Apoprotein B degradation is promoted by the molecular chaperones hsp90 and hsp70

Apoprotein B (apoB) is the major protein of liver-derived atherogenic lipoproteins. The net production of apoB can be regulated by presecretory degradation mediated by the ubiquitin-proteasome pathway and cytosolic hsp70. To further explore the mechanisms of apoB degradation, we have established a cell-free system in which degradation can be faithfully recapitulated. Human apoB48 synthesized in vitro was translocated into microsomes, glycosylated, and ubiquitinylated. Subsequent incubation with rat hepatic cytosol led to proteasome-mediated degradation. To explore whether hsp90 is required for apoB degradation, geldanamycin (GA) was added during the degradation assay. GA increased the recovery of microsomal apoB48 approximately 3-fold and disrupted the interaction between hsp90 and apoB48. Confirming the hsp90 effect in the cell-free system, we also found that transfection of hsp90 cDNA into rat hepatoma cells enhanced apoB48 degradation. Finally, apoB48 degradation was reconstituted in vitro using cytosol prepared from wild type yeast. Notably, degradation was attenuated when apoB48-containing microsomes were incubated with cytosol supplemented with GA or with cytosol prepared from yeast strains with mutations in the homologues of mammalian hsp70 and hsp90. Overall, our data suggest that hsp90 facilitates the interaction between endoplasmic reticulum-associated apoB and components of the proteasomal pathway, perhaps in cooperation with hsp70.

Introduction mixture was overlaid onto 300µl of 2.3 M sucrose in MSB in a centrifuge tube. Solutions of 1.5 M sucrose (600µl) and 0.25 M sucrose (500µl) in MSB were overlaid and the discontinuous gradient was centrifuged at 100 000 x g for 5 h at 4°C, after which 150µl aliquots were successively removed from the top of the gradient. Proteins in the fractions were resolved by SDS-PAGE and the radiolabeled apoB48 was detected by phosphorimager analysis of a dried gel.
ApoB48 degradation assay -At the end of the transcription-translation procedure, the reaction mixture was layered onto SH buffer (0.25M sucrose, 5mM HEPES, pH 7.4) and centrifuged at 100 000 x g for 30min at 4°C. The pelleted microsomal membranes were resuspended in 50% (vol/vol) of the appropriate lysate and 50% of 2xPh buffer (40mM HEPES pH 7.4, 220mM KCl, 10mM MgCl 2 ). At the following final concentrations, 2mM ATP, 10mM creatine phosphate, and 100µg/ml of creatine kinase were added to the complete mixture. Aliquots of the mixture were incubated for 2 h either on ice (control) or at 37°C. At the end of the incubation period, gel loading buffer for SDS-PAGE was added directly to the samples. In some experiments, the proteasome inhibitors (dissolved in DMSO) indicated in the Results were added to the degradation mixture just prior to the control and depleted lysates were then used for the degradation assay. To examine the dependence on the ubiquitinylation of apoB48 for degradation in RRL, just prior to its use in the degradation assay, RRL was supplemented with ubiquitin aldehyde (ubal), an inhibitor of deubiquitinylation (22), at a concentration range of 0-20 µM.
Effect of geldanomycin (GA) on apoB48 degradation -At the end of the transcriptiontranslation procedure, the reaction mixture was layered onto SH buffer and centrifuged at 100 000 x g for 30 min at 4°C. The pelleted microsomal membranes were resuspended in 50% (vol/vol) of the rat hepatic lysate and 50% of 2xPh buffer. At the following final concentrations, 2mM ATP, 10mM creatine phosphate, and 100µg/ml of creatine kinase were added to the complete mixture. The mixture was halved and GA (National Cancer Institute, Bethesda, MD) dissolved in DMSO was added to one aliquot at a final concentration of 30µM; an equal volume of DMSO was added to the other aliquot.
Samples were taken every 30 min during a 2 h incubation (37°C) period. At the end of the incubation, gel loading buffer for SDS-PAGE was added directly to the samples and the apoB48 content analyzed as above.

