The role of CDC48 in the retro-translocation of non-ubiquitinated toxin substrates in plant cells

When the catalytic A subunits of the castor bean toxins ricin and Ricinus communis agglutinin (denoted as RTA and RCA A, respectively) are delivered into the endoplasmic reticulum (ER) of tobacco protoplasts, they become substrates for ER-associated protein degradation (ERAD). As such, these orphan polypeptides are retro-translocated to the cytosol, where a significant proportion of each protein is degraded by proteasomes. Here we begin to characterize the ERAD pathway in plant cells, showing that retro-translocation of these lysine-deficient glycoproteins requires the ATPase activity of cytosolic CDC48. Lysine polyubiquitination is not obligatory for this step. We also show that although RCA A is found in a mannose-untrimmed form prior to its retro-translocation, a significant proportion of newly synthesized RTA cycles via the Golgi and becomes modified by downstream glycosylation enzymes. Despite these differences, both proteins are similarly retro-translocated.


When the catalytic A subunits of the castor bean toxins ricin and Ricinus communis agglutinin (denoted as RTA and RCA A, respectively)
are delivered into the endoplasmic reticulum (ER) of tobacco protoplasts, they become substrates for ERassociated protein degradation (ERAD). As such, these orphan polypeptides are retrotranslocated to the cytosol, where a significant proportion of each protein is degraded by proteasomes. Here we begin to characterise the ERAD pathway in plant cells, showing that retro-translocation of these lysinedeficient glycoproteins requires the ATPase activity of cytosolic CDC48. Lysine polyubiquitination is not obligatory for this step. We also show that while RCA A is found in a mannose-untrimmed form prior to its retro-translocation, a significant proportion of newly synthesised RTA cycles via the Golgi and becomes modified by downstream glycosylation enzymes. Despite these differences, both proteins are similarly retrotranslocated.
As in mammalian and yeast cells, the plant cell endoplasmic reticulum 3 (ER) is a major protein folding compartment. Newly synthesised soluble and membrane proteins destined to remain in the ER, or to be transported to the Golgi complex, vacuoles or apoplast, are scrutinized by a stringent quality control system (1). Such surveillance ensures that newly synthesized proteins assume their native conformation and, where appropriate, assemble into oligomers. In mammalian and yeast cells, proteins that fail to fold and/or assemble correctly are usually retrotranslocated to the cytosol and degraded by cytosolic proteasomes, in a tightly coupled pathway referred to as ER-associated protein degradation (ERAD) (2,3). For the majority of misfolded or unassembled substrates, extraction from the ER membrane requires a cytosolic ubiquitin-interacting AAA-ATPase complex designated CDC48 in yeast (p97, also known as valosin-containing protein, in mammalian cells), in association with the adaptor proteins Ufd1p and Npl4p (4)(5)(6)(7)(8). This complex is believed to use its ATPase activity to segregate ubiquitinated ERAD substrates (7,9), and/or mechanically extract them from the ER membrane in readiness for delivery to proteasomes (5,8). However, not all ERAD substrates are modified by ubiquitin and extracted by this ATPase (8,(10)(11)(12), leading to the suggestion that an alternative driving force must exist. For pre-pro α-factor, this has been shown to be the ATPases located at the base of the 19S proteasome regulatory complex (10).
A pathway closely resembling ERAD also operates in plant cells but, by comparison, relatively little is known about substrate recognition, extraction from the ER membrane, or the pathway(s) and machineries for degradation (13). The Arabidopsis thaliana genome contains sequences for three CDC48-like proteins, one of which has been functionally shown to complement a yeast CDC48 mutant (14,15). This AAA-ATPase has been found to play a role in the quality control of a mutant 2 polytopic membrane protein, barley (Hordeum vulgare) powdery mildew resistance protein (MLO), but its involvement in ERAD is thus far limited to this one example (16). The process of retro-translocation, deglycosylation and proteasomal degradation in plant cells was first characterized by studying the biosynthesis of the catalytic A subunit (RTA) of the A-B plant toxin ricin (17). This orphan polypeptide, once delivered into the ER by its signal peptide, was shown to retro-translocate in a manner unrelated to its propensity for ubiquitination (18). While it was shown that ubiquitination of internal lysine residues is not a strict requirement for ER-tocytosol transport of this protein, the involvement or otherwise of CDC48 was not investigated.
