Inhibition of Translocation of (cid:1) -Lactamase into the Yeast Endoplasmic Reticulum by Covalently Bound Benzylpenicillin*

We found recently that (cid:1) -lactamase folds in the yeast cytosol to a native-like, catalytically active, and trypsin-resistant conformation, and is thereafter translocated into the ER and secreted to the medium. Previously, it was thought that pre-folded proteins cannot be translocated. Here we have studied in living yeast cells whether (cid:1) -lactamase, a tight globule in authentic form, must be unfolded for ER translocation. A (cid:1) -lactamase mutant (E166A) binds irreversibly benzylpenicillin via Ser 70 in the active site. We fused E166A to the C terminus of a yeast-derived polypeptide having a post-translational signal peptide. In the presence of benzylpenicillin, the E166A fusion protein was not translocated into the endoplasmic reticulum, whereas translocation of the unmutated variant was not affected. The benzylpenicillin-bound protein adhered to the endoplasmic reticulum membrane, where it prevented translocation of BiP, carboxypeptidase Y, and secretory proteins. Although the 321-amino acid-long N-terminal fusion partner adopts no regular secondary structure and should have no constraints for pore penetration, the benzylpenicillin-bound protein remained fully exposed to the cytosol, maintaining its signal peptide. Our data suggest that the (cid:1) -lactamase portion must unfold for translocation, (cid:3) -AGCTCCG- GTGCCCAACGAT), and 82629 (5 (cid:3) -GCAACCAAGCTTGAGTAAACTT-GGTCTGACAG) for the E166A mutant, and oligonucleotides MS2, 92397 (5 (cid:3) -CAGCTCCGGGTCCCAACGA), 92398 (5 (cid:3) -TCGTTGGGAC-CCGGAGCTG), and 82629 for the E166D mutant. The mutated frag- ments were digested with Kpn I- Hin dIII and cloned into plasmid pKTH4539 to replace the wild type gene. The plasmids were named pKTH4825 (E166D) and pKTH4826 (E166A) and the mutations verified by sequencing. The Bam HI fragments containing the mutated HSP150 (cid:1) - (cid:2) -lactamase -ADC1 terminator cassettes were cloned into plasmid pFL34, designated pKTH4828 (E166D) and pKTH4830 (E166A), and into plasmid pFL26, designated pKTH4829 (E166A). The mutant (cid:2) -lactamase gene (E166A) with a His 6 tag was created by PCR plasmid pKTH4829 template, MS2 C2500 (5 (cid:3) -* plasmid pKTH4956 HI fragment -lactamaseE166A- -ADC1 terminator was into in pKTH4960, to lack- ing PCR-amplified as and 82629 (cid:3) GGCCTATGCTCCATCTGAGCC) and Hin dIII (Promega) to plasmid pKTH4700 containing promotor terminator to produce plasmid pKTH4716. I- Nhe I fragment of pKTH4716 with and terminator sequences was into pFL26 to plasmid to Nhe I- Kpn I

Depending on the hydrophobicity of the signal peptide, newly synthesized polypeptides are translocated into the yeast endoplasmic reticulum (ER) 1 either during translation or after completion of translation and release from the ribosomes (1). Post-translational and co-translational translocation occur through heterotrimeric Sec61p complexes, which are composed of subunits Sec61p, Sbh1p, and Sss1p. Sec61p spans the ER membrane 10 times and forms an aqueous pore. In addition to the heterotrimeric translocon complex, posttranslational translocation requires also the Sec62-63 complex (Sec62p, Sec63p, Sec71p, and Sec72p) embedded in the ER membrane, plus the soluble ER chaperone BiP, also named Kar2p in yeast (2)(3)(4)(5)(6). By chemical cross-linking of prepro-␣-factor to isolated microsomal membranes, it was possible to dissect two steps that precede post-translational pore penetration, docking of the precursor onto a receptor site where it interacts with the translocon subcomplex composed of Sec62p, Sec71p, and Sec72p, and its subsequent release for pore insertion (7). The release of the translocation substrate from the subcomplex is mediated by BiP and its co-chaperone Sec63p, and requires ATP (8). Thereafter, the signal peptide intercalates into transmembrane domains 2 and 7 of Sec61p, perhaps opening the pore, whereafter passage of pre-pro-␣factor proceeds in an ATP-and BiP-dependent manner (9).
