapiens PSMB4 (NCBI accession no. NP_002787) and S. cerevisiae Pre4 (NCBI accession no. NP_116708) and PSMB4 and C. elegans pbs-7 (NCBI accession no. NP_492354) were performed using the GenBank® database . Additionally, species sequence comparison of human, yeast and worm homologues was performed using LaserGene, 4.0 (DNASTAR, Madison, WI, U.S.A.). The structure of the bovine β7 20 S subunit (PDB accession no. 1IRU), integrated within the proteasome, was modelled using Swiss-PdbViewer 3.7  and the structure of yeast Pre4 (PDB accession no. 1FNT), integrated within the proteasome, was modelled using RasMol Version 2.6-beta-2 .
Construction of plasmids
General methods like PCR, transformation of Escherichia coli cells, restriction reactions, DNA ligations and other recombinant DNA techniques were performed following standard procedures . In brief, the coding sequences of SNEV and PSMB4 were amplified by RT–PCR. Total RNA was prepared using Trizol® reagent (Invitrogen, Carlsbad, CA, U.S.A.). Total RNA (1 μg) from HUVEC (human umbilical-vein endothelial cells) was reverse-transcribed using oligo(dT)25 primers, and standard PCR with the primers described in Table 1 was performed. Prp19 and Pre4 were directly amplified by PCR using genomic DNA from the yeast strain W303 as a template. Primer sequences are shown in Table 1. The inserts were ligated into pGADT7 and pGBKT7 for Y2H analysis, into pEYFP-C1, pEYFP-N1, pECFP-C1 and pECFP-N1 for FRET (fluorescence resonance energy transfer) analysis (ClonTech Laboratories, Palo Alto, CA, U.S.A.) and into pGEX-6P-1 (Amersham Biosciences, Uppsala, Sweden) for GST (glutathione S-transferase) pull-down assays. All plasmid constructs were amplified in E. coli and the inserts were confirmed to contain no mutations by sequence analysis. The His6–SNEV-containing plasmid for baculoviral expression was kindly provided by S. Hatakeyama.
Screening for SNEV-interacting proteins was performed using the MATCHMAKER GAL4 Two-Hybrid System3 (ClonTech Laboratories) according to the manufacturer's guidelines. Therefore the cDNAs of SNEV and SNEV deletion constructs: SNEV(ΔWD40) amino acids 1–205, SNEV(WD40) amino acids 206–504, SNEV(ΔU67) amino acids 67–504, SNEV(ΔU90) amino acids 91–504 and SNEV(U96) as well as Pre4 cDNAs were inserted into the bait vector pGBKT7 (ClonTech Laboratories) in frame with the GAL4 DNA BD (binding domain) and a c-Myc epitope. PSMB4 and its deletion mutants: PSMB4(PP) propeptide amino acids 1–45, PSMB4(P122) amino acids 1–122, PSMB4(PΔ102) amino acids 102–264, PSMB4(P101) amino acids 1–101, PSMB4(PΔ122) amino acids 123–264 as well as Prp19 were inserted into the vector pGADT7 (ClonTech Laboratories) providing the GAL4 AD (activation domain) and the HA (haemagglutinin) tag. All constructs were amplified in E. coli and sequence analysed to confirm correct proteins. Then the constructs were introduced into S. cerevisiae strain AH109 (ClonTech Laboratories) by LiAc-co-transformation . We selected interaction-positive clones by growth on high stringency SD (synthetic dropout) medium (4×SD: −Trp/−Leu/−His/−Ade; four times the SD medium lacking the amino acids tryptophan, leucine, histidine and the nucleotide adenine).
