Substrates destined for degradation by the 26 S proteasome are labelled with polyubiquitin chains. Rpn11/Mpr1, situated in the lid subcomplex, partakes in the processing of these chains or in their removal from substrates bound to the proteasome. Rpn11 also plays a role in maintaining mitochondrial integrity, tubular structure and proper function. The recent finding that Rpn11 participates in proteasome-associated deubiquitination focuses interest on the MPN+ (Mpr1, Pad1, N-terminal)/JAMM (JAB1/MPN/Mov34) metalloprotease site in its N-terminal domain. However, Rpn11 damaged at its C-terminus (the mpr1-1 mutant) causes pleiotropic effects, including proteasome instability and mitochondrial morphology defects, resulting in both proteolysis and respiratory malfunctions. We find that overexpression of WT (wild-type) RPN8, encoding a paralogous subunit that does not contain the catalytic MPN+ motif, corrects proteasome conformations and rescues cell cycle phenotypes, but is unable to correct defects in the mitochondrial tubular system or respiratory malfunctions associated with the mpr1-1 mutation. Transforming mpr1-1 with various RPN8–RPN11 chimaeras or with other rpn11 mutants reveals that a WT C-terminal region of Rpn11 is necessary, and more surprisingly sufficient, to rescue the mpr1-1 mitochondrial phenotype. Interestingly, single-site mutants in the catalytic MPN+ motif at the N-terminus of Rpn11 lead to reduced proteasome-dependent deubiquitination connected with proteolysis defects. Nevertheless, these rpn11 mutants suppress the mitochondrial phenotypes associated with mpr1-1 by intragene complementation. Together, these results point to a unique role for the C-terminal region of Rpn11 in mitochondrial maintenance that may be independent of its role in proteasome-associated deubiquitination.
- MPN (Mpr1, Pad1, N-terminal) domain
Most cellular proteins that need to be removed in a timely or regulatory manner are covalently labelled by polyubiquitin chains, and targeted to the proteasome for degradation . Ubiquitination is reversible, due to multiple deubiquitinating enzymes that process these chains or remove them from the tagged substrate [2,3]. The proteasome consists of a 20 S core particle (CP) to which one or two 19 S regulatory particles (RPs) attach, one at either end. Proteolysis takes place within the 20 S CP, while the 19 S RP binds polyubiquitinated substrates, unfolds and translocates them into the 20 S CP for proteolysis . The 19 S RP also removes ubiquitin from the target substrate, recycling ubiquitin even as the substrate is hydrolysed by the 20 S CP . The 19 S RP can be separated further into two subcomplexes: the lid and the base. The base attaches to the outer surface of the 20 S CP, and probably unravels the substrate, simultaneously with gating the channel into the proteolytic chamber [6,7]. The base may also be pivotal in anchoring polyubiquitin substrates, either directly or indirectly, during this process [8,9]. Attachment of the lid to the base is required for proteolysis of ubiquitin–protein conjugates, but not of unstructured polypeptides [10,11]. Interestingly, both the lid and the base have been found to contribute independently to 19 S RP deubiquitination activity. The subunits responsible for this deubiquitinase activity have been identified as Rpn11 in the lid and Ubp6 in the base [11–14].
Both the lid and the evolutionarily related CSN (COP9 signalosome) are composed of eight subunits, six containing a PCI (proteasome/COP9/eIF3) domain, while the other two (Rpn11 and Rpn8 in the case of the lid) contain an MPN (Mpr1/Pad1/N-terminal) domain . Rpn11, the most highly conserved subunit of the lid, is situated within a crosspatch cluster encompassing Rpn5, 11, 9 and 8 . A similar structure is present among the synonymous subunits of the CSN. The central position of Rpn11 within the lid structure and its proposed enzymic role as a deubiquitinase are key features for understanding the diverse phenotypes associated with Rpn11. For instance, Rpn11 plays an important role in the transcriptional response to UV irradiation . Moreover, Rpn11 is also involved in mitochondrial tubular organization and functioning, cell cycle progression and response to cellular damage [18,19]. Underscoring its apparent enzymic role, RPN11 and its orthologues exhibit dominant phenotypes upon overexpression. For example, high dosage of Rpn11 orthologues in human or Schizosaccharomyces pombe cells confers multidrug and UV resistance [20–23]. Overexpression of Rpn11 in Schistosoma results in stabilization of c-Jun . In Saccharomyces cerevisiae, overexpression of RPN11 can suppress a mutation in the importin-α, srp1 .