Events preceding these steps have been much less studied. It has been thought that cytosolic Hsp70s bind to completed precursor proteins to prevent them from folding and to keep them in a translocation-competent form (10,11). However, we showed recently that newly synthesized Escherichia coli RTEM-1 ␤-lactamase folded to a native-like, catalytically active, and trypsin-resistant conformation in the cytosol of Saccharomyces cerevisiae. Thereafter, it was translocated into the ER lumen and secreted in active form to the medium (12). ␤-Lactamase was expressed as a chimeric protein, fused to a yeast-derived polypeptide (Hsp150⌬) having a signal peptide conferring post-translational translocation. The crystal structure of authentic RTEM-1 ␤-lactamase is a tight two-domain globule measuring 32 ϫ 37 ϫ 53 Å (13), whereas the 321-amino acid-long N-terminal Hsp150⌬ fragment occurs as a random coil (14). As the translocon pore has been estimated to be able to enlarge to a maximal width of 60 Å (15), folded ␤-lactamase could traverse the pore without unfolding. Here we show that benzylpenicillin, which was irreversibly bound to a mutated ␤-lactamase portion, prevented ER translocation of the fusion protein at a stage preceding signal peptide cleavage, suggesting that unfolding by a cytosolic machinery was required for initiation and completion of pore penetration.

EXPERIMENTAL PROCEDURES
Strain Construction-The DNA fragment coding for the ␤-lactamase mutants E166A or E166D was PCR-amplified with Pfu polymerase (Stratagene) using wild type ␤-lactamase gene in plasmid pKTH4539 (16) as template, and oligonucleotides MS2 (5Ј-TTCTGCAGCCGCTAC-CTC), 92395 (5Ј-ATCGTTGGGCACCGGAGCT), 92396 (5Ј-AGCTCCG-GTGCCCAACGAT), and 82629 (5Ј-GCAACCAAGCTTGAGTAAACTT-GGTCTGACAG) for the E166A mutant, and oligonucleotides MS2, 92397 (5Ј-CAGCTCCGGGTCCCAACGA), 92398 (5Ј-TCGTTGGGAC-CCGGAGCTG), and 82629 for the E166D mutant. The mutated fragments were digested with KpnI-HindIII and cloned into plasmid pKTH4539 to replace the wild type gene. The plasmids were named pKTH4825 (E166D) and pKTH4826 (E166A) and the mutations verified by sequencing. The BamHI fragments containing the mutated HSP150⌬-␤-lactamase-ADC1 terminator cassettes were cloned into plasmid pFL34, designated pKTH4828 (E166D) and pKTH4830 (E166A), and into plasmid pFL26, designated pKTH4829 (E166A). The mutant ␤-lactamase gene (E166A) with a His 6 tag was created by PCR using plasmid pKTH4829 as a template, and MS2 and C2500 (5Ј-* This work was supported by Academy of Finland Grant 38017 and by grants from the Technology Development Center (Tekes) and the Juselius Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
TTAAGCTTAGTGATGGTGATGGTGATGCCAATGCTTAATCAGT) as primers. The PCR fragments were digested with KpnI and HindIII and ligated to pKTH4539. The resulting plasmid pKTH4956 was verified by sequencing. The BamHI fragment containing the HSP150⌬-␤-lactamaseE166A-His 6 -ADC1 terminator cassette was cloned into pFL26, resulting in plasmid pKTH4960, which was transformed to yield strain H1248. The DNA fragment of HSP150⌬-␤-lactamase lacking the signal peptide codons was PCR-amplified using pKTH4539 as template, and oligonucleotides 82629 and 82239 (5Ј-ATAAATGCATAT-GGCCTATGCTCCATCTGAGCC) as primers. The product was digested with NsiI and HindIII (Promega) and ligated to plasmid pKTH4700 containing the HSP150 promotor and the ADH1 terminator to produce plasmid pKTH4716. The XbaI-NheI fragment of pKTH4716 with the truncated ⌬1-18HSP150⌬-␤ -lactamase fragment with promotor and terminator sequences was cloned into pFL26 to produce plasmid pKTH4757, which was transformed to Sey2101a (R. Schekman) to produce strain H977. The NheI-KpnI fragment derived from pKTH4716 containing the truncated version of the HSP150 gene lacking the signal sequence and flanked by the HSP150 promoter, and the KpnI-SacI fragment of pKTH4828 containing the ␤-lactamaseE166A mutant gene flanked by the ADC1 terminator, were successively cloned into pFL34, to create pKTH4995 containing the signal sequence-less Hsp150⌬-␤lactamase E166A mutant (⌬1-18E166A). pKTH4544 (16), pKTH4830, pKTH4828, and pKTH4995 were transformed into CJY004 (17) to produce strains H987, H1045, H1046, and H1376, respectively (Table I). Yeast cells were grown overnight, in synthetic complete medium lacking appropriate amino acids or nucleotides or in YPD medium, in shake flasks at 24°C.