In vitro CoIP (co-immunoprecipitation)
Seize primary immunoprecipitation kit (Pierce, Rockford, IL, U.S.A.) was used to couple the c-Myc-tag antibody (ClonTech Laboratories) to agarose beads according to the manufacturer's guidelines. In vitro translation of SNEV, PSMB4 and Pre4 was performed using the pGADT7- and pGBKT7-derived plasmids as templates, and thus c-Myc-/HA-tagged proteins were synthesized using TNT Quick Coupled Transcription/Translation System (Promega, Madison, WI, U.S.A.). SNEV and PSMB4 proteins were labelled with 35S-methionine (Amersham Biosciences). After mixing the in vitro translates and incubating at 37 °C for 1 h, precipitation was performed by addition of the c-Myc-tag antibody-coupled beads and incubation at room temperature (22 °C) for 3 h in binding buffer [20 mM Tris/HCl (pH 7.4), 140 mM NaCl, 10% (v/v) glycerol, 1 mM CaCl2, 0.1% Triton X-100; and one tablet of Complete protease inhibitor cocktail (Roche, Basel, Switzerland) to 50 ml of buffer]. The beads were washed four times in binding buffer and elution was achieved by heating for 10 min in SDS sample buffer. After SDS/PAGE analysis, according to standard procedures , the gel was fixed, dried and proteins were detected by autoradiography.
For CoIP of Prp19–His6 and Myc–Pre4 (in vitro translated), Prp19–His6 cloned into pYPGE15 vector and transformed into yeast strain MG5128 (J. Grillari, G. Stadler and M. Grey, unpublished work) was grown overnight in 10 ml of SD/−Ura medium. The yeast cells were harvested by centrifugation and washed three times in breaking buffer (50 mM sodium phosphate, pH7.4, 1 mM PMSF, 1 mM EDTA and 5% glycerol). After another centrifugation, the cells were resuspended in SET buffer (1 M Sorbit, 10 mM Tris/HCl, pH 8.0, and 50 mM EDTA), mixed on a vortex and incubated for 30 min at 37 °C. The pellet was again resuspended in breaking buffer containing 1% NaN3, 1% EDTA and 1% PMSF. Lysis was performed by an addition of glass beads and six cycles of 30 s mixing on a vortex and 30 s of incubation on ice.
CoIP was performed by incubation of the Prp19–His6-containing yeast cell lysate and in vitro translated c-Myc–Pre4 protein for 1 h at 4 °C, followed by an incubation with Ni2+-NTA (Ni2+-nitrilotriacetate) agarose beads (Qiagen, Hilden, Germany) for 1.5 h at 4 °C. The beads were pre-equilibrated with TBST buffer (150 mM NaCl, 20 mM Tris/HCl, pH 7.5, and 0.1%, v/v, Tween 20). Then the beads were washed five times with TBST buffer and elution was performed by heating for 10 min with SDS/PAGE sample buffer. After SDS/PAGE, precipitated Pre4 was detected using anti-c-Myc antibody (ClonTech Laboratories) and anti-mouse peroxidase conjugate as secondary antibody (Sigma–Aldrich, St. Louis, MO, U.S.A.).
GST pull-down assay
Mature PSMB4 (without propeptide) was cloned into pGEX-6P-1 as a GST-fusion, then transformed into E. coli BL21 and expressed using Overnight Express Autoinduction System 1 (Novagen, Darmstadt, Germany). Purification was performed using the Micro Spin GST Purification Module (Amersham Biosciences) according to the manufacturer's instructions. His6–SNEV was expressed in Sf9 insect cells and affinity-purified on a Ni2+-NTA column (E. Böhm, J. Grillari, S. Gross, W. Ernst, B. Ferko, N. Borth and H. Katinger, unpublished work). For the pull-down assay, equal amounts of the purified proteins (0.5 μg) were mixed together and incubated for 2 h at 4 °C in CoIP buffer (200 mM NaCl, 25 mM Tris/HCl, pH 7.4, and 0.5% Triton X-100). After that, glutathione–Sepharose 4B beads (Amersham Biosciences), equilibrated with CoIP buffer, were added, followed by incubation for 2 h at 4 °C. Then the beads were washed three times with CoIP buffer and proteins were eluted with SDS/PAGE sample buffer. After SDS/PAGE, precipitated SNEV was detected using anti-His4 antibody (Qiagen). GST was detected using anti-GST antibody (Amersham Biosciences) and anti-goat peroxidase conjugate as secondary antibody (Sigma–Aldrich).
