For correspondence (fax +41 61 6973976; e-mail firstname.lastname@example.org).
Targeted protein degradation plays an important regulatory role in the cell, but only a few protein degradation signals have been characterized in plants. Here we describe three instability determinants in the termini of the cauliflower mosaic virus (CaMV) capsid protein precursor, of which one is still present in the mature capsid protein p44. A modified ubiquitin protein reference technique was used to show that these motifs are still active when fused to a heterologous reporter gene. The N-terminus of p44 contains a degradation motif characterized by proline, glutamate, aspartate, serine and threonine residues (PEST), which can be inactivated by mutation of three glutamic acid residues to alanines. The signals from the precursor do not correspond to known degradation motifs, although they confer high instability on proteins expressed in plant protoplasts. All three instability determinants were also active in mammalian cells. The PEST signal had a significantly higher degradation activity in HeLa cells, whereas the precursor signals were less active. Inhibition studies suggest that only the signal within the N-terminus of the precursor is targeting the proteasome in plants. This implies that the other two signals may target a novel degradation pathway.
Protein degradation plays an important role in many cellular processes: it allows much faster alteration of the amount of regulatory proteins than transcriptional or translational regulation, and is important for the relocation of biochemical resources. Although protein degradation has not been extensively studied in plants, accumulating evidence indicates that it is a complex process involving a multitude of proteolytic pathways, with each cellular compartment likely to use one or more of them (Vierstra, 1997).
The most widely used pathway in both nucleus and cytoplasm is the 26S proteasome, a multicatalytic protease complex that is conserved in all eukaryotes. Proteins are targeted to the 26S proteasome mostly by covalent linking of ubiquitin chains. This attachment is catalysed by three enzymes: ubiquitin-activating enzyme (E1); one or more ubiquitin-conjugating enzymes (E2); and, usually, a ubiquitin protein ligase (E3) (Hershko and Ciechanover, 1998).
In animals and yeast, the ubiquitin/26S proteasome pathway plays a role in a number of cellular processes, primarily by controlling the degradation of short-lived enzymes and regulatory proteins. Understanding of the role of this pathway in plants is still rudimentary. Although numerous components have been identified, only phytochrome A and the A- and B-type mitotic cyclins are known to be natural targets in plants (Clough et al., 1999; Genschik et al., 1998), whereas the tobacco mosaic virus movement protein is the first example of a plant virus using this pathway (Reichel and Beachy, 2000).
Little is known about other degradation pathways in either the nucleus or cytoplasm of plant cells. In animals and fungi, a family of Ca2+-activated neutral proteases, calpains, is involved in targeted proteolysis in the cytoplasm (Shumway et al., 1999). However, so far no plant homologues have been identified, making it unlikely that this pathway exists in plants.
Viruses can use the degradation pathways of their host for their own purposes (Mulder and Muesing, 2000). We have observed previously that although cauliflower mosaic virus (CaMV) particles are stable, its capsid protein (CP) precursor is very unstable in plant protoplasts if expressed outside the viral context (Leclerc et al., 1999). This protein might therefore be a good model for the study of targeted protein degradation in plants.
CaMV is the type member of the caulimoviruses, which contain a circular double-stranded DNA genome and replicate via reverse transcription (Rothnie et al., 1994). Virions are approximately 53 nm in diameter with a T = 7 icosahedral symmetry (Cheng et al., 1992). The capsid is built from three related proteins: p44, p39 and p37. These are derived from bi-terminal processing of the 57 kDa CP precursor, the product of ORF IV. Both terminal domains of the CP precursor are very acidic (Figure 1), and the function of these domains is not yet known. The N-terminus of p44 results from proteolytic cleavage of the CP precursor between amino acids 76 and 77 by the viral protease (Martinez-Izquierdo and Hohn, 1987; Torruella et al., 1989). The smaller CP forms p39 and p37 are created from p44 by further processing on both termini, but how this processing occurs has not been elucidated to date.
The sequence between amino acids 32 and 97 of the CP precursor is characterized by prolines, glutamic and aspartic acids, serines and threonines, and is not interrupted by basic amino acids. Such motifs are termed PEST motifs – these have been found in many unstable mammalian proteins, and it has been shown in several cases that degradation is mediated by the 26S proteasome (Rechsteiner and Rogers, 1996). It is still unknown if PEST motifs also play a role as degradation signals in plants. In phytochrome A, a PEST motif has been implicated in protein degradation, but in this case the PEST domain is not sufficient to confer protein instability (Clough et al., 1999).