Rpn11 belongs to a subset of MPN-domain proteins that harbour an MPN+ or JAMM (JAB1/MPN/Mov34) motif. The MPN+ motif is defined by a consensus sequence E-HxHx7Sx2D (where ‘x’ denotes ‘any amino acid’) with similarity to the active site of zinc metalloproteases. Recent structure determination of an archaeal member of the MPN+ family, AfJAMM, identified a zinc active site chelated to the histidine and aspartate residues of the MPN+ motif, with a water molecule probably serving as the fourth ligand and active-site nucleophile [26,27]. These properties lie behind the assertion that members of this family could be hydrolytic enzymes for removal of ubiquitin or ubiquitin-like proteins from their targets. Thus Rpn11 partakes in removal of ubiquitin from substrates bound to the proteasome [11–13, 28], and the CSN subunit Csn5/Jab1 is responsible for removal of the ubiquitin-like modifier Rub1/Nedd8 from the cullin subunit of cullin-based E3 ubiquitin–ligase complexes [29,30]. Indeed, single-site mutations in MPN+ residues of Rpn11 caused a decrease in proteasome deubiquitination, resulting in marked proteolysis defects [11,13,28]. In contrast, simultaneous substitution of both histidine residues located in this motif (the rpn11AXA mutant) was non-viable ; likewise, double substitution of both these histidine residues could not rescue RNA-interference treatment of DmS13/rpn11 in insect cells . Equivalent mutations in Csn5, the MPN+ component of the CSN, abolished the ability of this complex to remove the ubiquitin or ubiquitin-like Rub1 modification from their target proteins [27,30]. This catalytic motif is highly conserved within some MPN domain proteins such as Rpn11 and Csn5, but is not present in others, such as the proteasomal subunit Rpn8, suggesting that the latter are not catalytically active but may play a structural role in their respective complexes.
Although the above results clearly point to a function of Rpn11 in proteolysis, they do not easily explain its participation in maintaining mitochondrial integrity that stems from regions outside the MPN domain. For instance, many studies involving Rpn11 in yeast have been obtained by taking advantage of the mpr1-1 mutant: a nonsense mutation in the C-terminal region that results in a truncated Rpn11 protein lacking the last 31 amino acids [18,19]. This mpr1-1 mutant exhibits pleiotropic phenotypes, such as a temperature-sensitive cell cycle arrest at the G2–M phase with severe abnormalities in the bud, and disruption of the mitochondrial tubular network [18,19]. Interestingly, these mitochondrial defects could be dissociated from defects in the cell cycle by extragenic suppressor mutations . So far, this has been the only case in which a mutation in a proteasomal subunit was shown to cause defects in mitochondrial morphology and function.
In the present study, we have investigated the role of the Rpn11 lid subunit in mitochondrial tubular organization and in protein degradation, and attempt to map regions of the protein essential for either function. We find that overexpression of WT (wild-type) RPN8 is able to partially rescue the proteolysis phenotypes of mpr1-1, but not the mitochondrial defects. We also constructed chimaeric proteins by exchanging the MPN domain of Rpn8 with that of Rpn11. Only chimaeric constructs containing the natural C-terminal region of Rpn11 could suppress the mitochondrial defects. Rpn11 mutants defective in their MPN+ catalytic motif were able to rescue most mpr1-1 phenotypes. These results point to a new function of the C-terminus of Rpn11 in addition to the deubiquitinating function.
