Mutations in the Sec61p Channel Affecting Signal Sequence Recognition and Membrane Protein Topology*

The orientation of most single-spanning membrane proteins obeys the “positive-inside rule”, i.e. the flanking region of the transmembrane segment that is more positively charged remains in the cytosol. These membrane proteins are integrated by the Sec61/SecY translocon, but how their orientation is achieved is unknown. We have screened for mutations in yeast Sec61p that alter the orientation of single-spanning membrane proteins. We identified a class of mutants that are less efficient in retaining the positively charged flanking region in the cytosol. Surprisingly, these mutations are located at many different sites in the Sec61/SecY molecule, and they do not only involve charged amino acid residues. All these mutants have a prl phenotype that so far have only been seen in bacteria; they allow proteins with defective signal sequences to be translocated, likely because the Sec61p channel opens more easily. A similar correlation between topology defects and prl phenotype was also seen with previously identified yeast Sec61 mutants. Our results suggest a model in which the regulated opening of the translocon is required for the faithful orientation of membrane proteins.

The Sec61/SecY translocon complex mediates translocation of proteins across and integration into the endoplasmic reticulum (ER) 3 membrane (1,2). Proteins are targeted to this complex by hydrophobic signal sequences (3), and the signals and subsequent hydrophobic segments of the polypeptide are integrated into the bilayer as transmembrane helices. The translocon consists of Sec61␣ (Sec61p in yeast) with 10 transmembrane domains, and the single spanning proteins Sec61␤ (Sbh1p) and Sec61␥ (Sss1p), corresponding in bacteria to SecY/ SecG/SecE. The crystal structure of the SecY complex of the archaebacterium Methanococcus jannaschii (4) revealed a compact helix bundle that can form a hydrophilic channel with a lateral exit site. The central passage is blocked by a lumenal plug domain and a central constriction ring. To open, the plug needs to move out, and the ring must expand (5,6). The crystal structure and cross-linking experiments (7) suggest that the translocation channel is formed by a single heterotrimer, although it has been proposed that upon opening of the channel two or more complexes may cooperate (8,9).
The simplest case of protein topogenesis is the orientation of a signal sequence upon entering the translocon. Cleavable signal sequences assume a N cyt /C exo orientation (cytoplasmic N and exoplasmic C terminus), whereas noncleavable signal-anchors of membrane proteins insert in N exo /C cyt or N cyt /C exo direction to translocate either their N-or their C-terminal end, respectively. Best established is the role of charged residues on either side of the hydrophobic core of the signal, which according to the "positive-inside rule" (10,11) generally results in the more positive end to be cytosolic. Additional determinants are the folding of sequences N-terminal to an internal signal that may sterically hinder N-translocation (12) and the hydrophobicity of the apolar signal core (13)(14)(15). Analysis of model proteins with very hydrophobic N-terminal signal-anchors and positive N-terminal flanking charge indicated that the signal initially enters the translocon to produce an N exo /C cyt orientation followed by inversion (16). Inversion appeared to be driven by charge interactions and to be slowed down by hydrophobic interactions of the signal core with the translocon or the membrane. At which point in the process, the plug domain moves out of its lumenal binding cavity is not known.
Systematic mutation of conserved charged residues in Sec61p identified three residues that affected model protein topogenesis as predicted by the positive-inside rule (17): Glu 382 at the cytoplasmic-end of transmembrane segment TM8 and Arg 67 and Arg 74 in the plug domain. Additional mutations in the plug domain, including its complete deletion, all prove viable without growth defect, but show the same effect on signal orientation (22). The plug deletion was partially defective in coand post-translational translocation and appeared to be inefficient in Sec61 complex assembly.