N-terminal Sequencing of E166A-His 6 -Cells (2 liters, optical density of 1) treated with benzylpenicillin (5 mg/ml) were lysed with a Glass-Beater (Biospec) in 30 ml of Tris-HCl, pH 8.0, containing 300 mM NaCl, 1% Triton X-100, 2 mM phenylmethylsulfonyl fluoride, and 100 l of Yeast Protease Inhibitor Mixture (Sigma). The lysate was clarified by centrifugation at 10,000 ϫ g for 50 min at 4°C, and urea powder was added to 8 M concentration and the pH adjusted to 8.0. The lysate (80 ml) was mixed with 1 ml of Ni 2ϩ -nitrilotriacetic acid-agarose (Qiagen) overnight at 4°C. Further procedures were at room temperature. The resin was loaded into a 2-ml column and washed successively with 5-ml batches of buffers B, C, D, and E, which consisted of 8 M urea, 100 mM NaH 2 PO 4 , and 20 mM Tris, the pH of which was 8.0, 6.3, 5.9, and 4.5, respectively. The final wash was with 2 ml of discharging buffer (100 mM EDTA, 300 mM NaCl, 20 mM Tris-HCl, pH 8.0). Samples of the flow-through and the 1-ml fractions were analyzed by SDS-PAGE (7.5%) and Western blotting using His 5 antibody (Qiagen). The second fraction eluted with buffer D, which contained the E166A-His 6 protein of 66 kDa, was resolved in SDS-PAGE (7.5%) and blotted onto PDVF membrane (ProBlott) in 10 mM CAPS buffer, pH 11, containing 10% methanol at 50 V for 2.5 h at 4°C. The filter was rinsed with MilliQ water, stained with Coomassie Blue, and washed several times with 50% methanol, air-dried, and used for N-terminal amino acid sequencing of the 66-kDa protein (18).

Covalently Bound Penicillin G Prevents Translocation of
Hsp150⌬-␤-lactamaseE166A-E. coli RTEM-1 ␤-lactamase was fused to the C terminus of Hsp150⌬, an N-terminal portion of 321 amino acids of the secretory yeast glycoprotein Hsp150 (18). The Hsp150 signal peptide of 18 amino acids confers post-translational translocation (12). The fusion protein is designated Hsp150⌬-␤-lactamase (or ␤la for short). The Hsp150⌬ fragment has 106 serine and threonine residues, most of which acquire in the ER single mannose residues that are elongated in the Golgi (14,19), allowing distinction of the cytoplasmic (66 kDa), ER (110 kDa), and mature (145 kDa) forms (12). Hsp150⌬ consists mostly of 11 repeats of a 19-amino acid peptide, which do not adopt any regular secondary structure, as determined for the glycosylated protein by CD spectroscopy and for the synthetic unglycosylated consensus peptide by NMR spectrometry (14). Thus, conformational restrictions that would require unfolding before ER translocation should be due to the ␤-lactamase portion only.