Results
Biogenesis of apoB48 in vitro -For our studies, we have used the apoB48 species which is expressed in the intestine of all mammals and in the rodent liver. Though shorter than apoB100 (expressed only in mammalian liver), apoB48 still displays in cell culture studies many similar properties with respect to translocation, membrane integration and degradation (23). ApoB48 (MW ~240kD) was synthesized from its cDNA by in vitro transcription/translation in rabbit reticulocyte lysate (RRL) ( apoB48. This interpretation is consistent with the data reported by Rusinol and colleagues (24). In the presence of TritonX100 (used to destroy membrane integrity), the protein represented by the upper band was digested by trypsin ( Figure 2A, lane 3). Thus, the protease resistance in the absence of Triton X100 resulted from proper integration of apoB48 and not from protein aggregation or inadequate protease activity.
To investigate whether the upper band in Figure 1 represented apoB48 that was modified by glycosylation, microsomal associated apoB48 was separated from the reaction mixture by centrifugation and treated with N-glycosidase F (PNGF). The effect of PNGF treatment is shown in Figure 2B. Note that PNGF reduced the apparent molecular weight of the upper band to that of the lower band (lane 2 vs. 3). For comparison, the total reaction (lane 1) and supernatant after pelleting the microsomes (lane 4) are also displayed. This reduction of apparent molecular weight of the upper band by PNGF treatment not only demonstrates modification of apoB48 by glycosylation, an event also known to occur with apoB in intact cells (25), but also confirms that in the presence of DPM, there was translocation of apoB48. Treatment with sodium carbonate at pH 11.5 (26), however, released less than 5% of apoB48 from the microsomal membrane (data not shown).
To confirm that apoB48 was associated with the microsomal membrane, an in vitro transcription/translation reaction in the presence of DPM was performed, after which the microsomes were floated through a sucrose gradient as previously described (21). As shown in Figure 3, we found that the majority of apoB48 had migrated into fractions of lower density than that at which it was loaded onto the gradient (denoted by the arrow), indicative of it being membrane associated.
Together, the sodium carbonate and sucrose gradient data imply that translocated apoB48 strongly interacts with membrane lipid or protein components, consistent with previous results using isolated microsomes or intact cells (5,6,27).
ApoB48 is ubiquitinylated and degraded in the cell-free system -Recent studies have shown that degradation of apoB in HepG2 cells is mediated by the ubiquitin-proteasome pathway (e.g., (4,28,29)). To determine whether nascent apoB48 was ubiquitinylated in vitro, HA-tagged ubiquitin was added at the end of the transcription/translation procedure and the incubation was continued at 37°C for 1 h in the presence of ATP and an ATP regenerating system ( Figure 4). Ubiquitinylated apoB48 was immunoprecipitated with anti-HA-tag antibody (Figure 4, lane 4). That the high molecular weight material represents bona fide modification of apoB by ubiquitinylation was supported by 3 results: 1) if the antibody to HA tag was omitted, no labeled material was detected ( Figure 4 Overall, the data in Figures 1-4 demonstrate that the early events during apoB biogenesis are faithfully reproduced in the cell-free system, and suggest that this represents a valid model for the study of the mechanisms leading to apoB degradation. Towards this goal, microsomal-associated apoB48 was studied in the next series of experiments because we have previously shown in HepG2 cells that apoB destined for proteasomal degradation remains stably associated with microsomes (4)(5)(6). Thus, at the end of transcription/translation, microsomes were pelleted and resuspended in RRL or rat hepatic cytosol. The latter cytosol was utilized to study apoB48 degradation because it derives from the tissue in which it is normally expressed.