Here we have studied the effects of a dominant negative form of Arabidopsis CDC48 on the export of the orphan catalytic polypeptides of ricin and its close relative, Ricinus communis agglutinin, in tobacco protoplasts. The data presented clearly demonstrate that the retrotranslocation of these proteins requires the participation of CDC48, and is irrespective of their glycosylation or ubiquitination status.
Tobacco leaf infiltration -The lower epidermis of 3 to 5 cm long tobacco leaves from a 3 to 4 week old plant were pressure-infiltrated with a culture of Agrobacterium tumefaciens (transformed with empty vector, pGreenII-0029, or with either wild type or mutant CDC48) diluted to an OD 600 of 0.1 in infiltration media (50mM MES pH 5.6, 0.5% (w/v) glucose, 2mM Na 3 PO 4 , 100µM acetosyringone (10mM stock in EtOH)). The plant was then incubated in greenhouse conditions for a further 3 days before preparation of protoplasts.
Protoplast fractionation -Protoplast pellets (from 500,000 cells) were resuspended in 140µl of 12% sucrose buffer (100mM Tris-HCl pH 7.6, 10mM KCl, 1mM EDTA, 12% (w/w) sucrose, supplemented immediately before use with Complete™ protease inhibitor cocktail (Roche Applied Science, Mannheim, Germany)) and homogenized by pipetting 50 times with a Gilson-type micropipette through a 200µl tip. Intact cells and debris were removed by centrifugation at 500×g for 5 minutes at 4°C. 130µl was loaded onto a 17% (w/w) sucrose pad and centrifuged at 100,000×g for 30 minutes at 4°C. Pellets (microsomes) and supernatants (soluble proteins) were frozen on dry ice and stored at -80°C.
Protease protection assay -Protoplast pellets (from 500,000 cells) were homogenized in 12% sucrose buffer as described above, this time omitting protease inhibitors, and cell debris was removed by centrifugation at 500×g for 5 minutes at 4°C. Supernatants were divided into three aliquots and incubated for 30 minutes at 25°C with either buffer (negative control) or proteinase K (5mg/ml stock in 50mM Tris-HCl pH 8.0, 1mM CaCl 2 ) at a final concentration of 75µg/ml in the presence or absence of 1% Triton X-100. Phenylmethylsulfonyl fluoride was added to a final concentration of 20mM to inhibit proteinase K before immunoprecipitation. Samples were frozen on dry ice and stored at -80°C.
Toxicity measurements -Triplicate aliquots of 330,000 protoplasts were co-transfected with a toxin-encoding plasmid or empty vector (pDHA), a CDC48-encoding plasmid or empty vector (pamPAT-MCS), and the phaseolin-encoding plasmid (pDHE-T343F). After 16 hours of recovery, protoplasts were pulse-labeled for 1 hour before being pelleted as described above. Polypeptides immunoselected from homogenates using anti-phaseolin antiserum were separated by SDS-PAGE, before fluorography and densitometry as before. Toxicity of the various constructs was expressed as a percentage of phaseolin synthesis with respect to protoplasts co-transfected with empty vector instead of toxin.
Confocal microscopy -Transfected protoplasts were mounted in K3 and imaged with a Leica TCS SP5 confocal laser-scanning microscope, using a 40× oil immersion objective lens (NA 1.25). YFP was excited at 514 nm and detected in the 525 to 583 nm range. RFP was excited at 561 nm and detected in the 592 to 635 nm range. Simultaneous detection of YFP or RFP was performed by combining the settings indicated above in the sequential scanning facility of the microscope, as instructed by the manufacturer.