Yeast strains, media and genetic strains, plasmids and media
The yeast strains and plasmids used in the present study are listed in Tables 1 and 2. All strains used in this study were derived from the S288C strain. Y25683 (MATa/MATα; his3Δ1/his3Δ1; leu2Δ0/leu2Δ0; met15Δ0/MET15; LYS2/lys2Δ0; ura3Δ0/ura3Δ0; RPN11::kanMX4/RPN11) was a derivative of BY4743 obtained from EUROSCARF (European Saccharomyces cerevisiaeArchive for Functional Analysis; Institut für Mikrobiologie, Johann Wolfgang Goethe-Universität, Frankfurt am Main, Germany), and was used to generate haploid strains to study complementation of rpn11 null. In brief, the diploid strain was transformed with a URA3-marked centromeric plasmid expressing WT RPN11 under its own promoter, pYC-RPN11, and sporulated to obtain a haploid strain in which pYC-RPN11 complements the lethal phenotype of the chromosomal knockout of RPN11 . The resulting haploid strain is referred to in this study as WT-BY, and was later used for introduction of plasmids expressing single-site mutations or chimaeras by the shuffle-in/shuffle-out method. Strains harbouring the mpr1-1 truncation mutation in the RPN11 gene were based on the S288C derivative strain CMY826, referred to herein as WT-CMY, which was obtained from Carl Mann (CEA/Saclay, Gif-sur-Yvette, France).
Standard methods for cloning procedures by PCR were performed according to the manufacturer's instructions. Plasmids for expression of RPN8 or RPN11 from either ADH1 or RPN11 promoters were generated by PCR amplification from genomic yeast DNA by using the primers 5′-GCCC AAG CTT CCG AAG AAA TGA TGG-3′ and 5′-GTCC CCC GGG CTC GAG TTG ATT GTA TGC TTG-3′ to obtain the ADH1 promoter, or the primers 5′-GC CCA AGC TTC AAG TTT ACG ATG GCG C-3′ and 5′-GTC CCC CCG GGC TCG AGT AtG TCT CGT CTT TCT TG-3′ to obtain the RPN11 promoter; each was cloned into the HindIII/Xho1 sites.
RPN8- and RPN11-coding regions were also generated by PCR amplification from genomic yeast DNA by using the primers 5′-GTC CCC CGG GCT CGA GAT GGA ACG ACT ACA GAG AT-3′ and 5′-GTA CCG AGC TCT AGA CCA TTC ATT CCA TTA ACT-3′ for RPN11 or 5′-GTC CCC CGG GCT CGA GAT GTC TCT ACA ACA-3′ and 5′-GTA CCG AGC TCT AGA GTG GAC GAG AAT AGA GG-3′ for RPN8, and cloning into the XhoI/SacI sites. Tagging of different clones at the C-terminus with GFP (green fluorescent protein) was performed between the BglII/BamHI mismatch.
Construction of chimaeric proteins
The synthesis of chimaeric DNA was achieved by inserting a SalI restriction site at the end of the MPN-domain-coding region of each gene. The SalI site was inserted by single-site silent substitutions in RPN11 or RPN8, and generated using PCR site-directed mutagenesis on plasmids with the LEU2 marker for selection (either centromeric YpLac111 or 2 micron). The SalI site was added at the end of the MPN-domain-coding region by using the overlapping primers of RPN8 (5′-CCG CTA TTA TTA ATT GTC GAC GTC AAA CAA CAA GGT GTT G-3′ and 5′-CAC CTT GTT GTT TGA CGT CGA CAA TTA ATA ATA GCG G-3′) or of RPN11 (primers 5′-GTT GCT GTC GTT GTC GAC CCT ATT CAA TCC G-3′ and 5′-GAT TGA ATA GGG TCG ACA AcG ACA GCA ACA G-3′). The area which encodes the N-terminus of each of the proteins was switched with the one from the other by ‘cutting and pasting’ with the restriction enzymes XhoI and SalI.
Yeast culture media
YPD (1% bactopeptone, 1% yeast extract and 2% glucose) and YPG (1% bactopeptone, 1% yeast extract and 2% glycerol) media were used as rich media. WO medium (0.17% yeast nitrogen base, 0.5% ammonium sulphate and 2% glucose) was used as minimal medium. All media were supplemented with 2.3% Bactoagar (Difco) for solid media; WO medium was supplemented with the appropriate nutritional requirements according to the phenotype of the strains. MMS (methyl ethane sulphonate) was used at the concentration of 0.01% or 0.03% in YPD medium. Canavanine was added to minimal medium with other nutritional requirements, but without arginine, at the concentrations of 1, 3, 5, 7 and 10 μg/ml. Cycloheximide (CHX) was used at concentrations of 0.05, 0.1 and 0.5 μg/ml.