When glutamate 166 of RTEM-1 ␤-lactamase is exchanged  into alanine, the enzyme becomes deacylation-defective, and PenG is covalently and irreversibly bound to Ser 70 of the active site (21). We fused the E166A ␤-lactamase variant to Hsp150⌬, creating Hsp150⌬-␤-lactamaseE166A (designated E166A), and expressed it in S. cerevisiae, from which the ERG6 gene was deleted (resulting in strain H1045, see Table I). In ⌬erg6 cells synthesis of the main yeast membrane sterol, ergosterol, is blocked, resulting in efficient penetration of drugs across membranes (17). The H1045 cells were preincubated with PenG and pulse-labeled with [ 35 S]methionine/cysteine. Immunoprecipitation with ␤-lactamase antiserum followed by SDS-PAGE analysis showed that E166A was cell-associated and migrated mostly like the cytosolic 66-kDa protein (Fig. 1A, lane 2). A small fraction migrated like the ER form (110 kDa) and the mature protein (145 kDa), and very little was in the medium (lane 1). After a chase of 20 min, the cytosolic form persisted and the ER form migrated more slowly (lane 4). Some protein was found in the medium (lane 3). The ␤-lactamase portion of the cytosolic PenG-bound E166A molecules was trypsin-resistant (data not shown), as shown previously for native cytosolic Hsp150⌬-␤-lactamase molecules (12). The mutation per se did not effect translocation or secretion of E166A. In the absence of PenG after the pulse, cytosolic, ER, and mature forms could be immunoprecipitated from the lysate (Fig. 1B, lane 2) and some from the medium (lane 1), whereas after chase most of the mutant protein was in the medium (lane

3), and some persisted in mature form with the cells (lane 4).
The mutation inactivated ␤-lactamase completely (Fig. 2C, asterisks). In WT ( Fig. 2A) or ⌬erg6 (Fig. 2B) cells expressing native Hsp150⌬-␤-lactamase, catalytic activity increased in the medium (EX) and some remained intracellular (IN) similarly in the absence (circles) and presence (squares) of PenG. Thus, PenG did not effect translocation, folding, or secretion of native Hsp150⌬-␤-lactamase. We conclude that, when bound to PenG, the E166A fusion protein was unable to translocate into the ER lumen.
ER Association of PenG-bound E166A-Next we studied whether cytosolic PenG-bound E166A was attached to the ER membrane. PenG-bound E166A was accumulated for 30 min in H1045 cells (E166A/⌬erg6), the cells were lysed under mild detergent conditions, and the microsomal membranes were isolated. The cytosolic and membrane fractions were subjected to SDS-PAGE and Western analysis using ␤-lactamase antiserum. Most of the 66-kDa form was detected in the microsomal fraction (Fig. 3A, lane 2) and some in the soluble fraction (lane 1). The sec63-1 and sec18 -1 mutants expressing native  Hsp150⌬-␤-lactamase served as controls. At 37°C in sec63-1 an early step of translocation is blocked, leading to cytosolic accumulation of pre-proteins and in the latter mutant membrane traffic is halted before arrival in the Golgi (22,23). In both mutants most of the 66-kDa form was pelleted with the microsomes (lanes 4 and 6). In sec18 -1 part of the Hsp150⌬-␤lactamase pool had been translocated and could be visualized as the glycosylated 110-kDa form (lane 6). The 62-kDa form probably was an artifactual degradation product rather than a biosynthetic intermediate, because the signal peptide-less form runs like a 64-kDa protein and no 62-kDa form was found after metabolic labeling (see below). Moreover, the 62-kDa form was found in lysates of the sec63-1 mutant, where no signal peptide cleavage should occur, and even in the cytosolic fraction (lane 1). H1045 cells (E166A/⌬erg6) were then incubated with PenG and subjected to immunofluorescent staining using ␤-lactamase antiserum. Mostly the plasma membrane was stained (Fig. 3B, a). In yeast cells the ER is mostly located beneath the plasma membrane, as shown in Fig. 3B (b), where another sample of H1045 cells was stained with antiserum against Lhs1p, an ER-resident protein (24). As control we used the E166A mutant lacking a signal peptide, expressed in a ⌬erg6 background (strain H1376) in the presence of PenG. This variant was not membrane-associated, as it stained the entire cytosol (Fig. 3A, c). Thus, PenG-bound E166A was mostly attached to the ER membrane in a signal peptide-dependent fashion.