Statistical analysis -Each experiment was repeated 3 or 4 times (see individual figures) and data groups were analysed using one-way 4 ANOVA. When a significant effect between sample groups was detected, the groups were compared using Tukey post-hoc tests. Statistical analysis was performed using SPSS v14.0.

Dominant negative CDC48QQ induces an upregulation of ER chaperones -
We rationalized that the presence of a dominant negative CDC48 would lead to the accumulation of aberrant proteins within the ER lumen as a prelude to an up-regulation of ER chaperones as part of the unfolded protein response (UPR). Such a phenotype would confirm the efficacy of the expressed CDC48 mutant. We therefore prepared protoplasts from sections of leaf tissue that had been subjected to Agrobacterium-mediated infiltration.
The transformed protoplasts expressed either empty vector, wild type CDC48, or a mutant CDC48 in which the conserved glutamate residues of the Walker B motifs (E308 and E581) of the two ATPase domains had been replaced by glutamine (denoted CDC48QQ) (16). After three days, the levels of BiP, calreticulin and GRP94 were increased in cells that were expressing CDC48QQ (Fig. 1A). The quantitation of these data is shown in Fig. 1B, where it can be seen that a statistically significant difference is observed in chaperone levels when CDC48QQ is expressed. It should be noted that the up-regulation observed is likely to be an underestimate since the agrobacterial infiltration method used in these experiments, although more efficient that other methods of plant cell transformation, is unlikely to be 100% efficient. ER chaperones will therefore not be induced in a proportion of the cells taken for analysis.

Expression of dominant negative CDC48
increases the stability of ER-sequestered RTA -We then followed the fate of ER-sequestered RTA in tobacco protoplasts 16 hours after PEGmediated co-transfection of plasmids encoding toxin together with either wild-type CDC48 or mutant CDC48, or in the presence of the glycosylation inhibitor (and UPR inducer) tunicamycin (31). To minimize inhibition of protein synthesis and thereby maximize the amount of newly made toxin that could be visualized, we used in these (and some of the subsequent) experiments a catalytic point mutant, RTA E177D , which has been shown to have virtually native structure (32) but a 70-fold reduced potency to ribosomes (24). As shown from the representative pulse-chase experiment in Fig. 2A, the glycosylated RTA made in a one hour pulse with radiolabeled cysteine and methionine was degraded with a half-life of approximately 3.5 hours in cells expressing wildtype CDC48. In cells expressing CDC48QQ, the rate of RTA degradation was significantly and reproducibly reduced as shown by quantification and statistical analysis of bands taken from four independent repeats of this experiment (Fig. 2B). Although the expression of CDC48QQ was in itself toxic to cells, such that transfected cells synthesized only 40% as much RTA as cells expressing wild-type CDC48, the RTA that was made was more stable (Fig. 2), with a half-life of greater than 5 hours (data not shown). The slightly increased gel mobility of the A chain bands during the chase ( Fig. 2A, lanes 6-8, 10-12) most likely represents the emergence of a mannose-trimmed species. Fig. 2B quantifies the kinetics of degradation averaged from four separate experiments. It is noticeable that, unlike CDC48QQ, tunicamycin treatment to block glycosylation and to promote a generalised UPR, does not impede the disappearance of RTA ( Fig.  2A, lanes 13-16).
