Transient ER retention as stress response: Conformational repair of heat damaged proteins to secretion-competent structures

Mechanisms ensuring survival after severe heat stress have evolved in all organisms. Prokaryotes and eukaryotes acquire tolerance towards brief exposure to otherwise lethal temperatures, if they are preconditioned at a moderately elevated temperature where heat shock genes are activated (37°C for yeast) (Lindquist and Craig, 1988). Though the primary lesions caused by heat are unknown, denaturation of proteins must be fundamentally important, and survival may depend on the ability of cells to refold heat-damaged vital proteins. Indeed, the Hsp70 homolog DnaK, together with DnaJ and GrpE, is capable of repairing heat-damaged proteins in the cytosol of Escherichia coli (Skowyra et al., 1990; Gaitanaris et al., 1990; Schröder et al., 1993; Ziemienowicz et al., 1993; Georgopoulos and Welch, 1993). Moreover, Hsp104 can refold and solubilize heat-aggregated proteins in the yeast cytosol, and is essential for acquisition of thermotolerance (Sanchez and Lindquist, 1990; Parsell et al., 1994). Hsp78, an Hsp104 homolog of the yeast mitochondrial matrix, functions in restoration of heat-inactivated mitochondrial protein synthesis (Schmitt et al., 1996).

We have focused on stress physiology of the secretory compartment. We found recently a novel stress-related chaperone activity in the yeast ER, namely refolding of a completely translocated and folded, and thereafter heat- denatured reporter enzyme (Jämsä et al., 1995b; Saris et al., 1997). In our experimental system a secretory β-lactamase fusion protein, Hsp150Δ-β-lactamase, was accumulated in the ER by incubating temperature-sensitive mutants (sec18 or sec23) at the restrictive temperature 37°C, which results in arrest of membrane traffic from the ER to the Golgi (Kaiser and Schekman, 1990). The bacterial β-lactamase portion folded to an enzymatically active conformation in the yeast ER by the aid of the Hsp150?-carrier polypeptide, an N-terminal signal peptide-containing fragment of the natural secretory yeast protein Hsp150 (Russo et al., 1992; Simonen et al., 1994; Jämsä et al., 1995a). The cells were then exposed to 50°C, which resulted in loss of the β-lactamase activity. After shift back to physiological temperature, 24°C, the β-lactamase activity was slowly resumed in an ATP-dependent process (Jämsä et al., 1995b) (see scheme of temperature treatments in Fig. 1A). The Hsp70 homolog Lhs1p was required for stabilization and reactivation of the heat-inactivated reporter enzyme (Saris et al., 1997). Lhs1p, also named Ssi1p and Cer1p, shares 24% of identical amino acids with the other Hsp70 chaperone of the ER, BiP/Kar2p (Rasmussen, 1994; Baxter et al., 1996; Craven et al., 1996; Hamilton and Flynn, 1996). Here we study the fate of the secretory apparatus and ER-located natural and recombinant cargo proteins after severe heat stress. We found that thermal insult inactivated transiently the secretory machinery, and had dramatic but reversible effects on newly synthesised ER-located glycoproteins. Though the secretory capacity was soon reactivated, heat- affected glycoproteins were immobilized in the ER for different periods of time, and gained secretion competence with protein-specific kinetics. Lhs1p was required for resumption of secretion competence, but not for reactivation of the secretory apparatus.

Yeast strains H4 (Mata sec18-1 ura3-52 trp1-289 leu2-3,112), H393 (Mata sec18-1 ura3-52 trp1-289 leu2-3,112 URA3::HSP150Δ-β- lactamase) and H621 (Mata sec18-1 trp1-1 ade2-1 his3 ura3-52 lhs1::TRP1 URA3::HSP150Δ-β-lactamase) were grown at 24°C in shaking flasks overnight to early logarithmic phase in YPD medium, or in synthetic complete (SC) medium lacking methionine and cysteine.