Aliquots of each reaction mixture were incubated on ice (control condition) and at 37°C for 2 h in the presence of ATP and an ATP regeneration system. Independent of the source of the cytosol, after 2 h, the majority (~75%) of apoB48 was degraded compared to the control ( their results, they hypothesized that RRL contains a factor or factors that interfere with the actions of some proteasome inhibitors. Because of this limitation, to support the interpretation that in RRL, apoB48 degradation is accomplished by the ubiquitinproteasomal pathway and not by another protease, we performed degradation assays in which RRL was supplemented with ubiquitin aldehyde (Ubal), an inhibitor of deubiquitinating enzymes. Ubal has previously been reported to decrease apoB degradation in semi-permeabilized HepG2 cells (32) and exerts a negative effect on proteasomal degradation by interfering with the removal of ubiquitin chains from substrates prior to their entry into the proteasome (e.g., see (22)). The addition of Ubal to RRL resulted in a dose-dependent increase in the recovery of apoB48 at the end of the 2 h degradation assay, from 50% to 80% and 93% at 0, 5 and 20 µM of Ubal, respectively.
Overall, the results from these and the prior (31) studies imply that the degradation of apoB48 in RRL is accomplished by the ubiquitin-proteasome pathway, but the response of proteasomes in RRL to some inhibitors may be anomalous.
In contrast, the proteasome inhibitors effectively inhibited degradation in rat hepatic cytosol ( Figure 5, panels B and D). In addition to the increased recovery of apoB48 in the presence of these inhibitors, further evidence for a role of the ubiquitinproteasome pathway was suggested by a reduction in apoB48 degradation when the cytosol was pre-treated with the ATP-depleting agent apyrase (data not shown).
Consistent with the rat hepatic lysate studies, similar results were obtained using human hepatic cytosol (data not shown).
Hsp90 is required for apoB degradation -As reviewed in the Introduction, hsp90 has been shown to either promote or inhibit the interaction of substrate proteins with components of the ubiquitin-proteasome pathway. In light of this controversy, and our finding that in the cell-free system apoB48 can be ubiqutinylated and degraded by proteasomes ( Figures 4 and 5), we next explored whether hsp90 plays a role in apoB degradation.
Cytosolic hsp90 has been shown to be the specific target for the ansamycin benzoquinone antibiotics (33,34), particularly geldanamycin (GA). Therefore, GA, the most potent drug from this class antibiotics (35), was added to the degradation mixture and apoB recovery was monitored every 30 min during the subsequent 2 h incubation.
The results shown in Figure 6 indicate that the recovery of apoB48 was increased over 3fold in the presence of GA. It has been previously shown with purified proteasomes, GA did not reduce the degradation of the model substrate SLLVT-AMC (36), suggesting that GA did not function simply as a proteasome inhibitor in the cell-free degradation assay, but directly interfered with a pro-degradative effect of hsp90 on apoB48.
The homologue of hsp90 in the ER, grp94, which also binds the same class of antibiotics (37), has been shown to interact with apoB100 in HepG2 cells (38). In that study, grp94 was found to interact with apoB100 upon a reduction of the ATP level by apyrase treatment. Note that under our conditions, ATP and an ATP regeneration system are added (Experimental Procedures), making it unlikely that GA's effect was mediated by disrupting an apoB48-grp94 interaction. Nonetheless, to show that the GA effect was specific for cytosolic hsp90, we determined whether either chaperone interacts with apoB48. Using antibodies to either hsp90 or grp94, we attempted to coimmunoprecipitate apoB48 and the results are presented in Figure 7. As shown, apoB48 could be co-immunoprecitpitated with hsp90, but not with grp94 ( Figure  These cells express primarily the apoB100 form of native apoB, the recovery of which we have shown to be increased by lactacystin treatment (39). We took this transfection approach rather than treating the cells with GA because treatment of cells with ansamycin antibiotics not only directly affects protein-chaperone interactions, but increases the level of multiple heat shock proteins by promoting the activation of heat shock transcription factor (HSF1) (40,41). For example, treatment of yeast with ansamycin antibiotics inhibits ERAD due to a secondary increase in BiP concentration (J.L. Brodsky, unpublished data).