Expression of mutant CDC48 hampers the retrotranslocation of toxin subunits -We have previously shown that a proportion of retrotranslocated RTA uncouples from the degradation pathway and refolds to inactivate ribosomes (18). A reduction in the protein biosynthetic capacity of these cells is therefore indicative of retro-translocation activity. To monitor protein synthesis in the transfected cell population alone we quantified, in triplicate, the levels of a reporter protein (phaseolin) encoded by a plasmid that was co-transfected (in a triple transfection) into protoplasts along with a plasmid encoding either a cytosolic RTA (cRTA), an ER-targeted RTA (RTA), or the non recombinant vector pDHA (no toxin) together with one of the CDC48 constructs (or vector alone (pamPAT-MCS)). Following overnight expression and a one hour pulse with radiolabeled cysteine and methionine, the triplicate phaseolin immunoprecipitates were quantified from gels. As shown in Fig. 3A, protein synthesis in cells making ER-targeted RTA (with and without the wild-type CDC48 construct) is reduced by ~60%. In contrast, protein synthesis in cells expressing RTA and the dominant negative CDC48QQ was reduced by only 30%, a statistically significant difference 5 (P<0.05). That such rescue was a general consequence of expressing CDC48QQ was ruled out from controls that show a consistent inhibition of protein synthesis (by more than 80%) when ricin A chain was deliberately expressed in the cytosol without a signal peptide (cRTA), either in the presence or absence of wild type or mutant CDC48 (Fig. 3A). The significant rescue of protein synthesis shown in Fig. 3 therefore occurred only when RTA was initially targeted into the ER lumen in the presence of CDC48QQ. This would suggest that CDC48QQ mitigates the toxic effect of RTA by impeding its retrotranslocation to the cytosol. Since this experiment relies on efficient co-transfection, we checked the ability of protoplasts to take up multiple plasmids. Fig. 3B is a representative set of images showing that protoplasts competent to take up a YFP expressing plasmid in all cases concomitantly coexpress an RFP plasmid.
To confirm toxin retro-translocation, we determined the location of the stabilized RTA. Cells expressing ER-targeted RTA E177D alone or together with either wild-type or mutant CDC48 were pulsed with [ 35 S]-labeled amino acids before being fractionated into membranes (M) and cytosol (C). As shown in Fig. 4A (lane 4), a significant amount of the RTA E177D synthesized was recovered in the cytosol fraction under these conditions. By contrast, virtually no toxin accumulated in the cytosol of cells expressing CDC48QQ (Fig.  4A, lane 8). BiP immunoprecipitates from the same samples indicate the integrity of the membrane fractions and show that the cytosolic RTA was unlikely to result from membrane leakage during cell lysis and membrane preparation. It is noticeable that the form accumulating in the cytosol was equivalent to the slowest migrating species observed in the membrane fractions. This higher RTA band is reminiscent of the mannoseuntrimmed species seen in previous pulse-chase experiments (18). Regardless of the precise glycan forms however, it is clear that coexpression of mutant CDC48QQ impaired the retro-translocation of RTA to the cytosol.
To investigate whether RTA was retained inside the membranes we carried out a protease protection experiment using cell extracts containing ER microsomes (25). Accordingly, at the 3 hour chase point, when RTA was largely sensitive to proteinase K under normal conditions (Fig. 4B, lane 5), a proportion of this protein was clearly protected from protease when made in the presence of CDC48QQ (Fig. 4B, lane 11, and quantified from three independent repeats in Fig.  4C). Based on the statistical analysis, there is significantly more protease-protected RTA in the cells expressing CDC48QQ (Fig. 4C). The faster migrating species, visualized upon incubation of the samples with protease, most likely represents a membrane protected remnant of the fraction of RTA that was in the process of being retrotranslocated at the time of cell lysis and that may have been partially exposed to the cytosol. Consistent with this, we observed the complete disappearance of this fragment when the membranes were treated with protease concomitant with membrane solubilization (Fig.  4B, lanes 3, 6, 9 and 12).
Toxins can be retro-translocated in mannosetrimmed and untrimmed forms, but only when CDC48 is active -We next used cells expressing either ER-targeted RTA E177D or a comparable inactive version of its relative, Ricinus communis agglutinin (RCA A E176D (33)), and analyzed the presence or absence of these orphan proteins in the cytosol and membrane fractions over time. It was noticeable that RTA, but not RCA A, underwent a slight shift in size during pulsechase experiments (Fig. 5, top panel). This slight downsizing of RTA can be blocked using the ER mannosidase inhibitor 1-deoxymannojirimycin (DMM) (Fig. 5, second panel) to generate a species that is now comparable with the sharper bands observed for RCA A. Clearly, the appearance of RTA in the cytosol does not critically depend on prior mannose trimming events. Taken together, these observations suggest that the core glycan of RCA A may not normally be a substrate for extensive mannosidase action, and that the retrotranslocation machinery does not differentiate between the different glycosylated forms presented by RTA and RCA A.