Metabolic labeling of cells (25×10⁶/500 µl) was with 20 µCi of [³⁵S]methionine/cysteine (1,000 Ci/mmol, Amersham, UK) in SC medium lacking methionine and cysteine. The labelings were terminated with NaN₃, and the culture medium and cell lysate samples were immunoprecipitated as described (Saris et al., 1997). Briefly, cells were lysed mechanically with glass beads in NET-buffer (0.05 M Tris-HCl, pH 8.0, containing 0.4 M NaCl, 5 mM EDTA, 1% NP- 40 and 100 units/ml of aprotinin) in the presence of 2 mM phenylmethylsulfonyl fluoride (PMSF). The lysates were boiled and pre-cleared for 1 hour at 4°C with Protein A-Sepharose (Pharmacia). Immunoprecipitation was with anti-β-lactamase antiserum (1:100) or anti-CPY antiserum (1:100) and Protein A-Sepharose for 2 hours at 4°C. After washing with diluted NET-buffer (1:1), wash buffer (0.1 M Tris-HCl, pH 7.5, containing 0.2 M NaCl, 2 M urea and 0.5% Tween-20), and with 0.1% SDS, the precipitates were analyzed by SDS-PAGE.

Duplicate cell samples (25×10⁶/0.5 ml of YPD medium) were incubated at indicated temperatures, and the cell-associated and secreted β-lactamase activity was assayed using nitrocefin as a substrate (Simonen et al., 1994). Invertase synthesis was induced in YPD medium containing 0.1% glucose, and intracellular and cell wall bound invertase activity of duplicate samples was measured (Jämsä et al., 1994).

Cell samples were lysed in native conditions as for the determination of β-lactamase activity, and subjected to electrophoresis in 5% gels in the absence of SDS and reducing agents. The gel was then incubated in 0.1 M sodium acetate buffer, pH 5.1, containing 0.1 M sucrose, for 10-60 minutes in a 30°C waterbath with gentle shaking. The buffer was removed and the gel was rinsed twice with distilled water and incubated on a hot cooking plate in preheated 2,3,5-triphenyl tetrazolium chloride (1 mg/ml in 0.5 M NaOH). Staining was stopped by adding distilled water, and the gel was fixed with 10% acetic acid and dried.

Cell viability was assayed by incubating duplicate cell samples (3×10⁶/ml) in YPD medium in Wassermann tubes successively at 37°C and 48°C or 50°C. After cooling on ice, the cells were diluted and plated on YPD medium for 3-4 days at 24°C (Sanchez and Lindquist, 1990). Glucose consumption was assayed by suspending the cells after the thermal treatments in fresh YPD medium. During the 24°C incubation, duplicate samples were removed for the determination of the glucose concentration of the medium using the Gluco-quant kit (Boehringer Mannheim). CHX and NaN₃ were from Sigma and used in final concentrations of 100 µg/ml and 10 mM, respectively. SDS-PAGE was in 8% reducing gels.