After transfection either with pcDNA3 vector (control condition) or pcDNA3-hsp90ß plasmid, cells were pulse-labeled with [ 35 S]-methionine/cysteine for 15 min and chased in isotope-free medium for 30 and 90 min. Cell lysates and conditioned media were then analyzed by immunoprecipitation with antibodies to apoB and rat albumin (as a control). Successful transfection of the cells and expression of the plasmids were confirmed by RT-PCR detection of the plasmid region encoding neomycin antibiotic resistance (data not shown).
As shown in Figure 9 (panel B), transfection with the hsp90 cDNA did not affect albumin synthesis or secretion. In contrast, the total recovery of apoB100 from lysate and medium was lower at the 30 min (21% reduction) and 90 min (48% reduction) chase time points in the cells transfected with the hsp90 cDNA ( Figure 9, panel A). These results are consistent with the data from the cell-free system, and support a pro-degradative role for hsp90 in apoB degradation in the liver.
In addition to studying the effects of increasing hsp90 on apoB degradation in hepatic cells, it would have been desirable to also determine the consequences of decreasing the level of a chaperone. Perhaps because of the abundance and relatively long half-lives of many chaperones, it has been difficult to achieve major decreases in intact cells. For example, a ribozyme approach undertaken independently by two laboratories resulted in no more than a 25-30% reduction in cellular grp94 levels (42,43).
Thus, we tried to immunodeplete hsp90 from rat hepatic lysate (Experimental Procedures) and to use the depleted lysate in the cell-free degradation assay. Compared by guest on March 24, 2020 http://www.jbc.org/ Downloaded from to control rat hepatic lysate, there was a 4-fold increase in apoB recovery from the reaction mixtures containing lysate that had been depleted of ~75% of its hsp90. This increase in apoB recovery is again consistent with the transfection results and the other data from the cell-free system that hsp90 is a pro-degradative factor for apoB.
Degradation of microsomal apoB48 in yeast cytosols -The availability of specific hsp70 and hsp90 chaperone mutants in yeast (18)(19)(20)44) prompted us to explore whether yeast cytosol would support apoB48 degradation. Cytosols from wild type yeast or from isogenic strains with either a temperature-sensitive mutation in hsp82 (the yeast homologue of mammalian hsp90) and deleted for the gene for the constitutive hsc90 homologue, hsc82, or a temperature-sensitive mutation in the hsp70 homologue ssa1 (ssa1-45) were prepared from cells shifted for 1 h to the non-permissive temperature.
These cytosols were then used in the degradation assay (Experimental Procedures).
As shown in Figure 10  Note that GA (lane 3) blocked this degradation. The degradation of apo48 in yeast cytosols was also reversed by proteasome inhibitors (data not shown). Thus, the characteristics of apoB48 degradation in the presence of yeast cytosols were quite similar to those in the presence of rat hepatic cytosol (Figures 5 and 6).
The experiments were then repeated using the cytosols from the mutant strains.
The results clearly show that apoB48 is almost completely protected from degradation when the hsp82 or ssa1 mutant cytosols were used ( Figure 10, panels B, C, and E), similar to the effect of GA on wild type cytosol.
It is known that hsp90 acts in association with several different co-chaperones, which bind to hsp90 and organize it into discrete subcomplexes (12). In particular, hsp70 has been found in such a complex, associated with hsp90 through the Hop protein (45).
We previously found that hsp70 promotes apoB degradation (4). Thus, the protection of apoB48 from degradation when only one of these chaperones is mutated suggested that hsp70 and hsp90 may have cooperative effects on apoB degradation. To test this hypothesis, cytosols from the two mutant yeast strains were mixed at a 1:1 ratio and used in the degradation assay. The degradation of apoB48 was restored to wild type levels, and under these conditions, apoB48 was protected when GA was added to the reaction

Discussion
The majority of apoB is subjected to proteasome-mediated ERAD in cells of hepatic origin that are either deprived of exogenous fatty acids (which stimulate lipid synthesis) or deficient for microsomal triglyceride transfer (MTP) activity (4,28,29). It is assumed that the association of apoB with lipid-ligands is required for the achievement of its native conformation, and when the concentration or transfer of lipids is reduced, apoB is detected by quality control mechanisms that prevent the exit of malfolded proteins from the ER (46).