Strains were grown at the concentration of 2×107 cells/ml, plated on YPD medium and irradiated at 0, 30, 60, 90 and 120 J/m2 (200 cells/plate; three plates for each point). Plates were incubated at 28 °C in the dark for 3 days and colonies were counted.
Strains were streaked on 3 μg/ml canavanine minimal media plates without arginine and grown at 30 °C for 2 days.
Isolation of human suppressors of the mpr1-1 mutation
The mpr1-1 strain was transformed with the human library (generously provided by Anna Chelstowska, University of Warsaw, Poland). From 40000 transformants, 32 were able to grow on glucose and glycerol media at 34 °C and 11 were able to grow only on glucose medium at 34 °C. After checking that the growth of the transformants was associated with the presence of a plasmid, the inserts were sequenced. The transformants with a wild-type phenotype contained the POH1 gene (human homologue of RPN11; ); transformants able to grow only on glucose medium at 34 °C contained the human proteasomal RPN8 gene.
The suppressor gene RPN8 obtained from the human library was identified using the M13 reverse primer. The primers used for RPN8 amplification were RPN8-5′ (5′-CCT CCA TTA GGG CCT AAC-3′) and RPN8-3′ (5′-GGA CGA GAA TAG AGG CG-3′).
For DAPI (4,6-diamidino-2-phenylindole) staining, cells were harvested in exponential phase, and fixed with 1% formaldehyde for 30 min. DAPI was added at a concentration of 1 μg/ml and cells were observed by fluorescence microscopy. The vital dye DASPMI [2-(4-dimethylaminostyryl)-N-methylpyridinium iodide] was used at a final concentration of 10−6 M; cells were then observed by fluorescence microscopy and photographed directly from the culture.
Protein gel electrophoresis and immunoblotting
SDS/PAGE was usually performed using 12% (w/v) separating gels, unless stated otherwise. For immunoblotting, proteins were wet-transferred to nitrocellulose membranes, and blotted with appropriate antibodies. For in-gel proteasome visualization, rapidly prepared whole-cell extracts were resolved by non-denaturing PAGE, as published previously .
Multicopy RPN8 suppresses the cell cycle, but not mitochondrial defects of mpr1-1
The mpr1-1 mutation results in a truncated Rpn11 protein . As previously reported, the mpr1-1 mutant is slow growing in the 24–28 °C temperature range, and sensitive to higher temperatures; a shift to 34 °C induces a cell cycle arrest at the G2–M phase, resulting in lethality. Figure 1(A) shows highly aberrant cell morphology following this temperature shift, in particular elongated buds with an undivided nucleus (Figure 1A left panel). We screened for possible multicopy suppressor genes present in homologous or heterologous libraries. One interesting suppressor gene we found in both human and S. cerevisiae libraries was RPN8. Overexpression of RPN8 suppresses the temperature-induced cell cycle arrest of mpr1-1, and allows for growth at the non-permissive temperature (Figure 1A right panel). In addition to the cell cycle phenotype, mpr1-1 cells have malformed and malfunctioning mitochondria. Even at the lower permissive temperature, mpr1-1 harbours multiple spherical mitochondria (Figure 1B centre panel), which differ from the elongated tubular mitochondria present in WT cells (left panel). These mutants are unable to grow on a non-fermentable carbon source at 34 °C. Overexpression of RPN8 does not suppress this phenotype (Figure 1B, right panel).