We reported earlier that Hsp150Δ-β-lactamase, preaccumulated in the ER at 37°C, was inactivated and aggregated during exposure to 50°C. After shift to 24°C, it was solubilized and reactivated, but resided still in the ER after 6 hours (Saris et al., 1997), though it is normally secreted rapidly to the medium (Simonen et al., 1994). Thus, either the structural features of the refolded molecules were not compatible with exit from the ER after 6 hours of recovery, or the secretory machinery did not yet function. To study this, we first established whether secretion of the heat-affected molecules could be resumed later (see Fig. 1A). H393 cells (sec18) were labeled with [³⁵S]methionine/cysteine for 1 hour at 37°C. The culture medium was separated from the cells, which were lysed, and both preparations were subjected to immunoprecipitation with anti-β-lactamase antiserum. SDS- PAGE analysis revealed Hsp150Δ-β-lactamase of 110 kDa in the cell lysate (Fig. 1B, lane 2), whereas no protein could be detected in the medium (lane 1). The 110 kDa protein is the ER-specific form of the protein, which carries primary Oglycans, i.e. single mannose residues on multiple serine and threonine residues of the Hsp150Δ-fragment (Simonen et al., 1994; Jämsä et al., 1994, 1995b). After labeling, a parallel cell sample received cycloheximide (CHX) to stop further protein synthesis, and was incubated at 24°C for 2 hours to relieve the sec18 block. Hsp150Δ-β-lactamase, migrating like a 145 kDa protein, was now detected in the culture medium (lane 3), and only a little protein remained in the cells (lane 4). Mature secretory Hsp150Δ-β-lactamase migrates like a 145 kDa protein due to its O-glycans, extended up to pentamannosides in the Golgi, and lacks an N-terminal 54 amino acid propeptide, which is cleaved in the Golgi at a Kex2 protease recognition site (Russo et al., 1992; Simonen et al., 1994; Jämsä et al., 1995a). Another set of parallel samples similarly ³⁵S-labeled at 37°C were incubated for 20 minutes at 50°C (lanes 5-14). One sample was removed immediately after the thermal insult. The 110 kDa form was detected in the cells lysate (lane 6), and no protein was in the medium (lane 5). The rest of the samples were pelleted and resuspended in chase medium containing excess unlabeled methionine and cysteine, and incubated at 24°C for 4 (lanes 7 and 8) or 8 hours (lanes 9 and 10). Even after 8 hours, the protein still resided in the cells in the ER-specific form (lane 10) and no protein was detected in the medium (lane 9). However, when the recovery period at 24°C was extended to 16 hours, most of the protein was detected in the culture medium (lane 11), some of it remaining cell-associated (lane 12). The secreted protein comigrated with mature secretory Hsp150Δ-β-lactamase (compare lanes 11 and 3). Since membrane traffic requires metabolic energy, one sample received sodium azide for the 16 hour incubation at 24°C, to serve as a negative control. Under these conditions transfer of Hsp150Δ-β-lactamase to the medium was inhibited (lane 13). The cell-associated protein (lane 14) migrated more slowly than the native ER- accumulated molecules (lane 2), apparently due to increased O-glycosylation, resulting perhaps from escape of molecules to the Golgi during the long incubation. The cells did not multiply during the experiment, as cell division started only after 20 hours at 24°C. Most cells were viable, since 80% of them formed colonies in 3-4 days when plated at 24°C after the heat treatments. Thus, secretion of Hsp150Δ-β-lactamase molecules denatured by a 20 minute thermal insult at 50°C could be resumed, but extremely slowly.

Next we studied whether a less severe thermal insult would facilitate resumption of secretion of the reporter protein. To this end, the above experiment was repeated performing the thermal insult for 15 minutes at 48°C (Fig. 1C). This treatment abolished about 80% of the ER-accumulated β-lactamase activity (see Fig. 2). After this treatment, the 110 kDa form again persisted in the cells (Fig. 1C, lane 6), and no reporter protein was detected in the medium (lane 5). When the cells were returned to 24°C, after 2 hours about half of the reporter protein was detected in mature form in the medium (lane 7). The rest was in the cell lysate in the ER-specific form (110 kDa) (lane 8). After 4 hours, most of the protein was in the medium (lane 9), very little remaining in the cells (lane 10). After 8 hours of recovery, all Hsp150Δ-β-lactamase persisted in the culture medium (lane 11), and no protein was detected in the cells (lane 12). Sodium azide in the recovery mixture prevented secretion (lanes 13 and 14). When CHX was present during the 24°C incubation, Hsp150Δ-β-lactamase was secreted as in the absence of CHX (not shown). Lanes 1-4, Fig. 1C, are the same controls as in Fig. 1B. We conclude that less severe denaturation conditions resulted in considerable acceleration of resumption of secretion of Hsp150Δ-β- lactamase, and that resumption of secretion did not require de novo synthesised proteins.

Next we studied resumption of secretion of heat-denatured Hsp150Δ-β-lactamase in the absence of Lhs1p. H621 cells (secl8 Ihsl) were labeled at 37°C as above, resulting in accumulation ofHsp150Δ-β-lactamase in the cells (Fig. 1D, lane 2), and no protein was secreted to the medium under these conditions (lane 1). When the cells were shifted directly to 24°C with CHX, Hsp150Δ-β-lactamase appeared in the medium in 2 hours (lane 3), very little remaining in the cells (lane 4). When cells labeled at 37°C were treated for 15 minutes at 48°C, the reporter protein still could be detected in the cells (lane 6), and not in the medium (lane 5). When the cells were after the 48°C treatment chased at 24°C, very little of the reporter protein appeared in the medium in 2.5-7.5 hours (lanes 7, 9 and 11). During this period, the intracellular 110 kDa form was degraded (lanes 8, 10 and 12). The thermal insult left about 20% of Hsp150Δ-β-lactamase enzymatically active (Fig. 2), and it apparently was the fraction that appeared in the culture medium in the absence of Lhs1p. Resumption of secretion of the reporter enzyme appeared thus to require functional Lhs1p. The H621 cells (secl8 lhs1) were viable for at least 6 hours of recovery, since they consumed glucose during this time as vigorously as H393 (secl8 LHS1) cells (see Materials and Methods). Moreover, they are capable of protein synthesis at least up to 5.5 hours of recovery at 24°C (Saris et al., 1997). However, most of the cells die later. In the absence of Lhs1p, only 5% of the cells were able to form colonies after 3-4 days at 24°C, showing that Lhs1p is required for efficient acquisition of thermotolerance (Saris et al., 1997).