HeLa cells were grown in DMEM (Dulbecco's modified Eagle's medium) supplemented with 4 mM L-glutamine and 10% (v/v) foetal calf serum. HEK-293 (human embryonic kidney 293) cell line was grown in DMEM/Ham's medium supplemented with 4 mM L-glutamine and 10% foetal calf serum. Proteasome activity of HeLa cells was blocked using 25 μM MG132 (Calbiochem, Darmstadt, Germany) in the medium.
Cell lysis for Western blots
Cells were lysed in 50 ml of nuclear lysis buffer containing 50 mM Tris (pH 7.5), 0.5 M NaCl, 1% Nonidet P40, 1% deoxycorticosterone, 0.1% SDS, 2 mM EDTA and 1 Complete protease inhibitor tablet (Roche Diagnostics, Vienna, Austria). Proteins were detected using anti-SNEV antibody Prp19–866, anti-p53 antibody (BP53-12; Sigma–Aldrich), anti-β-actin antibody (Sigma–Aldrich) and anti-ubiquitin antibody (MBL International, Woburn, MA, U.S.A.).
In vitro 26 S proteasome degradation assay
His6–SNEV (0.3 μg) was incubated in the presence and in the absence of 0.3 μg of 26 S proteasome, purified from rat liver as described previously , for 2 h at 37 °C. Reactions were performed in 50 mM Hepes buffer (pH 7.5) containing 5 mM ATP and 5 mM MgCl2. At various time points, aliquots were added to SDS/PAGE gel sample buffer and heated at 100 °C for 5 min and then frozen. For analysis, SDS/PAGE and Western blotting were performed. IκBα (inhibitory κBα) was used as a model substrate to confirm the activity of proteasome and was detected in Western blots using an anti-IκBα antibody (sc-1643; Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.). Densitometry was performed using an imaging densitometer (Model GS-690; Bio-Rad, Hercules, CA, U.S.A.).
In vivo CoIP
For proteasome preparation, HEK-293 cells were lysed in CoIP lysis buffer (20 mM Tris, pH 7.5, 10% glycerol, 5 mM ATP and 0.2% Nonidet P40). The lysate was added to anti-α2 20 S subunit antibody-coupled Protein A–agarose beads (Roche Diagnostics, Vienna, Austria) (coupling in 1×PBS, 3 h at 4 °C) and incubated overnight at 4 °C. As control, anti-c-Myc antibody was coupled on to agarose beads and incubated with HEK-293 cell lysate as described above. The beads were washed four times with CoIP washing buffer (50 mM Tris, pH 7.5, 10% glycerol, 5 mM ATP, 150 mM NaCl and 0.2% Triton X-100) and resuspended in SDS loading buffer. SNEV and PSMB4 were detected by Western-blot analysis using anti-PSMB4 antibody (Affiniti Research, Mamhead Castle, Mamhead, Exeter, Devon, U.K.) and anti-SNEV antibody Prp19–866 with anti-rabbit peroxidase conjugate as secondary antibody (Sigma–Aldrich). Co-precipitated Cdc5L was detected by rabbit anti-Cdc5L antibody.
SNEV, SNEVΔ66, SNEVΔ89 and PSMB4 inserted into pECFP-N1, pECFP-C1 and pEYFP-N1, pEYFP-C1 respectively were transiently co-transformed into COS-1 and HEK-293 cells by Lipofectamine™ 2000 (Invitrogen) according to the manufacturer's guidelines. After 24 h, FRET images from living cells were generated by the MicroFRET method as described by Youvan et al. . Photos were captured on a Nikon Diaphot TMD microscope with a cooled charge-coupled device camera (Kappa GmbH, Gleichen, Germany), with the YFP (yellow fluorescent protein), CFP (cyan fluorescent protein) and FRET filter sets (Omega Optical, Brattleboro, VT, U.S.A.), under identical conditions and processed with Scion Image software version beta 4.0.2 (Scion, Frederick, MD, U.S.A.). The images were aligned by pixel shifting, inverted, and the background was subtracted. Images from the YFP and CFP filter sets were multiplied with their previously assessed correction factors (0.19 for YFP and 0.59 for CFP) and subtracted from the FRET filter set picture. The remaining signals were multiplied by 3 for better visualization and they represent the corrected FRET.