Degradation signals are often studied by fusion to a stable reporter gene. To prove that differences in protein accumulation result from degradation, and are not due to changes in production, pulse-chase analysis followed by immunoprecipitation is generally used. To reduce the error in such studies, the ubiquitin protein reference (UPR) system was developed (Levy et al., 1996), in which the protein under investigation is produced as a translational fusion to a stable reference protein separated by a ubiquitin monomer. Such fusions are rapidly cleaved by ubiquitin-specific processing proteases, yielding equimolar amounts of test and reference proteins.
Measuring protein half-lives directly by pulse-chase labelling is difficult in plant cells due to the low abundance of transiently expressed proteins, therefore previous studies of differential stability of plant proteins have depended on analysis of steady-state levels of test proteins (Worley et al., 1998). However, with the UPR system differences in synthetic rates or transcript stability are corrected for, making pulse data sufficient to prove instability.
In this study we adapted the UPR technique by using reporter genes that can easily be measured in cell extracts as reference and test proteins, and show that this modified UPR system can be used in both plant and animal systems. We used this technique to show that the instability of the CP and its precursor can be localized to certain transferable instability motifs. We show that these signals are also active in mammalian cells. Furthermore, we tested whether the proteasome degradation pathway might control CP and CP precursor degradation, and discuss the role of this instability in the context of the virus life cycle.
Stabilization of CP and CP precursor fragments by mutations in the PEST signal
ORF IV fragments corresponding to amino acid positions 1–332, 77–332 and 125–332 were fused in-frame to an N-terminal hemagglutinin epitope (HA11). The resulting plasmids, p(1–332), p(77–332) and p(125–332), were transiently expressed under CaMV 35S promoter control in Nicotiana plumbaginifolia leaf protoplasts. As mentioned above, position 77 corresponds to the start of mature capsid protein p44. The protoplasts were harvested after 12 h and total protein extracted. Proteins were separated by SDS–PAGE and detected by Western blotting with anti-HA11 monoclonal antibodies. Significant amounts of p(125–332) could be isolated and easily identified by Western blotting (Figure 2). However, under similar conditions very little p(77–332) and no p(1–332) at all could be detected, indicating that these proteins were unstable. To test whether this instability correlated with an acidic domain that contains the phosphorylatable serines (the sequence between amino acids 82 and 91), two mutations were produced: serines 82, 86 and 88 were replaced by alanines in mutant 3S▸A, and glutamic acids 83, 84 and 85 were replaced by alanines in mutant 3E▸A. p(77–332) was stabilized considerably by the 3S▸A mutation, as had been demonstrated earlier (Leclerc et al., 1999). However, an even more profound stabilization was achieved by introducing the 3E▸A mutation (Figure 2).
This suggests that the cluster of serines and acidic amino acids between positions 82 and 91 is in fact an instability element. In the context of the CP precursor [p(1–332)], the effect of 3S▸A was not visible (not shown), but a small stabilization effect of the 3E▸A mutation became evident. We concluded that additional instability elements are located between positions 1 and 76.
The UPR technique
To further characterize the putative degradation signals in the CP and CP precursor, the UPR technique (Levy et al., 1996) was employed (Figure 3a). In this system, a test protein is produced as a translational fusion to a stable reference protein separated by a ubiquitin monomer. This ubiquitin contained a K48R substitution to prevent the conjugation of further ubiquitin monomers, which would lead to protein degradation. Such fusions are rapidly and precisely cleaved at the C-terminus of ubiquitin by Ub-specific processing proteases (UBPs), yielding equimolar amounts of the test and reference proteins. This proteolytic activity is not affected by the nature of the downstream amino acid. The UPR technique avoids the need for pulse-chase labelling to study protein degradation signals, as determining the steady-state molar ratios of test and reference proteins in cell extracts can yield a direct ranking of their metabolic stabilities (Levy et al., 1996). We adapted the UPR technique by using two stable reporter genes. The β-glucuronidase (GUS) gene served as the internal control, and the CP fragments, fused to either the N- or the C-terminus of chloramphenicol acetyl transferase (CAT), as the test protein (Figure 3a).