The above data did not allow us to distinguish whether ER retention of heat-denatured Hsp150Δ-β-lactamase was due to structural distortion, which was recognized by the quality control machinery, or to transient inactivation of the secretory machinery. Thus we examined next whether the secretory machinery was affected by severe heat stress. For reference, we first established the level of synthesis and secretion of native Hsp150Δ-β-lactamase at 24°C in H393 cells (secl8) in the absence of any thermal treatments. In 4 hours, 0.38 unit/ml of β-lactamase activity was secreted to the culture medium (Fig. 3A, open circles), while 0.12 units/ml remained intracellular (closed circles). Hsp150Δ-β-lactamase was then collected in the ER of H393 cells at 37°C, and denatured for 20 minutes at 50°C (Fig. 3B, closed circles; bar on abscissa symbolizes thermal insult). The cells were then incubated at 24°C in the absence of CHX to allow de novo synthesis of Hsp150Δ-β-lactamase, simultaneously with reactivation of the pre-accumulated heat-inactivated molecules (Fig. 3B, circles). After 6 hours, 0.5 units/ml of activity could be detected inside of the cells (closed circles) and 0.24 units/ml in the medium (open circles). Since the total activity (0.74 units/ml) was more than the initially accumulated activity (0.52 units/ml), part of it must have been due to Hsp150Δ-β-lactamase molecules that were newly synthesized at 24°C. The HSPl50Δ-β-lactamase gene was under the control of the HSPl50 promoter, which is activated upon shift of the cells from 24°C to 37°C, resulting in higher expression level at 37°C as compared to 24°C (Russo et al., 1993).

To differentiate reactivation of Hsp150Δ-β-lactamase from de novo synthesised molecules, parallel cells incubated at 37°C and then at 50°C were shifted to 24°C in the presence of CHX (Fig. 3B, squares). Now, 0.4 units/ml was recovered inside of the cells (closed squares) and 0.055 units/ml in the culture medium (open squares). The total amount synthesized de novo between 2 and 6 hours of recovery was 0.285 units/ml, 65% of which was secreted. The secretory apparatus appeared thus to function at least after 2 hours of recovery. Hsp150Δ-β- lactamase, ³⁵S-labeled and secreted after 5 hours of recovery at 24°C, co-migrated in SDS-PAGE with the native molecules synthesized under normal conditions at 24°C (not shown). This indicates that O-glycosylation and processing by Kex2p functioned apparently normally in the recovering cells. For reference, Hsp150Δ-β-lactamase was pre-accumulated in the ER at 37°C in H393 cells (secl8), followed by addition of CHX and shift to 24°C to alleviate the secl8 block in the absence of any thermal insult (Fig. 3C). More than 80% of the intracellular activity (closed circles) was detected in the culture medium (open circles) within 90 minutes at 24°C, indicating that the secretory machinery was functional. Thus, ER retention of reactivated Hsp150Δ-β-lactamase was either due to structural anomalities not affecting enzymatic activity, or to localization of the molecules in a subcompartment of the ER, which had been damaged by heat and was unable to support membrane traffic. To decide between these alternatives, we extended our studies to authentic heat-affected yeast proteins.