Cell staining and immunofluorescence analyses
HeLa cells were washed with PBS and fixed for 5 min in 3.7% (w/v) paraformaldehyde in CSK buffer (10 mM Pipes, pH 6.8, 10 mM NaCl, 300 mM sucrose, 3 mM MgCl2 and 2 mM EDTA) at room temperature. Permeabilization was performed with 1% Triton X-100 in PBS for 15 min at room temperature. Cells were incubated with primary antibodies diluted in PBS with 1% (v/v) goat serum for 1 h, washed three times with PBS for 10 min, incubated for 1 h with the appropriate secondary antibodies diluted in PBS with 1% goat serum and washed three times for 10 min with PBS. Antibodies used were rabbit anti-SNEV antibody Prp19–867, anti-ubiquitin and anti-proteasome α2 subunit (Affiniti Research). As secondary antibodies, tetramethylrhodamine β-isothiocyanate-labelled anti-mouse antibody and FITC-labelled anti-rabbit antibody (Jackson Immunoresearch Laboratories, West Grove, PA, U.S.A.) were used. Microscopy and image analysis was performed using a Zeiss DeltaVision Restoration microscope as described previously .
Interaction of SNEV and Prp19 with the β7 subunit of the 20 S proteasome is evolutionarily conserved
In an attempt to further characterize SNEV, we found that its C. elegans homologue has been observed to bind to the β7 subunit of the C. elegans proteasome in a high-throughput Y2H screening . The corresponding human and yeast genes are evolutionarily conserved (Figure 1A) as shown by sequence comparison of human SNEV with S. cerevisiae Prp19 (23% identical and 41% similar amino acids), and with C. elegans T10F2.4 (50% identical and 67% similar amino acids). Similarly, human PSMB4 is conserved in comparison with S. cerevisiae Pre4 (43% identical and 64% similar amino acids) and with C. elegans pbs-7 (33% identical and 56% similar amino acids; Figure 1B). Interestingly, although the human to worm homologue of SNEV seems evolutionarily closer than to the yeast protein, homology of human PSMB4 to the S. cerevisiae Pre4 is higher than to C. elegans pbs-7.
Because of these homologies, we were interested to know if the interaction between these proteins is also conserved. In a first attempt to test a possible interaction of SNEV and Prp19 with the β7 subunit of the proteasome (PSMB4 and Pre4), Y2H experiments were performed. The cDNAs coding for the proteins of interest were genetically fused to the GAL4 AD or to the GAL4 BD. After co-transformation of yeast strain AH109 with vectors containing Prp19 and Pre4 or containing SNEV and PSMB4, interaction of SNEV with PSMB4 and of Prp19 with Pre4 (Figure 2A, iv and i) was indicated by growth on high stringency selection medium. No growth was observed with either of the proteins and the respective second vector containing only the GAL4 AD or BD alone (Figure 2A, ii, iii, v and vi).
Since our main interest lies in the mammalian proteins, further characterization of the domains necessary for their interaction was performed only with SNEV and PSMB4. Different deletion mutants of SNEV were generated for Y2H experiments and were tested by co-transformation with PSMB4 (Figure 2B). Only constructs containing the 68 N-terminal amino acids of SNEV resulted in growth of co-transformed yeast, indicating that these amino acids are necessary for the interaction. Interestingly, this domain corresponds to the U-box of SNEV, which is also necessary for the E3 ligase activity of SNEV . However, although this domain is necessary, it is not sufficient for the interaction. The U96 mutant of SNEV containing only the amino acids 1–96 does not give rise to the formation of yeast colonies.
For PSMB4, the propeptide (1–45) as well as the amino acids from 123–264 (PΔ122) were excluded from containing the binding site of SNEV. The amino acids from 1–101 (P101) however were sufficient for binding to SNEV, suggesting that the binding site maps on to the region from amino acids 46 to 101. However, the α-helical region (α-helix 1) between amino acids 102 and 122 seems to contribute to the interaction, since the construct PΔ102 containing this site interacts with SNEV (Figure 2C), although weakly, as indicated by the decreased growth rate of the positive transformants on high stringency selection medium. However, the fact that the P122 deletion mutant seems to interact more weakly than P101 lacking the α-helix is surprising. We suggest that although this α-helix contributes to the interaction, its presence alone (lacking the C-terminal half of PSMB4) might interfere with folding or accessibility of the interaction site contained in the region 45–101. This hypothesis is supported by the fact that interaction with full-length PSMB4 (containing the C-terminal half) is similar to the interaction of P101 alone.