All CAT fusions started with a methionine to exclude degradation by the N-end rule, which relates the in vivo half-life of a protein to the identity of its N-terminal residue (Varshavsky, 1996). These reporter proteins could be directly quantified in crude extracts of plant protoplasts or HeLa cells. For transient expression studies in N. plumbaginifolia protoplasts expression was driven by the duplicated 35S promoter of CaMV, and in HeLa cells by the human cytomegalovirus (hCMV) immediate early promoter.
Characterization of the protein instability element in the mature CP p44
To test whether the PEST signal identified at the N-terminus of p44 functions is a transferable degradation motif, the first 53 amino acids of p44 were fused to CAT in the UPR system (Figure 3a). Western blot analysis revealed a >99% cleavage efficiency (Figure 3b), indicating that the yeast ubiquitin is recognized efficiently by plant UBPs. Confirming the observations in Figure 2, the CP fusion 77–120::CAT accumulated to a steady-state level of only 20% compared to the non-fused CAT control. This level was increased to 40% by introducing the 3S▸A mutation. For unknown reasons, this value was highly variable. However, the 3E▸A mutation consistently produced a more stable protein. An additional triple replacement (D87A/E90A/E91A) or the combination of the 3E▸A with the 3S▸A mutation gave rise to proteins that were almost as stable as the control.
To study the effect of the 3S▸A and 3E▸A mutations on virus replication, mutant virus genomes carrying these modifications were used to inoculate turnip plants. The 3S▸A mutant virus was non-infectious, confirming earlier results (Leclerc et al., 1999). Symptoms originating from the 3E▸A mutant virus appeared with a 7-day delay (after 17 days compared to 10 days for the wild type). Virus particles were isolated from plants infected with mutant and wild-type viruses, and analysed by SDS–PAGE and silver staining. The ratio between the three forms of the capsid protein was similar in both samples, indicating that this mutation has no apparent influence on the processing of p44 to p39 and p37 (data not shown). After three passages, a partial revertant was obtained in which A85, but not A83 and A84, was changed back to the original glutamic acid.
Mapping of further destabilization signals in the CP precursor
In the UPR system, 1–120::CAT (containing the N-terminus of the CP precursor) was as unstable as the 77–120::CAT protein (containing the N-terminus of the largest capsid protein) (Figures 3 and 4). However, introduction of the 3E▸A replacement into 1–120::CAT did not lead to significant stabilization of the fusion, in contrast to its effect in 77–120::CAT (see above). This indicates the presence of additional instability elements between amino acids 1 and 76 of the CP precursor. Separate analysis of this sequence and two subfragments, 1–31 and 32–76 (Figure 4), revealed that in fact a strong instability element is located within the first 31 amino acids of the CP precursor.
To determine if the acidic C-terminus of the CP precursor protein also contains a destabilizing element, C-terminal fragments of decreasing length were fused to the C-terminus of CAT in the UPR system. CAT::439–489 and CAT::447–489 were very unstable; CAT::455–489 was still moderately unstable; but CAT::463–489 was stable (Figure 4). This shows that the C-terminus also contains a degradation signal.
Thus we have identified three transferable instability determinants within the CP precursor (Figure 1), one of which is still present in the mature CP p44.
CP precursor instability signals also function in HeLa cells
To see if the three mapped degradation signals are plant-specific, we studied the degradation of CAT containing these signals in HeLa cells. All three signals destabilized the reporter protein, but the CP-based signal was much stronger in the human system than in plant protoplasts, whereas degradation of CAT fusions containing the CP precursor-based signals was less pronounced (Table 1). This correlated with the high score of the CP-based signal using the PEST-find algorithm (Rechsteiner and Rogers, 1996) which was developed, based on motifs frequently occurring in rapidly degraded proteins in mammalian cells, to determine objectively whether a protein contains a PEST region.
Table 1. Activity of CP and CP precursor degradation signals in mammalian cells
na; Not applicable as the fragment does not contain a valid PEST signal.
Proteasome inhibition reduces degradation of 1–31::CAT
To study whether the CP precursor degradation signals target the proteasome, we employed proteasome inhibitors. As a positive control in our system, we created GUS-ub-CAT fusions with phenylalanine or arginine at the start of CAT. According to the N-end rule (Varshavsky, 1996), these amino acids should target the CAT protein for degradation through the ubiquitin pathway in mammalian systems and also in plants (Worley et al., 1998). However, these constructs were unexpectedly stable (data not shown), probably because the N-terminus of CAT was not exposed, or no lysine in CAT was accessible to the E3–E2 complex (Suzuki and Varshavsky, 1999). This problem was solved by using a construct with a 31 amino-acid lysine-free linker between the arginine at the start and CAT (R-CAT) (Levy et al., 1999).