Domain mapping of Rpn11
The suppressive effects of Rpn8 are not easy to explain; both RPN11 and RPN8 are essential genes, the products of which are present at approximately stoichiometric levels in the 19 S RP [32,33]. How then does an overdose of Rpn8 correct phenotypes associated with Rpn11? Rpn8 is a paralogue of Rpn11; both contain an MPN domain at their N-terminal region (Figure 2A). Interaction between those two proteins was shown previously (perhaps by physical interaction between these domains [16,34]), although it should be emphasized again that, in contrast with Rpn11, the MPN domain of Rpn8 does not contain the catalytic MPN+ motif; furthermore, the C-terminal parts of the two proteins do not share obvious similarities. In order to verify specific roles for these domains within Rpn11 or Rpn8, we created chimaeras by interchanging their N-terminal MPN domains (Figure 2A): CHR11-R8 (N-terminal region of Rpn11 with C-terminus of Rpn8) and CHR8-R11 (N-terminal region of Rpn8 with C-terminus of Rpn11). These chimaeras were tested for complementation and had no ability in rescuing the rpn11 null, either at natural abundance or when overexpressed (results not shown). However, both chimaeric proteins can rescue some of the phenotypes associated with mpr1-1, when expressed in the mpr1-1 mutant strain.
Chimaeras of Rpn11 with Rpn8 in either combination rescue the temperature-sensitivity of mpr1-1, allowing for growth at the restrictive temperature (Figure 2B left panel). Overexpression of RPN8, or of its human orthologue hsRPN8, does so as well. However, rescue of the respiratory defects associated with mpr1-1 is much more stringent. Thus Rpn8 supports growth of mpr1-1 on glucose-containing medium, but not on the non-fermentable carbon source, glycerol, while multicopy CHR8-R11 allows for growth on glycerol (Figure 2B right panel). This result suggests that properly functioning mitochondria require the C-terminal region of Rpn11. In order to corroborate this observation, we analysed the ability of various constructs to repair mitochondrial structural defects observed in mpr1-1 (Figure 3). Indeed, the C-terminal part of Rpn11 is necessary for the maintenance of a correct mitochondrial tubular network (Figure 3E). To conclude, a WT copy of the C-terminal region of Rpn11 is required for proper mitochondrial function, even if this domain is detached from the catalytic domain in its N-terminal region.
The C-terminal domain of Rpn11 functions independently of the N-terminal metalloprotease MPN+ motif
Having observed that the C-terminal part of Rpn11 is necessary for the maintenance of a correct mitochondrial tubular network, we wished to investigate the relative phenotypes associated with Rpn11 domains. Single-site mutants in the MPN+ sequence E-HxHx7Sx2D of Rpn11 exhibit general proteolytic defects, accumulate polyubiquitinated proteins and are temperature-sensitive , yet contain a normal-looking tubular mitochondrial network (Figure 4B). We transformed mpr1-1 with plasmids expressing these point mutations in Rpn11, and found that they are able to rescue the growth of mpr1-1 at the restrictive temperature of 34 °C and allow for utilization of glycerol as a carbon source as well (Figure 4A). Such intragene complementation of two slow-growing and temperature-sensitive mutants may indicate that the two domains of Rpn11 function independently, and there is no need for an intact molecule bearing both a functional MPN+ domain and the C-terminal part. We may deduce that the participation of Rpn11 in mitochondrial function may be independent of its deubiquitination activity.
Mutants in the C- and N-terminal regions of rpn11 exhibit distinct phenotypes
To distinguish between the relative roles of the C- and N-terminal regions of Rpn11 we compared additional phenotypes. Low levels of amino acid analogues such as canavanine are known to promote accumulation of damaged proteins, which must be removed by the proteasome. Thus many proteasome mutants are sensitive to low levels of canavanine [16,35]. As expected, the MPN+ mutant rpn11D122A is sensitive to 3 μg/ml canavanine in the growth medium, although surprisingly mpr1-1 behaves similarly to WT and is insensitive at these levels (Figure 5A). This result corroborates the conclusion that the effects of mpr1-1 are inherently different from those of mutations in the MPN+ motif. Underscoring this conclusion, mutants in the MPN+ motif were found to be sensitive to UV damage and exposure to low levels of the translation inhibitor CHX, while mpr1-1 was not (Figures 5B and 5C). Sensitivity to low levels of CHX is reported to be a consequence of depleted levels of ubiquitin that occur when degradation of ubiquitin is faster than synthesis [36,37]. Thus sensitivity to low levels of CHX (as shown in Figure 5C) can be a good indication of malfunctions in the ubiquitin-proteasome pathway, in contrast with sensitivity at high concentrations that reflect general translational arrest. Overall, the phenotypes of rpn11MPN mutants are typical of mutants in the ubiquitin-proteasome pathway, whereas those of mpr1-1 are pleiotropic. Curiously, both MPN+ mutants and mpr1-1 are sensitive to the alkylating agent MMS (Figure 5D). This sensitivity can be rescued by intragene complementation of rpn11MPN and mpr1-1 mutant copies, suggesting that different regions of Rpn11 participate differently to MMS adaptation. In contrast with suppression of the temperature-sensitivity of mpr1-1, Rpn8 cannot rescue the MMS sensitivity of mpr1-1 (Figure 5D), suggesting a requirement of an intact C-terminus of Rpn11 in the MMS-damage response, but surprisingly not for the UV-damage response.