Secretory invertase is a natural enzyme of S. cerevisiae, which is synthesized only in low glucose medium (0.1%). It is normally targeted to the cell wall, where it remains intercalated amongst other cell wall components. Invertase synthesis was derepressed by incubating H4 cells (secl8) in low glucose medium for 1 hour at 37°C. The activity accumulated in the ER (Fig. 4A, panel a, black circles), and very little appeared in the cell wall (open circles). The cells were then incubated for 20 minutes at 50°C (bar on abscissa). Though the thermal insult did not inactivate invertase (closed circles), its structure was distorted (see below). The glucose concentration of the medium was then reconstituted to 4% to inhibit further invertase synthesis, and the cells were shifted to 24°C for recovery (arrowhead). During the first hour at 24°C, the invertase activity remained intracellular. Then, the level of intracellular invertase decreased to 36% (closed circles) with simultaneous appearance of 64% of the pre-accumulated activity (average of 5 experiments was 60%; 56-66%) in the cell wall during the next 4 hours (open circles). No more activity appeared in the cell wall even after a prolonged incubation overnight, during which time the activity persisted inside of the cells and in the cell wall. The transfer of the activity to the cell wall occured by vesicular traffic, because it was inhibited by NaN3 (squares). CHX had no effect on secretion of invertase (not shown). 72% of the heat-treated H4 cells formed colonies at 24°C. When the 50°C treatment was omitted, pre-accumulated invertase was secreted to the cell wall at 24°C in less than 1 hour (Fig. 4A, panel b).

Next we studied expression and secretion of native invertase molecules synthesized during the recovery period. Secl8 cells (H4) were treated under repressing conditions (2% glucose) successively at 37°C and 50°C, and then shifted to 24°C. After various periods of time, invertase synthesis was derepressed for 90 minutes, and intracellular (Fig. 4B, panel a, stippled columns) and cell wall (whole columns) invertase activities were determined. After shift of the cells from 50°C to 24°C, very little invertase was synthesized during the first hours. Then, synthesis was slowly resumed, and reached in 6 hours a similar level as at 24°C under normal conditions in the absence of heat treatments (Fig. 4B, panel b). As soon as invertase was synthesized, it appeared in the cell wall. We then studied the kinetics of invertase secretion after 4 hours at 24°C more closely. Secl8 cells (H4) were subjected to the 37°C and 50°C treatments and incubated for 4 hours at 24°C under repressing conditions. The glucose concentration of the medium was lowered to 0.1% to derepress invertase synthesis and the cells continued for an hour at 24°C. Samples were withdrawn after various time periods for determination of cell wall bound (Fig. 4C, panel a, open circles) and intracellular (closed circles) activity. For reference, panel b shows a control, where invertase synthesis was derepressed under normal conditions at 24°C. In both the recovering cells and untreated cells, invertase was secreted similarly to the cell wall. Thus, secretion of heatdamaged invertase was resumed much faster than that of Hsp150Δ-β-lactamase. Invertase and Hsp150Δ-β-lactamase were unlikely to reside in different ER subcompartments recovering with different kinetics from heat-damage. Thus, we suggest that the varying rates of resumption of secretion of the two proteins were due to different kinetics of refolding to forms, which were approved by the quality control machinery for ER exit.

Though the enzymatic activity of invertase was not affected, we found that 20 minutes at 50°C distorted the molecules, whereas no indication for damage at 48°C was found. Invertase acquires 9-12 primary N-glycans in the ER, which under normal conditions are extensively and heterogeneously extended in the Golgi. Invertase activity blocked in the ER in secl8 cells (H4) at 37°C migrated in native gel electrophoresis mainly as a compact band, due to primary N-glycans (Fig. 5, lane 1). When parallel cells were incubated for 20 minutes at 50°C and then for 6 hours at 24°C, the activity still migrated similarly (lane 2), though two thirds of the molecules had reached the cell wall (see Fig. 4A, panel a). This indicates that the heat-affected molecules could not undergo N-glycan extension in the Golgi. When the thermal insult was for 10 minutes at 48°C, followed by 6 hours at 24°C, invertase migrated much more slowly and heterogenously, demonstrating extensive and heterogenous N-glycosylation (lane 3). When the cells were shifted directly from 37°C to 24°C for 2 hours, similar heterogenous glycosylation was observed (lane 4). Invertase molecules synthesized at 24°C after the 50°C treatment were extended normally, indicating that the glycosylation apparatus in the Golgi was functional (not shown).