We analysed the effect of the proteasome inhibitors MG132, lactacystin and clasto-lactacystin β-lactone. MG132 is a reversible, competitive inhibitor, whereas lactacystin and clasto-lactacystin β-lactone are irreversible inhibitors that modify all catalytic beta subunits (Craiu et al., 1997). Lactacystin had no effect, as was also reported for tobacco BY2 protoplasts (Reichel and Beachy, 2000). Both MG132 and clasto-lactacystin β-lactone had a strong inhibitory effect on protein production in the protoplasts. However, after correcting for this effect using the GUS internal control and by using a low inhibitor concentration, we could see a fourfold stabilization of R-CAT using clasto-lactacystin β-lactone (Figure 5), while the CP precursor fragment 1–31::CAT fusion was stabilized threefold.
The ratio for 77–120::CAT and CAT::455–489 was similar to that of the CAT control and a stabilized mutant, showing that their degradation was not inhibited. This indicates that the sequence 1–31 activates degradation by the proteasome, whereas the two other instability sequences may be subject to a different degradation pathway.
Variations in protein stability play an important role in the regulation of cellular processes. In eukaryotes, several motifs have been correlated with instability, including certain N-terminal residues: the N-end rule (Varshavsky, 1996); the cell cycle-related destruction (D) box (Glotzer et al., 1991); the KEN-box (Pfleger and Kirschner, 2000); and the PEST motif (Rechsteiner and Rogers, 1996). In many, but not all, cases instability correlates with ubiquitination and degradation by the proteasome. Instability motifs have been studied, mainly in yeast and mammalian systems, but the N-end rule and the D box have also been shown to function in plants (Genschik et al., 1998; Worley et al., 1998). However, no data showing the involvement of a PEST motif in plant protein degradation have been reported so far.
Here we describe three separate, transferable instability elements that contribute to the instability of the CaMV CP precursor. One of these is still present in the mature CP p44. The highest destabilization effect in plant protoplasts was achieved with polypeptides including the first 31 or the last 49 amino acids of the CP precursor, although neither of these sequences contains any of the established instability motifs. On the other hand, the N-terminal sequence of the mature CP protein, fragment 77–120, scores highly in the PEST algorithm, but ranks only third in destabilizing activity. Those motifs behaved differently when experiments were performed in HeLa cells, where sequence 77–120 caused the highest destabilization of the reporter protein (Table 1).
Phosphorylation often plays a role in the regulation of PEST signal-targeted degradation. Three serines at positions 82, 86 and 88 have been shown to be the major phosphorylation targets in the N-terminal part of p44 (Leclerc et al., 1999). Mutation of these serines to alanine in p(77–332) as well as in 77–120::CAT increases stability of the proteins, but to a lesser extent than mutation of glutamic acids 83, 84 and 85 to alanines. These data suggest that phosphorylation enhances, but is not essential for, targeting the proteins for degradation.
The degradation signal within the C-terminal part of the CP precursor (amino acids 439–489) is also acidic, but does not correspond to a valid PEST motif, because the prolines at positions 449 and 453 are separated from the acidic domain by a basic residue. However, we show that the domain containing the prolines still contributes to protein instability, indicating that the degradation signal may well be PEST-related. Based on these data, we suggest that PEST-related signals do act as degradation signals in plants, but that the PEST algorithm as developed for mammalian proteins needs to be adapted for plant systems.
The N-terminal sequence of the CaMV CP precursor includes a hydrophobic cluster embedded between charged residues. In this respect, it resembles the C-terminal signals involved in ER protein quality control via retrograde transport into the cytoplasm and degradation by the proteasome (Bonifacino et al., 1991; Wiertz et al., 1996). It is possible that such signals also act as degradation signals for proteins located in the cytoplasm.