To obtain more insight into the mechanism of topogenesis, we now used a screen to isolate mutations in Sec61p that influenced the orientation of diagnostic signal-anchor proteins. We identified 18 single-residue mutations that could be grouped into three classes with distinct effects on topology. One of them, also including all known plug mutations and the full plug deletion, showed suppression of signal sequence mutations, i.e. a prl phenotype, as previously observed only in prokaryotes (18 -20). In Escherichia coli, prlA mutations in SecY were isolated as altering protein localization of proteins with mutated signal sequences. In view of the translocon structure and supported by experimental data, prlA effects are explained by destabilization of the closed state of the translocon and facilitated pore opening (4 -6, 19, 20). The close correlation between a prl phenotype and a specific effect on signal orientation thus sheds light on the role of translocon gating in the process of protein insertion.  (21) in which pDQ1 (i.e. YCplac111 (LEU2 CEN) containing SEC61 with codons 2-6 replaced by codons for H 6 RS and with its own promoter) was exchanged for YCPlac33 (URA3 CEN) with the same SEC61 gene. This made it possible to introduce mutant sec61 in YCplac111 (LEU2 CEN) by plasmid shuffling using 5-fluoro-orotic acid (5FOA). VGY61 with a disruption of SSH1 is described previously (17). Sec61p mutant strains with the mutations L63N, L66N, L70N, the triple mutation LLLNNN, the deletions ⌬L70, ⌬tip (residues 67-72 replaced by a glycine), and ⌬plug (residues 52-74 replaced by a glycine) have been reported previously (22).
To analyze membrane insertion of dipeptidyl aminopeptidase B (DPAPB), CPY, invertase, and ␣-factor, the coding sequences of DAP2, PRC1, SUC2, and MF␣1 were amplified by polymerase chain reaction (PCR), fused to a C-terminal triple-HA epitope tag. Signal sequence truncations were produced by PCR using mutagenic primers. Codons 34 -235 of ATP2p (the ␤ subunit of mitochondrial ATP synthase, deleting the leader peptide) and codons 1-200 of Gal1p (cytosolic galactokinase) were fused to a sequence encoding EFNNSTEMMM-MMACYPYDVPDYAGYPYD-VPDYAYPYDVPDYA with a glycosylation site (in bold) and triple-HA tag (underlined), producing Atp2p⌬-g and Gal1p⌬-g, respectively. To test the glycosylation tag, codons 1-80 of H1 including its signal-anchor were fused in front of the Atp2p construct (H1Atp2p⌬-g). These constructs were cloned into pRS426 with a GPD promoter.
Mutagenesis of Sec61p and Library Construction-Site-directed mutagenesis was performed by PCR using appropriate mutagenic primers and Vent polymerase (New England Biolabs). All constructs were verified by sequencing. Random in vitro mutagenesis of SEC61 was performed by error-prone PCR using 0.1 g of template plasmid, 150 pmol of each primer, and 5 units of Taq polymerase (Roche Diagnostics) in 100 l of 10 mM Tris, pH 8.3, 50 mM KCl, 400 M MgCl 2 , 300 M MnCl 2 . Four sets of primers were designed to amplify four quarters of ϳ350 bp of the coding sequence delimited by unique restriction sites: XbaI, SacI (created by a silent mutation), StuI, AccI, and BamHI (introduced after the stop codon) at nucleotide positions 27, 343, 710, 1099, and 1470 from the initiation codon, respectively. For each segment, four parallel PCR reactions with three dNTPs at 200 M and the fourth at 40 M were performed (30 cycles of 30 s at 94°C, 40 s at 46°C, and 1 min at 72°C) and the products gel-purified, pooled, digested with the restrictions enzymes for the flanking sites, gel-purified, and ligated into pDQ1SacBam (pDQ1 with the additional SacI and BamHI sites). Upon electroporation into E. coli DH5␣, ϳ40,000 colonies were obtained for each mutagenized segment, scraped into medium, incubated at 37°C for 10 h followed by purification of the plasmid DNA.
Screening Procedure-The four plasmid libraries were transformed into the screening strain, grown for 4 days at 30°C on SD-Leu plates, replica-plated on 5FOA plates, and grown for 2 days at 30°C to select cells that had lost the wild-type SEC61 URA3 plasmid. Resistant colonies were replica-plated onto YPD with adenine (YPDA), incubated for 1 day at 30°C, overlaid with 0.5%-agarose in 0.5 M KHPO 4 , pH 7, containing 0.1% SDS, and 400 g/ml X-gal (Applichem) and incubated for 6 h at 30°C and then at 4°C for further color development. For each segment library, ϳ3000 colonies were screened. Colonies with darker or lighter color were streaked on YPDA plates, and the color assay was repeated. Plasmids were rescued by the procedure of Robzyk and Kassir (23), amplified in DH5␣, transformed into VGY61, selected on SD-Leu, and replica-plated on plates containing 5FOA to eliminate the wild-type SEC61 plasmid. The plasmids encoding the model constructs were separately transformed into the resulting strains.