PenG-bound E166A Is Associated with Translocons-Next
we showed that PenG-bound E166A molecules blocked translocation of other precursor proteins. First we used as markers BiP and CPY, mostly translocated during and after translation, respectively (1). Unlabeled PenG-bound E166A was allowed to accumulate for different times in H1045 cells (E166A/⌬erg6), whereafter the cell samples were 35 S-labeled, lysed, and immunoprecipitated with BiP antiserum. Before PenG treatment, only mature BiP was detected (Fig. 4A, lane 1). With PenG preincubation, pre-BiP started to accumulate (lane 2), until after 60 min less than half of the newly synthesized molecules were mature, and thus translocated (lane 3). The sec18 -1 mutant, where mature BiP accumulates (lane 4), and the kar2-159 mutant, where translocation of pre-BiP is partially blocked (lane 5) (2), served as controls. In the H1045 cells, the ER form of vacuolar carboxypeptidase Y (pro-CPY or p1) could be immunoprecipitated after a 5 min pulse (Fig. 4B, lane 1 Next, H1045 cells (E166A/⌬erg6) were preincubated with PenG for different times, followed by labeling with [ 35 S]methionine/cysteine for 1 h in continuous presence of the drug (Fig.  5A). The culture supernatants were trichloroacetic acid-precip- This must have been due to inhibition of their translocation, as PenG did not decrease protein synthesis. Incorporation of [ 35 S]methionine/cysteine into trichloroacetic acid-precipitable material was after 3 h as efficient as in the absence of the drug (Fig. 5B). Since pre-accumulated PenG-bound E166A inhibited translocation of an ER-resident protein (BiP), a vacuolar enzyme (CPY), and a number of secretory proteins, it must have been engaged with ER components required for both co-translational and post-translational translocation.
Cytosolic Exposure of Ligand-bound E166A-We then asked how far the Hsp150⌬ portion of PenG-bound E166A had advanced into the ER lumen, by examining signal peptide cleavage. The signal peptide appeared not to be cleaved, since PenG-bound E166A migrated in SDS-PAGE (Fig. 6, lane 1) like native pre-Hsp150⌬-␤-lactamase blocked in the cytosol before pore penetration in the sec63-1 mutant (lane 3). The cytosolic signal peptide-less Hsp150⌬-␤-lactamase variant served as a control; it migrated slightly faster (lane 2) than PenG-bound E166A. These data were confirmed by direct amino acid sequencing. An E166A fusion protein variant with a C-terminal histidine tag, E166A-His 6 , was expressed in a ⌬erg6 mutant (H1248). The cells were incubated with PenG at 37°C for 2 h and lysed by glass beads in the presence of Triton X-100. The lysate was subjected to affinity chromatography over a nickel column as described under "Experimental Procedures." Fractions that, according to SDS-PAGE and Western blotting using His 5 antibody, contained the E166A-His 6 protein of 66 kDa were subjected to SDS-PAGE, blotting onto a polyvinylidene difluoride filter, and N-terminal amino acid sequencing. The sequence was that of the signal peptide of Hsp150 (18). We conclude that the ER-attached PenG-bound E166A fusion protein was not translocated far enough to reach the signal peptidase, though the Hsp150⌬ fragment should have no structural constraints for pore penetration, and authentic Hsp150 is translocated extremely rapidly (12). This suggests that, in normal conditions, unfolding of the C-terminal ␤-lactamase portion has to occur before pore penetration can be initiated.