As observed before for RTA (17,18), there is a qualitative loss of both toxin chains from membrane fractions with time (Fig. 5, top two panels). That this is due to retro-translocation and cytosolic degradation is indicated from experiments in which proteasome activity was compromised using clasto-lactacystin β-lactone (Fig 5, third panel). We have previously observed that this conventional inhibitor of mammalian proteasomes does not completely block RTA degradation. Nevertheless, under these conditions, 6 the cytosolic turnover of both toxin A chains was sufficiently hindered to permit the more rapid appearance of a faster migrating intermediate (Fig. 5, third panel, asterisk) that was absent in non transfected controls. This species of RTA was previously shown to be a deglycosylation product of cytosolic peptide N:glycanase that becomes more noticeable when proteasomes are partially inhibited (17). Strikingly, when DMM was used (to maintain both toxins as tight homogenous bands) together with the proteasome inhibitor and CDC48QQ, neither A chain was observed in cytosolic fractions (Fig. 5, compare cytosol fractions in the top three panels with panel four). The bottom panel reveals the distribution of ER BiP to indicate the integrity of the membrane fractions. Clearly, the effect of CDC48QQ was to severely compromise emergence of both Ricinus toxins into the cytosol.
Cycling via the Golgi can preface retrotranslocation -RTA and RCA A, though very closely related, appear to be handled differently upon their synthesis in the plant cell. Both polypeptides become core-glycosylated at a single site (N 10 ) upon entering the ER lumen, and such glycans can be removed by treatment with endoglycosidase H. Although we know that RTA, in the absence of its normal partner B chain, is neither secreted nor delivered to vacuoles in tobacco cells (21), a significant proportion of this newly made protein was found to be Endo Hresistant (Fig. 6A, top panel, lanes 4-6). This supports a cycling of RTA from ER-to-Golgi-to-ER, and an encounter with Golgi glycanmodification enzymes. By contrast, RCA A does not acquire Endo H-resistance under identical conditions (Fig. 6A, lower panel, lanes 4-6) suggesting that cycling of these soluble proteins in the early secretory pathway is not obligatory for their retro-translocation. In the presence of CDC48QQ, a greater proportion of the total population of newly made RTA became Endo H resistant (Fig. 6A, top panel, compare lanes 4-6 with lanes 10-12), presumably because cycling continues even when retro-translocation is perturbed. Indeed, when retro-translocation is inhibited, a proportion of RCA A also reveals Endo H resistance (Fig. 6A, lower panel, lanes  11-12). The quantification of the proportion of Endo H-resistant species at the 5 h chase points from three independent experiments are shown in Fig. 6B. They reveal a statistically significant difference between the fraction of Endo H-resistant RTA and RCA A chains, and also between RCA A made in the absence or presence of CDC48QQ. A possible rationalisation for the latter is that when retrotranslocation is compromised, RCA A has greater opportunity to be cycled via the Golgi and to acquire Endo Hresistance. Note that both RTA and RCA A appear to be degraded at comparable rates during the chase when wild type CDC48 is present, indicating that retro-translocation is independent of glycan structure. This was confirmed for RTA where both the Endo H-sensitive and Endo Hresistant forms were observed in the cytosolic fraction (Fig. 6C).