Labeling and Immunoprecipitation-Yeast cells were in vivo pulse-labeled for 5 min with 150 Ci/ml [ 35 S]methionine (GE Healthcare), lysed with glass beads, heated at 95°C for 5 min with 1% SDS, cleared by centrifugation, subjected to immunoprecipitation, and analyzed by SDS-gel electrophoresis and autoradiography as described previously (22). Signals were quantified by phosphorimager.
Competition between Wild-type and Mutant Sec61p-To analyze the stability of Sec61p mutants in the presence of wildtype Sec61p, the sec61 coding sequences were extended by a sequence encoding a triple-HA epitope tag, cloned with the original promoter into YCplac111 (LEU2 CEN), transformed into VGY61, and grown on SD-Leu-Ura to maintain both wildtype and HA-tagged mutant copy of Sec61p. Cells were lysed in SDS-sample buffer with glass beads, boiled for 10 min, separated by SDS-gel electrophoresis, blotted onto nitrocellulose, and decorated with a rabbit antiserum against the C terminus of Sec61p, a gift by C. Stirling (University of Manchester, UK; 24), or with an anti-HA antibody. Antibody was detected using horseradish peroxidase-conjugated secondary antibody and the enhanced chemiluminescence kit (Amersham Biosciences).

A Screen for Sec61p Mutations Affecting Protein Topology-
To identify mutations in Sec61p that affect membrane protein topogenesis, we devised a screen based on the color reaction catalyzed by the model protein 40[H1](ϩ5)LacZ (Fig. 1). It is a derivative of the asialoglycoprotein receptor H1, a type II single-spanning membrane protein. Because of inversion of the flanking charges of its signal-anchor sequence (resulting in a charge difference ⌬(C-N) of ϩ5 according to 11), it inserts with mixed orientations into the ER membrane (25) and is thus a sensitive reporter of alterations affecting topology. The C-terminal sequence was replaced by a segment of yeast CPY and the coding sequence of bacterial ␤-galactosidase (LacZ), which is active in the cytosol but inactive, when translocated. Mutations in Sec61p that affect the membrane orientation of 40[H1](ϩ5)LacZ result in changes in enzyme activity detectable using the color substrate X-gal.
The coding sequence of Sec61p was mutagenized by errorprone PCR and transformed on a CEN plasmid with its own promoter into the yeast strain VGY61 with a chromosomally integrated 40[H1](ϩ5)LacZ gene. After elimination of the wildtype copy of SEC61, colonies were analyzed for LacZ activity. Colonies deviating in color intensity from the background of colonies with wild-type Sec61p were picked, and the plasmids were isolated and transformed into VGY61 cells expressing 40[H1](ϩ5) without the LacZ domain (construct A in Fig. 1). Upon elimination of wild-type SEC61, the cells were labeled with [ 35 S]methionine, and construct A was immunoprecipitated and analyzed by SDS-gel electrophoresis and autoradiography. In this manner, 29 Sec61 plasmids from a total of ϳ12,000 transformants were identified to produce an increase and 46 to produce a decrease in glycosylated, i.e. C-terminally translocated, forms of construct A. Upon sequencing, 18 single point mutations were identified to affect the topology of construct A.

Sec61p Mutations Display Distinct Topology Effects-
The effects of these mutants as well as of R67E, R74E, and E382R analyzed previously (17) and I91T (one of four mutations in   Fig. 1 were expressed in cells with the indicated Sec61p mutants, labeled with [ 35 S]methionine, and analyzed by immunoprecipitation, SDS-gel electrophoresis, and autoradiography. 2-and 3-fold glycosylated forms correspond to C-terminally translocated proteins. The fraction of C-terminally translocated products of the total is plotted as the deviation of that observed for the wild type in percentage points (left scale). The scale on the right shows the absolute percentage of C-translocated products. For constructs B, D, and E, averages of 2-8 determinations with S.D. are shown. In addition to the mutants identified in the screen, the effects of R67E, R74E, E382R, and I91T were included. Based on the effects, the mutants were grouped into three classes. P292S was classified because of its integration defect apparent from the increased unglycosylated species of constructs C and D. n, not tested; wt, wild-type.