Reversible Binding of Ligand to the ␤-Lactamase Portion Allows Translocation-Finally we examined the Hsp150⌬-␤-lactamaseE166D mutant (E166D), which also binds PenG, but reversibly (21). We anticipated that E166D molecules should be translocated immediately when PenG dissociates from the active site. As the crystal structure of PenG-bound molecules is almost identical to that of the native unmodified ␤-lactamase molecules (25), E166D molecules binding PenG in the ER lumen after translocation should be competent for ER exit and secretion. The ⌬erg6 mutation allows penetration of drugs also  1 and 2), and chased in the presence of CHX for 20 min (lanes 3 and 4). The media (odd lane numbers; m) and lysed cell samples (even numbers; c) were immunoprecipitated with ␤-lactamase antiserum, followed by SDS-PAGE analysis. Numbers on the right indicate migration of the cytosolic (66 kDa), ER (110 kDa), and mature (145 kDa) forms of Hsp150⌬-␤-lactamase (␤la). C and D, the same experiment as in A and B for strain H987 (native Hsp150⌬-␤-lactamase in ⌬erg6 background), except that CLX was used instead of PenG and chase was for 30 min. E, strain H987 (Hsp150⌬-␤-lactamase in ⌬erg6 background) was incubated at 37°C in the presence of CLX. Intracellular (IN, open squares) and extracellular (EX, filled squares) ␤-lactamase activity was determined and plotted against incubation time. All incubations were at 37°C. across the ER membrane. Pulse-chase experiments showed that E166D was translocated and secreted similarly in the presence (Fig. 7A) and absence (Fig. 7B) of PenG. As no more ER form could be detected after chase in the presence of PenG (Fig. 7A, lane 4) than in its absence (Fig. 7B, lane 4), E166D apparently exited the ER as rapidly in free and PenG-bound form. Since the E166D mutation inactivated the enzyme similarly as shown in Fig. 2C for the E166A mutation, and PenG had no effect on the fate of the E166D fusion protein, we needed to confirm that the drug bound to the reporter protein. To this end, parallel E166D/⌬erg6 cell samples (H1046) were incubated with [ 14 C]PenG and [ 35 S]methionine/cysteine for 30 min at 37°C. Immunoprecipitation of the medium with ␤-lactamase antiserum and SDS-PAGE analysis revealed a 14 C-labeled protein comigrating with 35 S-labeled E166D (data not shown).
The results on the E166D mutant were complemented using native, enzymatically active Hsp150⌬-␤-lactamase and the penicillinase inhibitor CLX, which is bound to authentic ␤-lactamase, hydrolyzed, and released (26). CLX inactivated Hsp150⌬-␤-lactamase, confirming binding (Fig. 7E). The control experiment demonstrating secretion of active molecules in the absence of CLX is shown in Fig. 2B (circles). Pulse-chase experiments showed that CLX had no effect on translocation and secretion of Hsp150⌬-␤-lactamase (Fig. 7, C and D). These data suggest that release of the ligand allowed unfolding, which in turn allowed translocation of native Hsp150⌬-␤-lactamase as well as the E166D variant. DISCUSSION Here we show that prefolded ␤-lactamase with a covalently bound ligand could not be translocated into the yeast ER. Our reporter protein was E. coli RTEM-1 ␤-lactamase, which in authentic form is a tight globule (13). It was fused to the C terminus of a 321-amino acid fragment (Hsp150⌬) of the yeast secretory glycoprotein Hsp150. Hsp150⌬-␤-lactamase is translocated post-translationally, but before that, the ␤-lactamase portion folds in the cytosol to a native-like catalytically active conformation (12). Here we introduced to the ␤-lactamase portion a point mutation (E166A), which causes covalent and irreversible binding of PenG to the active site residue Ser 70 (21). In addition, Glu 166 , Lys 73 , Ser 130 , Asn 132 , Lys 243 , and Ala 237 are involved in substrate binding, and the crystal structure of the PenG-bound mutant protein is nearly identical to that of the native unmodified protein (25). In the presence of PenG, the Hsp150⌬-␤-lactamaseE166A protein (designated E166A) was unable to translocate and remained cytosolic with a trypsin-resistant ␤-lactamase portion. PenG and another penicillin derivative, cloxacillin, which bind reversibly to variant E166D and native ␤-lactamase, respectively, did not prevent translocation of the respective fusion proteins. We suggest that irreversibly PenG-bound E166A molecules could not penetrate the translocon because the ligand prevented unfolding, whereas release of the reversibly bound ligands allowed unfolding and translocation.