Suppression of proteasome structural defects does not rescue mpr1-1 mitochondrial phenotypes
To verify that the chimaeric proteins are indeed expressed and can be stably incorporated into proteasomes, we analysed the migration pattern of tagged Rpn11 constructs. Fractionation of whole-cell extract using a glycerol gradient separates proteasomes from lower-molecular-mass proteins in their native state. Upon overexpression, tagged Rpn11 or CHR8-R11 chimaeras are incorporated into the proteasome, but are also found in low-molecular-mass fractions (Figure 6). Highly abundant Rpn8 behaves similarly (results not shown). Incorporation of these tagged proteins into assembled proteasomes is similarly efficient in WT and mpr1-1 strains (Figure 6). Finding two pools of Rpn11 indicates that excess Rpn11 can exist in a stable form that is not associated with the proteasome, and may explain the documented phenotypes observed upon overexpression of this subunit (see the Introduction).
A slight shift towards lower-molecular-mass fractions was found for proteasomes in mpr1-1 that were not covered by rescue plasmids (Figure 6 bottom panel). This observation agrees with a report showing structural defects in proteasomes purified from mpr1-1 . To test whether a proteasome structural defect is the source of other mpr1-1 phenotypes, we studied proteasome conformation in various rpn11 strains. Crude cell extracts from strains grown at the permissive temperature were resolved by nondenaturing PAGE, and proteasomes were visualized by fluorogenic peptide overlay (Figure 7A). The overwhelming majority of proteasomes from exponentially growing WT yeast are found as doubly capped or singly capped 26 S holoenzymes with nearly undetectable levels of free 20 S CP, in agreement with previous results . In contrast, proteasomes that were rapidly resolved from whole-cell extracts of the mpr1-1 strain are almost exclusively present as ‘lidless’ base–CP complexes (Figure 7A). An elevation in temperature promotes proteasome dissociation, resulting in increased levels of singly capped base–CP complexes and 20 S CP over the larger proteasome conformations (Figure 7B). Quite unexpectedly, these defects in proteasome structure are almost completely rescued by highly abundant Rpn8 or the chimaeric protein CHR8-R11, even though neither carries a functional MPN+ domain. It should be noted that no structural defects are observed for proteasomes purified from mutants in the MPN+ motif of Rpn11 , supporting a largely functional, rather than a structural, role for the MPN+ motif itself. An additional observation regarding proteasomes from the mpr1-1 strain is that they appear naturally repressed compared with WT (Figure 7). Lower levels of peptidase activity were measured for proteasomes prepared from mpr1-1 whole-cell extract (controlled for equal proteasome levels as determined by protein staining and immunoblotting), in agreement with diminished peptidase activity observed for purified proteasomes from mpr1-1 .
It would be expected that such dramatic structural and functional proteasomal defects would give rise to general proteolytic phenotypes. Indeed, slight accumulation of polyubiquitinated proteins is detected in mpr1-1, even at the permissive temperature (Figure 8A). Accumulation is greatly pronounced following a shift to the restrictive temperature for mpr1-1 cells (Figure 8B), in agreement with the marked decrease in intact proteasome levels (Figure 7B). That the mpr1-1 mutant does not show sensitivity to canavanine (Figure 5A) is surprising in light of these results, as it suggests that the proteolysis defect of mpr1-1 may be limited to a subset of proteins, rather than a general effect. Overexpression of Rpn8 almost completely rescues the accumulation of polyubiquitinated proteins, probably because of Rpn8 correcting proteasome structural defects (Figures 6–8). Independently, accumulation of polyubiquitinated substrates was also observed for MPN+ mutants . Interestingly, expression of rpn11D122A partially corrects proteolysis defects in mpr1-1 (Figure 8B), indicating that subunits lacking a functional MPN+ motif can rescue proteolysis defects associated with mpr1-1. We conclude that mpr1-1 and MPN+ mutants do not influence proteolysis in the same manner.