JOURNAL OF BIOLOGICAL CHEMISTRY 33203
Sec61-23; 21) on topology were tested for the five model proteins shown in Fig. 1. Construct B is identical to A except for the hydrophobic core of the signal-anchor that consists of a Leu 16 stretch instead of the natural transmembrane segment of H1. Upon expression in yeast cells, three forms were produced corresponding to the protein without glycans or with two or three glycans (17). Because the glycosylation sites are in the C-terminal domain, the glycosylated species represent polypeptides with a N cyt /C exo orientation. In cells expressing wild-type Sec61p, 40% of construct A and 46% of construct B were glycosylated (22 and not shown). In cells expressing mutant Sec61p, this fraction was either increased or reduced as summarized in Fig. 2. The effects were essentially the same for constructs A and B, indicating that the hydrophobic cores of the two signals were handled identically. No singly glycosylated forms were detectable for any mutant (data not shown) indicating that glycosylation efficiency was not affected by the mutations.
The signal-anchors of constructs A and B have a positive charge difference ⌬(C-N) typical of proteins with a N exo /C cyt orientation. Based on studies in mammalian cells (12,14,25), they insert with mixed topologies because the 40-amino acid N-terminal domain of H1 hinders N-translocation. In construct C, this domain was in part replaced by the N-terminal translocated sequence of Ste2p, resulting in reduced C-translocation with wild-type Sec61p (12%). All Sec61p mutants that increased C-translocation of A and B, also did so for C (Fig. 2).
Those that reduced C-translocation of A and B showed no significant effect, most likely because construct C was close to minimal C-translocation already with wild-typeSec61p. With a reduced charge difference ⌬(C-N) of ϩ1, i.e. with two positive flanking charges at the N-and three positive charges at the C-terminal end of the transmembrane domain, construct D was again at the center of the diagnostic range (60% C-translocation with wild-type Sec61p; 22), and all mutant translocons affected the ratio of orientations. Except for mutants W35R, T185K, and P200L, they did so in the same direction as with constructs A and B (Fig. 2). With the Ste2p sequence at the N terminus of constructs C and D, an efficient N-glycosylation site was introduced so that polypeptides integrated into the ER membrane in either orientation were glycosylated: two or three times when the C terminus was translocated, and once when the N terminus was translocated. The appearance of unglycosylated products thus revealed polypeptides that were not integrated into the membrane. When expressed in cells with wildtype Sec61p, 5.3% (Ϯ1.4%, n ϭ 8) of construct D were unglycosylated (17). Except for P292S, which generated ϳ25% nonintegrated products, none of the mutants produced more than 10% of this form, indicating that they did not cause a significant defect in membrane integration.
Finally, model protein E differs from all others in that its signal-an-

prl Mutations in Yeast Sec61p Affect Topogenesis
chor is at the very N terminus (Fig. 1). Despite a negative charge difference of Ϫ3, this protein is inserted with ϳ30% N cyt /C exo orientation by wild-type Sec61p, because of its highly hydro-phobic signal-anchor (22). Many Sec61p mutants also affected the insertion behavior of this protein, mostly by reducing C-terminal translocation (Fig. 2).
In our analysis, as summarized in Fig. 2, the observed changes in model protein orientation versus wild-type Sec61p generally ranged from 5 to 20% age points and are therefore in the range that is also observed for single charge mutations flanking the signals of substrate proteins (17). Based on their effects on the different model proteins, the mutants could be grouped into three classes. Class 1 mutants consistently increased translocation of the more positively charged end of the signal-anchor of all substrates tested. This corresponds to the phenotype of a weakened positive-inside rule. The plug deletion and all other plug mutants previously prepared by directed mutagenesis also belong to this class (22). In contrast, class 3 mutants generally reduced C-translocation. Class 2 mutants only differed from class 3 in having the opposite effect on substrate D.  Table 1 are labeled and highlighted as spheres in red, black, and blue to indicate class 1, 2, or 3 topology effects, respectively.   Fig. 2 and 26). P292S was not classified because of its co-translational translocation defect. d prl phenotype based the translocation efficiency of CPY⌬3 in comparison to wild-type Sec61p (Fig. 6). Ϯ, and ϩ indicates less and more than 2-fold the efficiency of wild-type Sec61p, respectively.