These results were also confirmed by co-transformation of the two PSMB4 deletion mutants, P122 and PΔ102, and the two SNEV deletion mutants, ΔWD40 and ΔU67 (Figure 2D). The SNEVΔWD40 construct showed the same weak interaction with both PSMB4 constructs as the full-length PSMB4, whereas SNEVΔU67 showed no interaction with either of the two PSMB4 deletion mutants.
To test if the putative interacting amino acids of PSMB4 as determined by Y2H domain mapping are on the outer surface of the proteasome and thus accessible for SNEV interaction, we visualized the region from amino acid 46 to the α helix 1 region within the structure of the proteasome. Therefore we used the models of the yeast and bovine proteasomes . Of these, the bovine structure is the most similar (β7 subunit: 99% conservation to PSMB4) to the human proteasome. Structure of the mature yeast proteasome has been modelled by RasMol 2.6-beta-2 (Figure 2E) whereas that of the mature bovine proteasome was modelled by Swiss-PdbViewer 3.7 (Figure 2F). The surface-exposed amino acids of the region from amino acid 46 to α-helix 1 of the β7 subunit integrated into the proteasome are coloured and show that interaction of either SNEV or Prp19 with the mature proteasomes is structurally possible and is consistent with our co-precipitation of SNEV with the proteasome. Domains of the bovine β7 subunit were coloured like the domains of PSMB4, indicated in Figure 2(C).
Interaction of in vitro translated SNEV and PSMB4
To exclude the possibility that the interactions observed by the Y2H system are false positives, we verified our results by CoIP experiments.
For in vitro CoIP of SNEV and PSMB4, the proteins were in vitro translated as radiolabelled fusions to the Myc and HA tag, using rabbit reticulocyte lysates. After incubation with anti-Myc antibody-coupled beads and three washing steps, SDS/PAGE was performed, and radiolabelled proteins were detected by autoradiography (Figure 3A). HA-tagged PSMB4 was co-precipitated with c-Myc-tagged SNEV on c-Myc-coupled beads, while no PSMB4 was detected in the control, where PSMB4 was incubated with c-Myc beads without SNEV.
Interaction of the yeast partners was similarly confirmed using in vitro translated Myc–Pre4 and Prp19–His6, which was ectopically expressed in S. cerevisiae. Precipitation of Prp19–His6 on Ni2+-NTA beads resulted in co-precipitation of Myc–Pre4 as detected by anti-c-Myc antibody. No signal was detected when Pre4 was incubated with Ni2+-NTA beads in the absence of Prp19–His6 (Figure 3B). Therefore we conclude that the interaction between the homologous proteins of SNEV and Prp19 with the respective β7 proteasome subunits is evolutionarily conserved.
Interaction of SNEV and PSMB4 is direct
CoIP using rabbit reticulocyte lysate does not exclude the possibility that other factors could mediate the observed interaction as bridging factors. Therefore we performed pull-down assays using bacterially expressed PSMB4 as GST fusion protein, together with Ni2+-NTA affinity-purified His6–SNEV produced in insect cells. Only a single band was visible in silver-stained gels of the purified His6–SNEV (results not shown). Incubation of these purified proteins with GST–Sepharose beads resulted in the precipitation of His6–SNEV, whereas neither incubation with GST on beads nor with glutathione-beads alone was able to pull down SNEV. Precipitated SNEV was detected using anti-His4 antibody and GST was detected using anti-GST antibody (Figure 4).