The ubiquitin–proteasome pathway represents the main protein-degradation machinery in eukaryotic cells (Hershko and Ciechanover, 1998), and many PEST-containing proteins have been shown to be degraded by this pathway (Rechsteiner and Rogers, 1996). However, from our data it appears that the signal contained in the first 31 amino acids of the CP precursors in fact targets the proteasome, whereas the two PEST-like instability determinants induce a different degradation pathway. Recent data show that there are many differences in the mechanisms by which PEST motifs can be targeted for degradation. For example, ODC is targeted to the proteasome by antizyme instead of ubiquitin (Murakami et al., 1992), and C-Myc contains a PEST that seems to influence stability after the ubiquitination step (Gregory and Hann, 2000). It has also been shown that PEST sequences can affect protein stability by targeting proteins to alternative degradation pathways that are different from the proteasome. For example, IκBα contains a PEST degradation signal that targets degradation by µ-calpain (Shumway et al., 1999). Calpains are Ca2+-activated cysteine proteases. They have been described in animals and fungi, but so far no plant homologues have been identified, making it unlikely that this pathway exists in plant cells. Thus two out of three of the instability determinants characterized here may target a novel, possibly plant-specific, degradation mechanism.
Although the instability elements defined here are transferable, their effects vary depending on the context and are not simply additive. For instance, in the UPR system amino acids 1–31 alone bring more instability to the fusion than the larger fragments 1–76 or 1–120. Improper folding of the smaller fragment may also enhance degradation disproportionately. However, in the context of the CP, where proper folding is likely to occur due to the presence of the assembly core (Chapdelaine and Hohn, 1998), the precursor fragment clearly confers instability on the protein, thus corroborating conclusions drawn from the heterologous UPR system. A similar confirmation of the effect of the C-terminus on CP stability is revealed by Western analysis: although a CP antiserum readily detects a C-terminal truncated CP precursor carrying the stabilizing mutation in the N-terminal degradation signal [CP p(1–332), 3E▸A], the full-length precursor carrying the same mutation [p(1–489), 3E▸A] cannot be detected, showing that the C-terminus also confers instability in the CP context (unpublished results).
In CaMV-infected cells, most of the CP precursor is protected from degradation by the viral inclusion bodies (Kobayashi et al., 1998). Virus particles themselves are also very stable, although the proportion of p44 relative to the smaller CP variants decreases upon storage of isolated particles. Capsid assembly probably protects p44 from degradation, but still allows it to be processed to yield p39 and p37.
This raises the question of the instability signals’ role in the virus life cycle. Viral genomes carrying the stabilizing 3E▸A mutation were infectious, but the appearance of symptoms was clearly delayed, and the mutant partially reverted. Although it remains possible that other functions of CP are affected by this mutation, these data nevertheless suggest a requirement for this degradation signal, and a role for CP / CP precursor instability in the virus life cycle. Although the amino acid sequence of the termini of caulimovirus CP precursors is not conserved, all contain PEST-like domains at both ends.
Some evidence that the precursor sequences are important for the virus is that deletion of 76 amino acids from the N-terminus, or 21 amino acids from the C-terminus, of the CP precursor resulted in non-replicating virus in a protoplast replication system (K. Kobayashi, Friedrich-Miescher Institute, personal communication). A regulatory function for the C-terminal CP precursor sequences has been proposed (Guerra-Peraza et al. 2000), namely to prevent viral RNA binding to the basic domain and zinc finger motif, thereby guaranteeing its availability as mRNA until enough viral proteins (including the protease) are produced. We suggest that the N-terminal acidic domain may also have a regulatory function in the inhibition of nuclear targeting (Leclerc et al., 1999; unpublished results).
The degradation signals might ensure removal of the N- and C-terminal propeptides created by proteolytic processing of the CP precursor by the viral protease, to avoid their charge-mediated association with the mature CP. Furthermore, the instability signals may also be important for degradation of free forms of the CP precursor and CP in the infected plant (resulting from over-production or misfolding of the protein). These might interfere with the virus life cycle, and/or cause damage to the plant cell due to their nucleic acid-binding activity (Chapdelaine and Hohn, 1998; Daubert et al., 1982). Such effects may explain the toxicity of the CP precursor expressed in Escherichia coli (Fütterer et al., 1988). Using the data presented here, we can further characterize the proposed regulatory functions and precisely map the precursor degradation signals, after which mutations affecting only degradation can be made, allowing its function for the virus to be elucidated.