The tubular organization and cellular distribution of mitochondria is a highly dynamic process that is shaped by a tight equilibrium between fusion and fission events [39,40]. The ubiquitin-proteasome pathway may influence mitochondrial structure and function in multiple ways. For instance, it has long been documented that breakdown of some mitochondrial proteins is an ATP- and ubiquitin-dependent process, attributed to the proteasome . Targets of proteasomal proteolysis could be proteins normally situated in the outer mitochondrial membrane. A fraction of proteasomes may even be involved in the quality-control system on the outer membrane of the mitochondria, similar to its documented role in ER-associated degradation. Alternatively, ubiquitination could be involved in targeting or degrading nuclear-encoded proteins on their way to be imported into the mitochondria [42,43]. Underscoring the link between the ubiquitin system and the mitochondrial network, aberrant mitochondrial morphology and inheritance have been observed in mutants encoding E3 ubiquitin ligases. Examples include Rsp5, an E3 that is also involved in endocytosis and cytoskeleton remodelling , or the F box protein Mdm30, which may be responsible for targeting the mitochondrial fusion protein fusion protein Fzo1 for ubiquitination . At the other end of the ubiquitin system, a deubiquitinating enzyme Ubp16 has been localized to the mitochondria outer membrane, although its purpose there is unknown at present . Finally, a connection between the proteasome itself and mitochondria structure can be inferred from phenotypes associated with mutations in two proteasome subunits. Mitochondrial defects caused by a mutation in the mitochondrial protease Yme1 are suppressed by a mutation in the rpt3 locus that encodes an ATPase subunit of the proteasome . Since Yme1 is localized to the inner membrane of the mitochondria, indirect participation of the ubiquitin-proteasome pathway is suggested in this case. In a more direct example, mutations in the extreme C-terminal region of the proteasome lid subunit Rpn11 (the mpr1-1 mutant) cause mitochondria malformation and malfunction [18,19].
In the present study, we observed pleiotropic effects of Rpn11 in both proteolysis and mitochondrial integrity. On the one hand, mutations in the MPN+ motif at the N-terminal domain cause proteolytic defects resulting from reduced deubiquitinating activity associated with this domain. On the other hand, mutations at the C-terminal region lead to both proteasome and mitochondria malfunctions. Interestingly, we now show that proteasome structural defects and general proteolytic defects associated with mpr1-1 can be suppressed, for the most part, by high amounts of another lid subunit, Rpn8 (Figures 1, 2, 5 and 8, and Table 3). The suppressive effects of Rpn8 are not easy to explain; both RPN11 and RPN8 are essential genes, the products of which are present at approximately stoichiometric levels in the 19 S RP. As mentioned above, Rpn8 does not contain the catalytic MPN+ motif and is not expected to participate in deubiquitination. We can speculate that high levels of Rpn8 could stabilize Rpn11-containing lid precursors, such as the Rpn5/11/9/8 cluster . Nevertheless, it is surprising that repairing proteasome conformations and proteolysis defects by expressing high levels of Rpn8 is insufficient to correct defects in the mitochondrial tubular system observed in mpr1-1 (Figures 1 and 2). Rpn8, or other constructs that do not contain the C-terminal region of Rpn11, cannot correct for phenotypes associated with mitochondrial structure and respiratory function (Table 3). In contrast, expression of mutant or chimaeric proteins that contain the natural C-terminal region of Rpn11 does suppress the mitochondrial phenotypes by intragene complementation (Figure 2 and Table 3). Overall, these results point to a unique role for the C-terminal region of Rpn11 in mitochondrial maintenance, beyond the function of Rpn11 as a deubiquitinase that emanates from its N-terminal domain.