SNEV interacts with the proteasome in vivo, but is not degraded by the proteasome
To confirm the in vivo relevance of the interaction of SNEV with the proteasome, we tested if endogenous SNEV is detectable in proteasome isolations. Therefore we precipitated the proteasomes from HEK-293 cell lysates using anti-α2 subunit antibodies coupled with Protein A beads. Indeed, SNEV was detected in the precipitate, in contrast with precipitates using Protein A beads charged with the irrelevant anti-c-Myc antibody (Figure 5A). In these precipitations, only a portion of SNEV seems proteasome-associated, as can be observed when comparing SNEV amounts contained in the precipitated proteasome with those contained in the total cell lysates.
The interaction with the proteasome might indicate that SNEV is degraded by this major proteolytic machinery. Therefore we treated HeLa cells for 6 h with the proteasome inhibitor MG132 and analysed the amount of SNEV protein on Western blots (Figure 5B). Blocking of proteasome activity was monitored by accumulation of ubiquitinated protein species in the cell lysates, as well as by accumulation of the proteasome substrate p53. In contrast, no significant increase in the protein levels of SNEV and β-actin (as protein loading control) was visible. In line with this result, no additional higher molecular mass band or smear that could represent accumulation of mono- or multi-ubiquitinated SNEV was detected using anti-SNEV antibodies.
Additional evidence that SNEV might not be a substrate to the proteasome is derived from in vitro experiments. When active purified proteasome was added to recombinant SNEV, no difference in SNEV degradation with or without 26 S proteasome was observed on Western blots, which were quantified by densitometry (Figure 5C). Activity of the proteasome preparation was controlled by using IκBα as substrate, which was readily degraded. These results suggest that SNEV is, at least under the conditions tested, not a substrate for proteasomal degradation. This is consistent with the pulse–chase labelling experiment by Gotzmann et al. , who have shown that SNEV amounts do not decrease in Jurkat cells during 24 h of observation.
Since SNEV is reported to be contained in a multiprotein complex, we also tested another protein in this complex, Cdc5L , in our proteasome precipitations (Figure 5A). Indeed, we were able to detect Cdc5L, indicating that at least one other protein of the Cdc5L-associated complex interacts with the proteasome. It should be noted, however, that no proteasomal subunits were found in a screening for proteins that co-precipitate together with the Cdc5L-associated complex .
As an additional confirmation and to localize the interaction of SNEV and PSMB4 within the cells, we inserted the cDNAs into pECFP and pEYFP vectors (Figure 6). After co-transformation of two different cell lines (COS-1 and HEK-293), interaction between these proteins was observed by MicroFRET analysis  in living cells. The interaction was primarily localized within the cytoplasm, but a fainter FRET signal was also detected in the nucleus. As negative controls, we used Δ66-SNEV, which lacks the U-box that is necessary for the interaction in the Y2H experiments, as well as Δ89-SNEV. On co-transfection with PSMB4, no FRET signal was observed (Figures 6P and 6R), similar to the co-transfection of SNEV–ECFP (enhanced CFP) and unfused EYFP (enhanced YFP; results not shown). As positive control, an EYFP–ECFP fusion protein was used (Figures 6M and 6O).
These results, however, have several limitations. An ECFP–SNEV fusion protein was found in the cytoplasm and nucleus, whereas the endogenous protein is mainly found in the nucleus (J. Grillari and R. Voglauer, unpublished work and ). Additionally, PSMB4–EYFP fusion proteins did not integrate into mature proteasomes. We were not able to detect the fusion protein in proteasome precipitations from HEK-293 cells transiently transfected with pEYFP–PSMB4 (results not shown), a result consistent with the observation of Lin et al.  who showed that C-terminally FLAG-tagged PSMB4 is also not integrated into the proteasome.
Finally, the cellular localization of SNEV was altered either by fusion to ECFP, by overexpression or by the interaction with free PSMB4–EYFP that might interfere with the import of SNEV into the nucleus. Therefore although our FRET analysis confirms an interaction of the recombinant proteins in living cells, it seems to have no relevance for the real localization.