JOURNAL OF BIOLOGICAL CHEMISTRY 33205
Translocation Efficiency and Stability of Mutant Translocons-In addition to Sec61p, Sbh1p, and Sss1p, yeast expresses a second nonessential translocon complex consisting of Ssh1p (Sec61 homolog), Sbh2p, and Sss1p (26). In ⌬ssh1 cells lacking the second translocon, topology changes by mutations D61N, W35R, and Q93R (as representatives of the three classes) were essentially the same 4 , as previously shown for R67E, R74E, and E382R (17). None of the Sec61p mutants displayed a growth defect at 30°C (not shown) or at 39°C even in the absence of Ssh1p (Fig. 3). Mutations I91T and P292S were cold sensitive and showed reduced growth in the presence of tunicamycin, which caused protein misfolding and ER stress (Fig. 3). I91T thus is responsible for the phenotype of Sec61-23, from which it is isolated. None of the other mutations showed sensitivity to cold or tunicamycin stress.
Efficiency of translocation was tested using CPY and DPAPB (Fig.  4A), established post-and co-translational substrates, respectively (27), for all mutants including the plug mutations prepared previously. Most mutants are as efficient in translocation as wild-type Sec61p. Exceptions are P292S and the full plug deletion ⌬plug that produces significant levels of unglycosylated CPY and DPAPB. A post-translational translocation defect was in addition observed for I91T and most severely, the partial plug deletion ⌬tip.
We previously observed that ⌬plug competes poorly with wildtype Sec61p for limiting factors in a heterozygous situation (22). When coexpressed with wild-type Sec61p, HA-tagged ⌬plug was hardly detectable in steady state by immunoblot analysis (Fig. 4B, lane 30). In contrast, all HA-tagged point mutants were present in similar amounts as HA-tagged wild type in the presence of untagged wild-type Sec61p (Fig. 4B, lanes 1-29). These results show that the full plug deletion has additional defects compared with the point mutations in the plug or elsewhere. These are most likely the result of joining the lumenal ends of TM1 and TM2 by a very short connection and so immobilizing them in fixed relative position.
Class 1 Topology Effects Correlate with a prl Phenotype-The mutations affecting topology are listed in Table 1, and their locations are indicated in the structure model of the yeast Sec61 complex in Fig. 5. Most mutated residues are well conserved between species. The mutations are not limited to a single domain of the protein, but are found throughout the sequence. New mutations were found in the plug domain (D61N) and in TM2 facing the plug (E79G). Plugs destabilizing mutations as well as partial or complete deletion of the plug have previously been shown to affect topology with class 1 effects (22). In our screen, mutations were also isolated in three of the six residues forming the central constriction ring (I86T, T185K, and M450K). Several mutations are concentrated where TM2 contacts TM8 (I91T, Q93R, Q96R, T379I, and E382R) and two mutations (K284E and E460K) disrupt the ion-bridge connecting the cytosolic ends of TM7 and TM10. 4 T. Junne and M. Spiess, unpublished observations. FIGURE 6. Class 1 mutants show a prl phenotype. A, wild-type and truncated signal sequences of CPY and DPAPB are shown in single-letter code. The signal cleavage site is indicated by --. B, CPY⌬3 or DPAPB⌬37 were expressed in cells with wild-type (wt) or the indicated mutant Sec61p, labeled, and analyzed as described in the legend to Fig. 4A. C, translocation efficiency was quantified by phosphorimager and plotted as black, white, light gray, and dark gray bars for cells with wild-type, class 1, 2, or 3 Sec61p mutants. Plug mutations prepared previously (22) are presented on the right. For CPY⌬3, the average of triplicate determinations is shown. The horizontal lines indicate the wild-type levels. D, CPY⌬6, CPY⌬9, and the complete signal deletions CPY⌬19 and DPAPB⌬45 were analyzed for selected Sec61p mutants as in B.

prl Mutations in Yeast Sec61p Affect Topogenesis
The distribution of the mutations in Sec61p is reminiscent of that of prlA mutations in E. coli SecY (19,20). Several prlA mutations were found in the plug domain and in ring residues. Three mutations are exactly at corresponding positions: Asn 65 in E. coli SecY (prlA8914), Ile 191 (prlA200), and Ile 408 (prlA4-2), matching Arg 67 , Thr 185 , and Met 450 , respectively, in Sec61p. In addition, it has very recently been shown that deletion of the plug domain of E. coli SecY has a prl phenotype (28,29). We therefore set out to test our Sec61p mutations for a prl phenotype in yeast.