A dual role for Rpn11 could explain how the mitochondrial effects of the mpr1-1 mutation (which are not common to other proteasomal mutants) can be dissociated from phenotypes associated with other rpn11 mutants. The situation regarding Rpn11 is reminiscent of the pleiotropic roles executed by two distinct domains of another 19 S RP subunit, Rpn10. On the one hand, Rpn10 has been shown to bind polyubiquitin chains via the ubiquitin-interacting motif present in its C-terminal region. On the other hand, Rpn10 independently participates in stabilizing the structure of the 19 S RP via a von Willebrand factor A domain in the N-terminal portion of the protein . Pertinent to its function, Rpn10 is found in two pools: proteasome-bound and ‘free form’ . The free form has been proposed to shuttle polyubiquitinated substrates to the proteasome, whereas the proteasome-bound form is pivotal for stabilizing lid–base interactions [9,10,16]. In this context, suppression of cell-cycle and temperature-sensitivity phenotypes of mpr1-1 by RPN8 overexpression, but not of mitochondrial ones, is in agreement with the possibility that the two functions ascribed to Rpn11 could be carried out independently (possibly, although not certainly, by two separate pools of Rpn11 molecules; Figures 5–8).
That mpr1-1 cells are viable, despite finding the majority of proteasomes in a lidless configuration, questions whether a tight interaction between lid and base is necessary for proteasome function. Clearly, all lid subunits are essential (with the exception of Rpn9), and are found at approximately stoichiometric levels in purified proteasomes [32,33,49]. The lid is also required for proper degradation of both mono- and poly-ubiquitinated substrates in vitro [10,11]. Nonetheless, under certain conditions the lid can detach rather easily from proteasomes, suggesting that in vivo dynamic association of the lid with the proteasome takes place [10,12,34,49]. It is not unreasonable to surmise that the lid has independent functions as well. Finding stable pools of Rpn11 that are not associated with the proteasome (Figure 6) supports such a possibility, and may explain the documented phenotypes observed upon overexpression of this subunit.
The recent revelation that Rpn11 participates in proteasome-associated deubiquitination focuses interest on the MPN+ metalloprotease site in its N-terminal MPN domain. However, in the present study we show that the role of Rpn11 in mitochondria maintenance does not necessarily invoke these properties of Rpn11. In particular, the observation that the mpr1-1 mutant is insensitive to conditions typically associated with defects in the proteasome or other components of the ubiquitin system, such as exposure to amino acid analogues, low levels of CHX or low UV dosage (Table 3), points to a distinct role of Rpn11. At this stage, it is unclear whether Rpn11 functions independently in mitochondrial maintenance, or whether it serves to link mitochondria with proteasomes for unique proteolytic needs. One hypothesis is that the C-terminal part of Rpn11 could partake in contacting proteins at the mitochondrial surface, or serves to recruit a subpopulation of proteasomes to the mitochondria. For instance, levels of Fzo1, a protein localized to the outer membrane and involved in mitochondrial fusion, are regulated by ubiquitination . Whether the proteasome in general, and Rpn11 in particular, partake in regulating Fzo1 levels should be a topic of future research.
We are grateful to Anna Chelstowska for many helpful discussions, and to Noa Reis for suggestions and technical assistance. Anna Chelstowska generously provided the human library. Many thanks to Benedikt Westermann for the mitochondrial GFP plasmid, and to Allen Taylor of the anti-ubiquitin antibody. This work was partially supported by Center of Excellence BEMM, MIUR 2003 University of Rome “La Sapienza” and FIRB (code RBNE01KMT9) to L.F., and grants from the Israel Science Foundation (ISF), the German–Israel Binational Foundation (GIF) and the Wolfson Foundation for research on ubiquitin to M.G.
Abbreviations: CHX, cycloheximide; CSN, COP9 signalosome; CP, core particle; DAPI, 4,6-diamidino-2-phenylindole; DASPMI, 2-(4-dimethylaminostyryl)-N-methylpyridinium iodide; GFP, green fluorescent protein; JAMM, JAB1/MPN/Mov34; MMS, methyl methane sulphonate; MPN, Mpr1/Pad1/N-terminal; RP, regulatory particle; WT, wild-type; Ub, ubiquitin
- The Biochemical Society, London