CDC48-mediated retro-translocation of RTA does not require polyubiquitination of lysine residues -The CDC48/p97 complex is believed to function in retro-translocation by interacting with polyubiquitin chains on exposed client proteins and mechanically extracting them from the ER membrane. Studies in mammalian cells with the non-ubiquitinated but nevertheless still retro-translocated substrate cholera toxin A1 chain (12) have shown that cytosolic p97 was not required. A similar situation may exist for RTA, since only the degradation of this protein rather than its retrotranslocation was affected by a modulation of lysine content (18). It is evident that a lysyl-free toxin variant (RTA E177D 0K) is able to retro-translocate when endogenous or endogenous plus over-expressed wild type CDC48 are present (Fig. 7, lanes 7 to 12 and 13 to 18, respectively). However, when the mutant ATPase is present, the dislocation of the lysinefree toxin to the cytosol is completely prevented (Fig. 7, lanes 20, 22 and 24).

DISCUSSION
To test for a dominant negative effect of CDC48, we monitored the up-regulation of ER chaperones that follows the inhibition of retrotranslocation and that forms part of the unfolded protein response. Figure 1 reveals a significant induction of chaperones in cells expressing CDC48QQ, but not in cells that make the wild type protein, and confirms, in plant cells, an effect of the mutant ATPase that is well known in other systems (12,34,35).
The process of ERAD typically begins with the recognition of a substrate as being misfolded or orphan. The structural features recognized are not known in most cases, but for newly made RTA 7 (and its close relative RCA A) it is likely that, in the absence of cognate B chain(s), exposed hydrophobic stretches of amino acids normally masked in the holotoxin (36) may trigger chaperone or lipid interactions (37) to prevent aggregation. Persistent absence of the partner B chain appears, however, to lead to a redirection of the catalytic A subunit to the retrotranslocation machinery (17), which is likely to be a protein-conducting channel in the ER membrane (38). Once exposed on the cytosolic face of the ER membrane, most ERAD substrates become oligoubiquitinated on lysyl residues in a process requiring specific membrane-associated ubiquitin conjugating enzymes and ligases (e.g. the Doa10p and Hrd1p complexes in yeast (39)(40)(41)(42)(43)). This is usually followed by an ATPdependent extraction process, during which the ubiquitin-binding CDC48 (p97 in mammals) and its cofactors Ufd1p and Npl4p are recruited to the membrane by Ubx2 (44). The substrate is usually then polyubiquitinated by E3 enzymes such as Ufd2 (45) or the N-glycan-specific SCF Fbs1 and SCF Fbs2 complexes (46,47), followed by deglycosylation where appropriate (48), and passage on to proteasomes for destruction.
In comparison to the extensive characterization of mammalian and yeast ERAD, the molecular and mechanistic details of this process in plant cells remains very poorly understood. Earlier identification of plant ERAD substrates, either misfolded (16,49) or orphan as here (17,18), now opens the way to identifying the key players and requirements in this system. It should be noted that the behaviour of newly synthesized ricin A chain in the ER of plant cells reflects, in outline at least, the situation observed when disulfide bonded A-B ricin holotoxin is endocytosed into the mammalian ER, whereupon reduction sparks the retro-translocation of the catalytic subunit. In both scenarios (biosynthesis of RTA in the plant cell or its liberation in the ER of mammalian cells following holotoxin endocytosis and reduction), most retro-translocated RTA appears to be locked in a tightly coupled pathway leading to proteasomal degradation (50). However, a fraction of RTA is known to uncouple from this pathway, and to refold and intoxicate either the ultra-sensitive mammalian ribosomes or the more recalcitrant plant ribosomes (51). Such degradation avoidance is known to be linked to the scarcity of lysyl residues in the toxin moiety that prevents this protein becoming efficiently ubiquitinated (52).