To construct the UPR cassette, the β-glucuronidase (GUS) gene was modified by PCR to remove the stop codon. The HA11-ubiquitin (K48R) fragment was amplified by PCR from pRc/dhaUbMβgal (Levy et al., 1996). The chloramphenicol acetyl transferase (CAT) gene was modified by PCR to remove an internal NcoI site. The UPR cassette was placed in plasmid pTAGNLS (Leclerc et al., 1999) replacing the HA11 epitope for transient expression in plant protoplasts, and in the mammalian expression vector pcDNA3.1 (Invitrogen, Carlsbad, CA, USA) for expression in HeLa cells. Fragments of ORF IV and mutants thereof were generated by PCR, using pCa37 as a template, which is the CAMV genome cloned in the SalI site of pBR322 (Lebeurier et al., 1980). These were cloned into pTAGNLS or in the UPR cassette. Details of PCR primers, PCR conditions and cloning steps are available upon request. To introduce the 3E▸A mutation into the viral genome, the KpnI–HpaI fragment of pCa37 was subcloned and mutated using the Quikchange method from Stratagene (La Jolla, CA, USA). The mutant fragment was cloned back into pCa37.
Transient expression in plant protoplasts and HeLa cells
Leaf mesophyll protoplasts of Nicotiana plumbaginifolia were transfected using polyethylene glycol (Goodall et al., 1990). DNA (10 or 20 µg) was added to 6 × 105 protoplasts. For the experiment shown in Figure 2 cells were harvested after 12 h incubation in the dark at 27°C, and for Figures 3 and 4 after 22 h incubation. HeLa cells were plated in six-well plates, transfected with 1 µg DNA per well using Fugene (Roche Molecular Biochemicals, Indianapolis, IN, USA), and harvested after approximately 30 h.
To both types of cells, 200 µl GUS extract buffer (Jefferson et al., 1987) was added and cells were lysed by three cycles of freeze–thawing (liquid nitrogen, −37°C). Samples were centrifuged for 10 min at 15.000 g and the supernatant used for the quantification of GUS activity and CAT ELISA levels.
Protoplast were transfected as above. Proteasome inhibitors were dissolved in water (lactacystin) or DMSO (others), and added to the protoplast culture 30 min after transfection. The final DMSO concentration in the cultures was 0.1%. We obtained MG132 from Peptides International (Louisville, USA), and lactacystin and clasto-lactacystin-β-lactone from Calbiochem (San Diego, USA). Protoplasts were harvested after 22 h and processed as above.
Detection of proteins
For Western blotting, proteins were isolated from protoplast extracts with phenol and precipitated with methanol (Hurkman and Tanaka, 1986). The proteins were separated on SDS–PAGE and electroblotted onto nitrocellulose at 80 V for 1 h. Blots were probed using HA11 monoclonal antibodies 16B12 (BAbCO, Richmond, CA, USA) or antibodies from the CAT ELISA kit (Roche Molecular Biochemicals). CAT and CAT fusion proteins were quantified using the CAT ELISA kit. GUS activities were assayed in 96-well microtitre plates in 200 µl reaction volumes containing 1 mm 4-methylumbelliferyl β-d-glucuronide (Sigma, St Louis, MO, USA) (Jefferson et al., 1987). 50 µl aliquots were taken after approximately 15 and 45 min, and the quantity of fluorescent product was recorded using a Titertek Fluoroskan II apparatus (Jefferson et al., 1990). The relative amounts of CAT and GUS were calculated relative to the average of the no-fusion control.
Inoculation of plants
Cloned virus was digested from the plasmid using SalI. Turnip plants (Brassica rapa var. Just Right) were grown from seed to the three- to four-leaf stage in a phytotron with 16 h light at 25°C. The youngest leaf was dusted with Celite, and 10 µg DNA in 20 µl H2O was gently applied. Plants were kept in the phytotron until symptoms developed. Virus was isolated from infected tissue (Gardner and Shepherd, 1980). Leaf sap of infected plants was used to infect new plants, and after three passages virus mutants were analysed by amplifying the DNA fragment containing the mutation and sequencing. The CP content of purified wild-type and mutant virus was analysed by SDS–PAGE and silver staining.
We thank M. Müller for the preparation of protoplasts, W. Krek for HeLa culture facilities, and F. Levy for providing the HA11-ubiquitin (K48R) and lysine free linker fragments. We also thank H. Rothnie, D. Kirk, K. Wiebauer, L. Stavalone and F. Levy for critical reading of the manuscript. This work was supported by the Novartis Research Foundation.