To inactivate the signal sequence of CPY, the first three residues of its hydrophobic core were deleted (CPY⌬3; Fig. 6A). This reduced CPY translocation by wild-type Sec61p to ϳ15%. Several of our Sec61p mutants significantly restored CPY⌬3 translocation, in some cases up to more than 70% (Fig. 6, B and  C; top). When grouped according to their topogenic effects, most of the class 1 mutants (Fig. 6C, white bars), including all the plug mutants displayed the prl phenotype. Only one other mutant showed more than twice the translocation efficiency of the wild type; mutation of Glu 460 , the ion-bridge partner of Lys 284 , which in mutation K284E produced, class 1 and prl phenotypes. Deletion of the N-terminal sequence of DPAPB including half of its signal-anchor completely abolished translocation by wild-type Sec61p. This defect was again partially suppressed by a majority of the class 1 mutants (Fig. 6, B and C;  bottom).
Further truncation or the complete deletion of the CPY signal abolished CPY translocation by the wild-type translocon, but was still rescued at a very low level by mutations L63N, ⌬plug, and L131P (Fig.  6D) that were tested as representatives of prl mutants identified with CPY⌬3. In contrast, signal-less DPAPB was not translocated at all (Fig. 6D). The observed background of signal-independent CPY translocation by Sec61p mutants appears to be restricted to this protein and may be the result of a cryptic targeting signal within its mature portion. Deletion of the signal sequences of invertase and ␣-factor, additional co-and post-translationally translocated proteins, completely abolished export (Fig. 7, A-C). Translocation of the partial signal deletions was largely restored by the Sec61p point mutants tested, whereas ⌬plug at least recovered its relative translocation defect. No translocation was furthermore detected for polypeptides derived from a cytosolic (Gal1p) or a mitochondrial (and thus intrinsically translocation competent) protein (Atp2p) equipped with a glycosylation site (Fig. 7D). There is thus no unspecific leakage of proteins through the mutant translocons.
As shown quite directly for the bacterial translocon (5,6), prl mutations facilitate the opening of the translocation pore. The strong correlation between the prl phenotype and a distinct topology effect for a subset of our Sec61p mutations therefore suggests a role of timely pore opening in the proper orientation of signal sequences.

Sec61p Mutations Affect Protein Orientation in Distinct Ways-
We have identified mutations in yeast Sec61p that affect the orientation of diagnostic signal-anchor proteins during insertion into the ER membrane. Point mutations that strongly interfere with transmembrane orientation (if they exist) are likely to have serious growth defects, precluding their isolation in our screen. This may explain why our mutations produce rather mild changes in topology, approximately in the range also caused by single charge mutations in signal flanking regions of the model substrates (17). This is generally not sufficient to perturb the topology of natural proteins. However, SecY mutations have recently been described that cause periplasmic stress and lactose permease misfolding in E. coli (30). Two of these mutations, secY238 and secY351, occurred at corresponding positions to our Q261R and M400K mutants in FIGURE 7. prl Mutants do not allow translocation of signal-less proteins. A-C, invertase (Inv) and ␣-factor (␣F) and the partial (Inv⌬4, ␣F⌬5) and complete (Inv⌬19, ␣F⌬18) signal deletion mutants were analyzed and quantified as in described in the legend to Fig. 6, B and C. D, as control proteins, the N-terminal portion of cytosolic galactokinase and of mature ␤-subunit of the mitochondrial ATP synthase fused to a glycosylation and HA tag (Gal1p⌬-g and Atp2p⌬-g, respectively) were analyzed for the appearance of nonspecifically translocated and thus glycosylated products. The expected position of glycosylated products is indicated by an arrowhead. To confirm the functionality of the glycosylation tag, Atp2p⌬-g with a signal-anchor of H1 (H1Atp2p⌬-g) was expressed in cells with wild-type Sec61p and incubated with (ϩeH) or without (--) endoglycosidase H.
Sec61p, suggesting that mild topology defects in endogenous proteins were not unlikely.