Co-localization of SNEV with the proteasome and ubiquitin
Since the FRET experiment did not reveal in which cellular compartment the interaction between SNEV and the proteasome is likely to occur, we performed indirect immunofluorescence experiments using antibodies against the endogenous proteins in HeLa cells for co-localization. Since we reasoned that the E3 ligase SNEV might transport the target of its ubiquitination activity to the proteasome and after ‘delivery’ might quickly dissociate again, we included HeLa cells treated with the proteasome inhibitor MG132 in this analysis. We observed rare co-localization of SNEV with the proteasome without MG132 (Figures 7A–7C), consistent with a low amount of SNEV in proteasome precipitates, but a marked increase in co-localizing speckle-like structures with proteasome inhibition (Figures 7D–7F). When we tested SNEV and ubiquitin for co-localization, the staining of ubiquitin was very weak in untreated HeLa cells (results not shown), but was clearly co-localizing with SNEV in MG132 treated cells in similar speckle-like structures as SNEV and the proteasome (Figures 7G–7J).
These results, in combination with our result that neither the total amount of SNEV nor that of ubiquitinated SNEV increases on proteasome inhibition, suggests that SNEV might display E3 ligase activity not only in vitro but also in vivo and that it might transport its ubiquitinated substrate to the proteasome, presumably for degradation.
By a subtractive hybridization method, various differentially expressed genes were identified from early passage and senescent HUVECs . One of these genes, the unknown gene 4, now termed SNEV, was selected for further characterization. The mRNA encoding the putative protein SNEV is present in all human tissues tested (kidney, lung, placenta, small intestine, liver, peripheral blood leucocytes, spleen, thymus, colon, skeletal muscle, heart and brain), and no alternative splicing products have been identified (J. Grillari, unpublished work; ).
SNEV is a multifaceted protein that is involved in DNA double-strand break repair , in splicing  and life span extension (R. Voglauer, W. M. F. Chang, M. Wieser, K. Baumann, H. Katinger and J. Grillari, unpublished work).
Within the first 68 amino acids, a U-box [a modified RING (really interesting new gene) finger] domain was identified by SMART search. This conserved domain is found in a large variety of proteins that show E3 ligase activity . Indeed, Hatakeyama et al.  have shown that SNEV has E3 ligase activity in vitro, when tested with Ubc3 (cdc34) as E2 enzyme, and that the ubiquitin transfer is dependent on the U-box . Additionally, SNEV has been shown to mediate multiubiquitination through the Lys48 residue , which is known to target substrates to the proteasome [1,4].
Different E3 ligases have been reported to physically interact with the proteasome, like the yeast proteins Ubr1p, a RING-H2 domain containing protein, and Ufd4p, a HECT (homology to E6-AP C-terminal)-domain E3 ligase . Additional evidence for an interaction between ubiquitin-conjugating enzymes and the 26 S proteasome in yeast is derived from the report that various E3 ligases also interact with the proteasome . In mammalian cells, however, only the von Hippel-Lindau protein, a component of the von Hippel-Lindau-Elongin B/C-cullin 2 (VEC) E3 ligase complex, has been shown to interact with the proteasome . All of these proteins bind to components of the 19 S lid of the proteasome.
To verify if an interaction of our E3 ligase SNEV with the proteasome can be observed, we looked for potential proteasomal interactors. Thereby, we found a putative candidate that was identified in a high-throughput Y2H study. Putative interaction between T10F2.4, the C. elegans homologue of SNEV, and F39H11.5, the β7 subunit of the proteasome, was reported (, electronic supplement).
In the present study, we have shown that SNEV indeed interacts directly with PSMB4 in vitro by different independent methods and that this interaction is conserved from yeast to mammalian cells. Additionally, proteasome precipitates contain endogenous SNEV protein, indicating that a portion of SNEV is associated with the proteasome in vivo and strongly suggesting that this association is dependent on binding to PSMB4.
Results of the present study are supported by indirect immunofluorescence experiments, where SNEV partially co-localizes with the proteasome. When the cells are treated with the proteasome inhibitor MG132, this co-localization increases significantly. Co-localizations of SNEV with the proteasome or with ubiquitin that we have observed occur in nuclear speckle-like structures. Similar structures, termed clastosomes, have already been described to contain ubiquitin conjugates, protein substrates and proteasomes .