Pro-CPY (67 kDa) is normally transported from the ER via the Golgi to the vacuole, where it is proteolytically processed to mature CPY (61 kDa). Electrophoretic migration thus indicates the location of the molecules (Stevens et al., 1982). To study the fate of heat-affected pro-CPY, secl8 cells (strain H4) were ³⁵S-labeled at 37°C (Fig. 6A, lanes 1-6). One sample was removed immediately (lane 2), and another was shifted directly to 24°C and incubated for 1 hour with CHX to serve as a positive control for vacuolar transport (lane 1). Immunoprecipitation with anti-CPY antiserum and SDS-PAGE analysis showed that after labeling at 37°C all CPY was in the ER in the pro-CPY form of 67 kDa (lane 2). After chase at 24°C, more than half of pro-CPY had reached the vacuole as it occured as mature CPY of 61 kDa (lane 1). Parallel cell samples, labeled at 37°C as above, were incubated then for 20 minutes at 50°C (lanes 3-6), continuing then in chase medium at 24°C for 3 (lane 3), 6 (lane 4), 9 (lane 5) or 12 hours (lane 6). In each sample only pro-CPY could be detected, suggesting ER retention. Even after 16 hours at 24°C, pro-CPY persisted unprocessed and undegraded (not shown).

To confirm that lack of maturation of pro-CPY reflected the location of the molecules in the ER, and was not due to inability of the vacuolar proteases to process it to mature CPY, cells were incubated successively in the absence of radiolabel at 37°C, 50°C and for 4 hours at 24°C, followed by ³⁵Slabeling at 24°C for 1 hour. Only mature CPY could be immunoprecipitated (Fig. 6A, lane 7). This demonstrated that the vacuolar proteolytic processing machinery was functional at least after a 4 hour recovery period, and that native pro-CPY molecules synthesized de novo at 24°C by-passed the heat- affected pro-CPY molecules in the ER and reached the correct destination. When the thermal insult was for 20 minutes at 48°C instead of 50°C, pro-CPY remained in the ER at least for 9 hours (not shown). However, when the heat treatment was only for 10 minutes at 48°C, more than half of the pro-CPY was in the vacuole after 3 hours of recovery at 24°C (Fig. 6B, lane 3), and most of it after 6 hours (lanes 4 and 5). Lack of metabolic energy stopped the transport of pro-CPY from the ER to the vacuole, since in the presence of NaN3 pro-CPY persisted (lane 6). Heat-affected pro-CPY thus resumed transport more slowly than Hsp150Δ-β-lactamase and invertase, and was perhaps more severely damaged as concerns the features required for secretion.

We then performed the experiment using strain H621 which lacks the LHSl gene (secl8 lhsl). Pro-CPY was again accumulated in the ER during ³⁵S-labeling at 37°C (Fig. 6C, lane 2). After 20 minutes at 50°C (lane 3), and after 2 hours of recovery at 24°C, pro-CPY remained in the ER (lane 4). However, after 4-6 hours at 24°C (lanes 5 and 6), most pro- CPY had disappeared, but no mature CPY could be detected. In the absence of Lhs1p, transport of de novo synthesized pro- CPY appeared normal. When H621 cells labeled at 37°C were shifted for 2 hours to 24°C in the absence of thermal insult, only mature CPY was detected (lane 1), and it persisted undegraded at least for 6 hours (not shown). Lhs1p was thus required for stability of heat-affected pro-CPY.

Completed whole genome sequencing projects have predicted a large number of proteins, more than one third of the proteome in the case of S. cerevisiae, whose functions are unknown. It has been proposed that a proportion of the orphan genes encode proteins required for tolerance towards natural stress conditions that laboratory strains have not been challenged with (Oliver, 1996). Here we studied the effect of heat stress on the physiology of the secretory compartment of yeast cells. We found that ER retention and subsequent conformational repair of heat-affected proteins is part of the cellular stress response.

Vacuolar CPY, cell wall invertase, and secretory recombinant Hsp150Δ-β-lactamase were accumulated in the ER of a reversible ts-mutant (sec18), followed by a brief treatment at 48-50°C after pre-conditioning at 37°C (see Fig. 1A for experimental design). When the cells were returned to 24°C, the proteins remained in the ER for different times, and were thereafter transported to their proper destinations with widely varying kinetics. After 20 minutes at 50°C, most of the invertase had reached the cell wall in 5 hours, whereas Hsp150Δ-β-lactamase was in the culture medium after approximately 16 hours, and pro-CPY remained irreversibly in the ER. When the thermal insult was less severe, for 10 and 15 minutes at 48°C, Hsp150Δ-β-lactamase reached the medium in about 4 hours and pro-CPY the vacuole in 6 hours, respectively. Severe heat stress resulted in intracellular retention of most newly synthesized ³H-mannose labeled cell wall mannans, more than half of which reached the cell wall during recovery (data not shown). Since transport of the reporter proteins was resumed with different rates, we asume that they were distorted to varying extents by the heat treatments. Thereafter they were refolded with different kinetics to structures that were accepted by the quality control machinery for ER exit. Under normal conditions, immature proteins are retained in the ER by the quality control machinery for characteristic times depending on their different folding and assembly rates (Lodish et al., 1983; Braakman et al., 1991; Ou et al., 1993; Hammond and Helenius, 1994). The large variety of heat-affected yeast proteins transiently retained in the ER of living cells shows that heat-denaturation and subsequent refolding are physiologically relevant phenomena.