That chaperones may have important functions in apoB degradation stems from the growing recognition that these molecules influence the ERAD process of other proteins (for a review, see (9)) and from our demonstration that increasing the expression of hsp70 in fatty acid-deprived human liver-derived HepG2 cells promoted the proteasomal degradation of apoB (4,47). Further hints that hsp70 and other chaperones may be involved in the quality control of apoB come from Linnik and Herscovitz (38), who showed in HepG2 cells that BiP/Grp78, calreticulin, Erp72 and grp94 coimmunoprecipitate with apoB , and from Olofsson and colleagues, who have found recently that BiP, PDI, calreticulin and grp94 are associated with apoB-containing lipoprotein particles in rat hepatoma cells (48).
Because of the limited ability in intact hepatic cells to modulate the interactions of chaperones without affecting their steady state levels, we developed a cell-free degradation assay system in order to best explore the roles of these proteins in apoB ERAD. The previous report that apoB48, synthesized in a coupled transcription-translation system, was translocated into dog pancreatic microsomes (24) suggested the feasibility of this goal. Under such conditions, there is neither MTP activity nor active lipid synthesis, thereby mimicking the lipid-ligand deficient state in hepatic-derived cells deprived of fatty acids or lipid transfer activity. In the present studies, we have extended the previous results by showing that the nascent protein was modified by glycosylation and ATP-dependent ubiquitinylation, both known to occur under native conditions (4,25,28,49). By isolating the microsome-associated apoB48 and demonstrating its proteasome-mediated degradation after resuspension in hepatic cytosol, we have established a cell-free system in which ERAD of apoB can be reconstituted. As with other mammalian and yeast ERAD substrates studied in similar systems (e.g., (8,50,51)), this now allows us to decipher the molecular requirements for the degradation of apoB by the ubiquitin-proteasome pathway.
As noted in the Introduction, we were interested in studying the potential influence of hsp90 on apoB degradation, given its abundance in mammalian cells and its controversial role as a protective or susceptibility factor in the proteasomal degradation of other proteins (14)(15)(16). Multiple lines of evidence implicate hsp90 as a pro-degradative factor for apoB. First, there was increased recovery of apoB in the cell-free degradation assay when GA was added to the rat hepatic cytosol ( Figure 6). GA is widely considered to be a specific inhibitor of protein interactions with members of the hsp90 family (e.g., (34,37)) and in fact, was used in two of studies cited above (14,15).
Further support that GA was operating through a specific hsp90-mediated mechanism was our finding that it disrupted the interaction between hsp90 and apoB48 by guest on March 24, 2020 http://www.jbc.org/ Downloaded from (Figure 7) without perturbing the total pool of hsp90. Importantly, in spite of the interaction of apoB and grp94 in HepG2 cells (38), there was no evidence for this in the cell-free system, consistent with the suggestion that grp94 plays a role in the maturation of lipoproteins containing apoB escaping ERAD (48).
The pro-degradative effect of hsp90 on apoB in the cell-free assay was also found in rat hepatoma cells, McArdle-RH7777, transfected with hsp90 cDNA (Figure 9). This cell line is a standard model of hepatic mammalian lipoprotein metabolism and also exhibits ERAD of apoB (39,52). Although the cells produced predominately apoB100, the small amount of apoB48 in the control cells was also reduced in the hsp90 cDNAtransfected cells (data not shown). These effects on cellular apoB were not attributable to a non-specific effect of hsp90 or the transfection procedure, as albumin recoveries from the cell and medium were not significantly changed and the control cells were subjected to the same transfection protocol. Overall, these results imply that the cell-free system is an accurate reflection of the role of hsp90 in intact cells of hepatic origin.