These findings suggest that either SNEV itself is a substrate of proteasomal degradation, or that SNEV as an E3 ligase escorts its substrate(s) to the proteasome and accumulates there, since ‘delivery’ and in turn dissociation fails due to proteasome inhibition. The latter hypothesis is supported by the fact that SNEV does not accumulate within the cells on inhibition of the proteasome and that no additional bands indicating SNEV ubiquitination are detected on Western blots. Additionally, SNEV is not degraded by the 26 S proteasome, when tested in vitro. However, ubiquitin does co-localize with SNEV on proteasome inhibition, suggesting that SNEV might have E3 ligase activity not only in vitro , but also in vivo.
There are several reports available on the function of homologuous yeast and human subunits of the 20 S proteasome Pre4 and PSMB4. For the yeast subunit Pre4, these results are mainly derived from genetic mutant strains. Pre4 is located within the inner ring of the 20 S proteasome and is a crucial component in assembling the proteasome . Two yeast mutant strains with defects in Pre4 are presently known. Pre4-1 mutants show a defect in peptidyl glutamyl peptide hydrolase activity of the proteasome  and, like several other mutants of proteasome subunits, are resistant to cycloheximide . Pre4-2 mutants suppress a mutation of mitochondrial RNAse P that leads to lack of growth on fermentative carbon sources . Another protein that is functionally connected to Pre4 is Sit4 phosphatase, the concerted action of which is necessary for cell maintenance during starvation-induced G1 arrest .
The human protein PSMB4 is known mainly as an interaction partner for several proteins, suggesting an important function of binding not only to the lid, but also to the 20 S core of the proteasome. The HIV-derived protein Nef1 binds to the amino acids 73–249, where the amino acids 219–249 are of major importance, although not sufficient . Another viral protein binding to PSMB4 is Tax (human T-cell leukaemia virus), an interaction that is supposed to influence the nuclear factor κB pathway .
Signalling of the transforming growth factor β superfamily is also influenced by binding of Smad1 to PSMB4 (reviewed in ), an interaction that occurs with the immature half proteasomes . This is in contrast with SNEV–proteasome interaction that occurs with the mature proteasome, since we do not detect unprocessed PSMB4 in our precipitates, but only the mature form at 24 kDa.
Other interacting partners of PSMB4 are lipopolysaccharides of microbial origin, which have been shown to influence the activity of the proteasome by increasing the chymotrypsin-like activity on binding to PSMB4 .
These reports together with our results suggest that PSMB4 might be a major site for proteasome regulation, where signals from the outside might be transduced to the protease activities inside. However, we were not able to detect changes in proteasome activity on addition of recombinant SNEV in vitro (results not shown).
Although this is the first report showing that a U-box E3 ligase binds to the proteasome and that this binding is not directed to the lid but to the core, co-localization of nuclear proteins, especially of splicing factors, that are targets of proteasomal degradation has already been observed . Since SNEV is also contained within the spliceosome , these results suggest that SNEV might be involved in the degradation of spliceosomal proteins to allow the large structural rearrangements that are necessary for spliceosome activation and splicing catalysis. However, SNEV is also reported to be involved in DNA double-strand break repair  and it is presently not clear which of these functions is dependent on the enzymatic activity of SNEV in the ubiquitin–proteasome degradation pathway. We will get closer to answering this question as soon as the target of SNEV's E3 ligase activity has been identified.
We are grateful to Polymun Scientific, Vienna for generous funding of this project. W. Ernst provided baculovirally expressed SNEV protein. We thank N. Chondrogianni for helpful discussions.
Abbreviations: AD, activation domain; BD, binding domain; CFP, cyan fluorescent protein; CoIP, co-immunoprecipitation; ECFP, enhanced CFP; YFP, yellow fluorescent protein; EYFP, enhanced YFP; FRET, fluorescence resonance energy transfer; GST, glutathione S-transferase; HA, haemagglutinin; HEK-293, cell, human embryonic kidney 293 cell; IκBα, inhibitory κBα; Ni2+-NTA, Ni2+-nitrilotriacetate; SD, synthetic dropout; SNEV, senescence evasion factor; Y2H, yeast two-hybrid
- The Biochemical Society, London