PenG-stabilized E166A was attached in a signal peptide-dependent fashion to the ER membrane, where it inhibited translocation of BiP, CPY, and a number of secretory proteins. This shows that the PenG-bound molecules were on a productive translocation pathway, normally shared by co-and posttranslationally translocated polypeptides. Whether the binding site was Sec63p or Sec61p (7), or perhaps an as yet unknown receptor upstream of these components, remains to be studied. Anyhow, PenG-bound E166A did not penetrate deep enough into the translocon pore to be processed by the signal peptidase. The failure of the 321-amino acid-long Hsp150⌬ fragment to penetrate into the ER lumen is surprising, since most of it adopts no regular secondary structure as determined by NMR spectrometry and CD spectroscopy (14), and should thus have no structural constraints for translocation. Moreover, authentic Hsp150 translocates so rapidly that no cytosolic form can be detected even after a 1-min 35 S pulse (12). The cytosolically exposed, signal peptide-containing, ER-attached PenG-bound E166A fusion protein must have represented a proper translocation intermediate, because the E166D variant binding PenG reversibly was readily translocated. It appears that unfolding of the C-terminal ␤-lactamase portion had to occur before translocation of the Hsp150⌬ portion could be initiated or advanced significantly. Pore opening is carefully controlled, and completion of the unfolding process may somehow trigger pore opening. The scenario is different from what has been suggested for mitochondrial import. Stably folded precursor proteins cannot traverse mitochondrial import sites (27,28). However, the N-terminal F 0 -ATPase subunit of 86 amino acids was fully translocated, and only the fusion partner dihydrofolate reductase, when stabilized by methotrexate, remained stalled against the mitochondrial outer membrane (29).
As PenG-bound molecules were exposed to the cytosol, the machinery unfolding the remote ␤-lactamase portion must be cytosolic. BiP has been shown not to actively pull precursor proteins, but to trap them passively by binding and preventing backwards sliding (30). This does not exclude the possibility that BiP exerted a pulling function on our reporter protein, but pulling as well as trapping could occur only after destabilization of the ␤-lactamase portion by cytosolic factors, and sufficient advancement of the polypeptide into the ER lumen for BiP to be able to grab it. It has been suggested that pore penetration becomes BiP-dependent after the signal peptide has intercalated into transmembrane domains 2 and 7 of Sec61p (9). Unfolding of cytosolic prefolded proteins for mitochondrial import has been proposed to occur as matrix Hsp70 actively pulls the polypeptide through import sites. In this scenario mtHsp70 is viewed as a motor, which interacts with the inner membrane protein Tim44 to generate pulling force acting on the translocation substrate (29,31). Other data suggest that mtHsp70, anchored to Tim44, acts like BiP as a ratchet minimizing retrograde movements, that unfolding occurs spontaneously, and that forward movement is driven by Brownian motion (32,33). Import of polypeptides into mitochondria requires a high degree of unfolding. Only 50 amino acids were sufficient to span both outer and inner membrane, demonstrating that the passenger polypeptide is imported into the matrix in an extended state (34). According to equilibrium measurements with 8 M urea, pre-␤-lactamase folds in vitro via a molten globule state, which retains native-like secondary structure but lacks catalytic activity, and whose compactness is between those of native and completely unfolded forms (35). Whether the translocation-competent ␤-lactamase portion retains secondary structure in yeast cells remains to be determined.
Pre-pro-␣-factor was post-translationally translocated, in the absence of cytosolic components, into reconstituted proteoliposomes, which contained the heptameric translocon complex in the membrane and BiP in the lumen (30). However, pre-pro-␣factor was urea-denatured before dilution and the translocation assay. Moreover, it is not known whether it folds prior to translocation in vivo. Loosely folded molecules may not require unfolding, and evolution may have selected such proteins for post-translational translocation, and directed tightly folded proteins to the co-translational pathway, which allows folding only on the luminal side of the ER membrane. Nevertheless, our data unravel new activities in the yeast cytosol, unfolding of tightly folded protein for post-translational ER translocation, and a connection between unfolding of the translocation substrate and pore opening. Such events were not anticipated, as it was thought that polypeptides do not fold to native-like conformation prior to translocation. Our experiments were performed on living cells, confirming the physiological relevance of the findings.