No degradation of heat-affected proteins could be detected even in the case of irreversible ER retention, unless Lhs1p was lacking. In contrast, misfolded de novo synthesized proteins and unassembled monomers are often degraded in normal mammalian cells (Klausner and Sitia, 1990; Bonifacino and Lippincott-Schwartz, 1991; Knittler et al., 1995). How Lhs1p contributes to the stability of the heat-affected proteins remains to be studied. BiP/Kar2p is involved in export of mutated pro- CPY from the ER to the cytoplasm for proteosomal degradation (Plemper et al., 1997).

Reactivation of heat-inactivated Hsp150Δ-β-lactamase was not sufficient for secretion competence, which was resumed much later than biological activity. On the other hand, invertase was not inactivated by the thermal insult, though it was distorted since its N-glycans were not extended in the Golgi as they would be normally. The quality control machinery thus appeared to recognize structural features, which were independent of biological activity. Structural features thought to lead to ER retention of newly synthesized immature proteins include exposed hydrophobic sequences, partially trimmed Nglycans, aggregation or unassembly of monomers, and free disulfides (Flynn et al., 1991; Blond-Elguindi et al., 1993; Reddy et al., 1996; Braakman et al., 1991; Jämsä et al., 1994; Hurtley and Helenius, 1989; Hammond et al., 1994). Exposure of hydrophobic sequences by heat-denaturation is likely, but N-glycans had no role in retention since Hsp150Δ-β-lactamase lacks them. Neither was aggregation critical, since Hsp150Δ- β-lactamase aggregates were solubilized long before resumption of secretion.

We found that exit of proteins from the ER was inhibited for an hour after thermal insult at 48-50°C, whereafter exocytosis was reactivated independently of de novo protein synthesis. Native reporter molecules, synthesized some hours after the cells had been returned to 24°C, were folded correctly and rapidly exited the ER as in untreated cells, while their heatdamaged counterparts were undergoing conformational repair. Since ER retention and refolding of a large amount of heatdamaged proteins did not disturb de novo folding, the two chaperoning activities may have distinct features. Moreover, Lhs1p was required for resumption of secretion competence of our heat-damaged reporter proteins, but not for their de novo folding. We showed earlier that Lhs1p was associated with heat-denatured reporter proteins, whereas no binding to the native counterparts could be observed (Saris et al., 1997). The best known mammalian ER chaperone BiP and its yeast homolog BiP/Kar2p are required for conformational maturation of many de novo synthesized proteins (Dorner et al., 1992; Knittler and Haas, 1992; Gething and Sambrook, 1992; Pittman et al., 1994; Hendershot et al., 1996; Simons et al., 1995). In yeast BiP/Kar2p is generally needed for ER translocation (Lyman and Schekman, 1995), whereas Lhs1p is needed under certain conditions for efficient ER translocation of a subset of polypeptides (Baxter et al., 1996; Craven et al., 1996; Hamilton and Flynn, 1996). The role of BiP/Kar2p in conformational repair of heat-denatured proteins is not yet known. Unlike BiP/Kar2p which is essential for life, Lhs1p is dispensable, except under severe heat stress (Saris et al., 1997). Like Hsp104, Lhs1p may be a genuine stress protein specialized in repair functions after heat stress.

We acknowledge the Academy of Finland for support (38017), and thank Dr Leevi Kääriäinen for critical reading of the manuscrpt, Ms Anna Liisa Nyfors for excellent assistance and Ms Helena Vihinen, MSc. Chem. Eng., for help with computers. M.M. is a Biocentrum Helsinki fellow.