The availability of yeast strains with mutations in the homologues of mammalian hsp90 and 70, hsp82 and ssa1, respectively, has allowed us to confirm and to examine further the roles of these chaperones in apoB degradation (Figure 10). The suitability of this system was first evidenced by the fact that the degradation of microsome-associated apoB48 in the presence of the wild type yeast cytosol had characteristics resembling those observed with rat hepatic cytosol; i.e., >60% of apoB48 was degraded, which was reversed by the proteasome inhibitor (data not shown) or GA. Second, consistent with the results summarized here and our previous studies implicating a pro-degradative effect by guest on March 24, 2020 http://www.jbc.org/ of hsp70 (4,47), cytosols from the strains with either chaperone mutated did not support significant degradation. This implied that both chaperones are required for the proteasomal degradation of apoB. This requirement was specific, as supported by our finding that degradation was reconstituted in the complementation (mixing) experiment.
Other studies have shown that hsp90 and hsp70 cooperate. For example, both chaperones participate in the activation of hormone binding by glucocorticoid receptor (53).
Multichaperone complexes containing hsp90 and hsp70 have also been shown to mediate refolding of denatured luciferase and β-galactosidase in a cooperative manner (54,55).
How do we envision hsp90 and hsp70 to function in the ERAD pathway for apoB? Based on our recent studies (5,6,47,56), we find no evidence for a complete retrotranslocation into the cytosol of apoB destined for proteasomal degradation, as for MHC Class I molecules in CMV-infected cells (57) or mutant pre-pro alpha factor (8); rather, there appears to be an extraction process in which factors interact with apoB domains as they appear on the cytosolic surface of the ER membrane. It has been proposed that hsp70 (58) can facilitate such an extraction, perhaps by binding to a domain and serving as a molecular motor or a ratchet. This would be consistent with our finding that increased expression of hsp70 decreased the secretion of apoB-containing lipoproteins even when lipid synthesis was stimulated in HepG2 cells (4), suggesting a competition between the cytosolic factors targeting apoB to ERAD and the ER lumenal factors promoting lipoprotein assembly and exit from the ER.
Increased expression of hsp70 enhances apoB ubiquitinylation (4,47). In contrast, GA did not change apoB ubiquitinylation (Figure 8), implying that the hsp90 acted after ubiquitin tagging of apoB48. Because hsp90 has been shown to be associated with the proteasome, it has been proposed recently that it may facilitate the unfolding of substrates into the relatively narrow mouth of the 19S cap of the proteasome (59,60). In this scenario, we envision hsp70 and hsp90 working sequentially -with hsp70 being an early participant as an "extraction" and, perhaps, an ubiquitinylation factor, and hsp90 as a factor acting later to facilitate the association with and entry into the proteasome of the ubiquitinylated apoB. Thus, for maximal degradation both chaperones should be expected to be required for apoB degradation. Because of the length of apoB (2152 and 4536 amino acids for apoB48 and apoB100, respectively) and the nature of an extraction process, however, it is likely that at steady state in the hepatic cell, both chaperones are bound to apoB targeted for ERAD, as we have recently observed in HepG2 cells ((6) and R. Pariyarath and E. Fisher, unpublished results). The participation of both chaperones in apoB degradation is consistent also with the recent finding that hsp70 and hsp90 have roles in the folding and ubiquitinylation of wild type and ∆F508 CFTR in mammalian cells (15) and that hsp70 facilitates the degradation of CFTR in yeast (21).
In summary, we have established a powerful and convenient cell-free system to study the ERAD pathway for apoB. With this system, and confirmed in transfection studies, we have now obtained clear evidence that both hsp90 and hsp70 promote apoB ERAD. The effects of these two chaperones and other cytosolic factors on this process can now be studied by using purified factors or cytosols from yeast strains with relevant mutations, singly or in combination, to allow the biochemical characterization and the kinetic ordering of events related to the targeting of apoB to the proteasome.