The observation that most retro-translocated substrates are ubiquitinated and pulled from the ER membrane by virtue of this modification begs the question, what happens to proteins that are not ubiquitinated (10)(11)(12)? It has been shown for non-ubiquitinated pre-pro α-factor that ATPases in the 19S proteasome regulatory complex can provide an alternative driving force (10,53) for membrane extraction. It is also evident from functional analyses that p97 is not required for the retro-translocation of ubiquitin-free cholera toxin A1 chain in mammalian cells (12). The driving force in this latter case is not known with certainty. Such findings have led to the suggestion that the p97 (CDC48)-Ufd1-Npl4 complex may operate only for substrates that are modified by ubiquitin (12). However, it is clear that there are other instances where CDC48 is known to be dispensable (54) or to play a more ancillary than obligatory role (55). Here, we have shown that CDC48 does participate in the retrotranslocation of the orphan toxin subunits (RTA and RCA A; Fig. 5), even when completely lacking in lysine residues (as with RTA 0K; Fig.  7). Assuming that atypical N-terminal ubiquitination is not occurring in this instance, this finding would suggest that the ATPase can facilitate the extraction of a non-ubiquitinated protein. It remains to be seen whether or not the requirement for CDC48 involves direct interaction with the toxin polypeptides.
In the presence of CDC48QQ, glycosylated RTA is partially stabilized (Fig. 2), is considerably less toxic to ribosomes (Fig. 3), and is localized to the membrane fraction (Fig. 4). Likewise, RCA A is similarly stabilised and retained in a membrane fraction when CDC48 function is impaired (Fig.  5). That CDC48QQ was responsible for these effects rather than the UPR itself is shown in Figure 2. From this, it is clear that induction of the UPR by a different means (blocking glycosylation by treatment of cells with tunicamycin, thus increasing the levels of misfolded proteins) does not result in a stabilization of RTA.
It is also noticeable that the two toxin subunits are processed differently in preparation for dislocation. In the case of RTA, its emergence into the cytosol appears to involve two speciesan initial mannose-untrimmed version and, at later chase points, a mannose-trimmed version (Fig. 5, top panel). The mannose-trimmed version 8 appearing at later chase times may well represent the fraction of RTA that has undergone cycling via the Golgi (Fig. 6A), since only at such a time point is an Endo H resistant form present in the cytosol fraction (Fig. 6C). This species would need to have undergone mannose trimming and anterograde export from the ER to reach the enzymes that render the toxin glycan resistant to Endo H. By contrast, RCA A is not mannosetrimmed or cycled via the Golgi (Figs. 5 and 6). Despite these differences, both polypeptides are retro-translocated with similar kinetics in a CDC48-dependent manner. The reason for such a striking difference between two proteins that are 94% identical at the primary sequence level and that possess an identical glycosylation site is not clear. It may be connected with the fact that RCA A, unlike RTA, is normally required to assemble into an A2-B2 tetramer in which disulfide bonds exist between each A and B chain and between the two A chains. This additional burden may lead to prolonged interactions and scrutiny by ER resident oxidoreductases that prevent the entry of this protein into an ER-to-Golgi cycling pathway. The cycling observed with RTA is reminiscent of some soluble substrates in yeast that rely on transport to the ER from the Golgi (56-58).
It is evident from the limited number of studies that have been carried out that an ERAD-like protein disposal pathway does indeed operate in plant cells. However, the present dearth of known substrates means that many details and variations are yet to emerge, not least the range of cytosolic proteins dedicated to the extraction, modification and targeting of ER-sequestered substrates to the plant proteasome, and possibly to other cytosolic proteases. The finding that a CDC48 complex is required for the natural plant substrates described to date (16) suggests it is likely that the plant homologues of other mammalian and yeast ERAD components, found in Arabidopsis (59) and in other plants (60), may well have a functional role in this disposal pathway. In Arabidopsis, over 1300 genes (5% of the proteome) are devoted to the selective breakdown of proteins by the extensive ubiquitin/26S proteasome pathway (61), although virtually nothing at present is known about these components in relation to ERAD. It should also be noted that since plant cells have other proteolytic repositories, in the form of lytic vacuoles (62), where proteins can be degraded (63,64), the impact of this ERAD pathway to overall protein degradation, including the turnover of conformationally regulated proteins that may have critical roles in photomorphogenesis, pathogen defense, and hormonal regulation in the plant cell, remains to be determined. BiP immunoprecipitates were exposed to film for 2 days while the GRP94 and calreticulin immunopreciptates were exposed for 10 days.