Interestingly, the mutants, although distributed over the entire sequence, show distinct topology phenotypes on the model substrates tested (Fig. 2). The effects of class 1 mutants are consistent with a weakened positive-inside rule. Class 3 mutations, in contrast, appear to generally hinder translocation of the more sizable or less flexible portion of the model proteins and therefore to favor transfer of the smaller N-terminal sequence. Three mutants (class 2) deviate from this behavior for one substrate protein. The clearest insight into the mechanism of how some Sec61p mutations influence signal orientation is provided by the striking correlation between class 1 topology effects and a prl phenotype.
Sec61 Mutations Can Cause a prl Phenotype in Eukaryotes-In E. coli, prlA alleles of SecY are initially isolated as suppressors of signal sequence mutations in secretory proteins (18 -20). This phenotype suggested a signal-dependent proofreading mechanism in SecY that prevents proteins with defective signals to be translocated by the wild-type translocon. Suppressor mutations are not allele-specific and suppress signal truncations as well as charge insertions. Interestingly, an effect on the topology of model membrane proteins is shown for two prlA alleles (31,32). Secretory proteins completely lacking the signal, but not cytosolic proteins, are still exported by prlA mutants in a SecB-dependent manner (33). This is accounted for by the fact that SecB recognizes secretory proteins within their mature sequences (34,35) and thus targets them to the translocon. Eukaryotic cells lack the SecB/SecA pathway of protein targeting, and no export-specific chaperones have been detected (36), which seems to explain why a prl phenotype has so far not been observed in eukaryotes. Yet, we found that a number of mutations in the Sec61 translocon of yeast also suppressed signal sequence mutations. However, unlike in the bacterial system, complete signal deletions were generally not suppressed, confirming the absence of signal-independent targeting mechanisms.
The closed state of the Sec61/SecY complex forms a compact structure stabilized from within by the bound plug domain. Many prlA mutations are altered in the plug domain of SecY in the pore ring or in residues contacting the plug (20), suggesting that they destabilize the closed conformation and facilitate plug movement and pore opening. This has been demonstrated experimentally by cysteine cross-linking of the plug with SecE outside the channel (5) and by observing spontaneous channel opening in conductance measurements (6). Accordingly, deletion of the plug domain is also found to have a prl phenotype (28,29).
In agreement with this, also all plug mutations in yeast Sec61p as well as mutation of Asp 61 forming part of the plug cavity and of the ring residue Ile 86 produced a prl phenotype. In addition, elimination of the salt bridge between Lys 284 and Glu 460 that connected the cytosolic ends of TM7 and TM10 by mutation of either partner showed a prl phenotype, consistent with a stabilizing role of this ionic bond. For three other prl mutations, the mechanism of destabilization is not as obvious, but may well be a conformational change with that effect.

The Correlation between prl and Topology Phenotypes
Suggests a Role for Regulated Pore Opening in Signal Orientation-The conspicuous correspondence between mutations causing a prl phenotype and a class 1 topology effect suggests a role for timely plug opening in the process of signal orientation. The mutants with facilitated plug movement and pore opening are apparently less efficient in rejecting the positive end of a signal sequence from entering the channel and translocation through. The central part of the plug domain of Sec61p contains two charged residues, Arg 67 and Arg 74 , at least one of which is conserved in Sec61␣ sequence of eukaryotes and archeae. Plug movement out of the pore thus entails the removal of these positive charges from the channel (schematically illustrated in Fig. 8), thereby reducing repulsion of the more positively charged end of the signal from entering the pore. The class 1 phenotype thus corresponds to the effect of premature pore opening, as induced by prl mutations. Accordingly, timely plug movement contributes to the fidelity of membrane protein orientation in the ER. This model also suggests that signals orient themselves in the cytoplasmic pore cavity before full channel opening and insertion of the hydrophilic polypeptide. This is particularly plausible considering the narrow space within a monomeric translocation channel.
Glu 382 is also identified to contribute to the positive-inside rule (17). Situated at the cytosolic end of TM8, it may attract the more positive end of signals. Its mutation to arginine FIGURE 8. A model how premature channel opening affects signal orientation. As the signal enters the cytoplasmic cavity of the wild-type translocon, an electrostatic field, to which Arg 67 and Arg 74 in the plug and Glu 382 on the cytosolic side contribute, acts on it to orient it according to its flanking charges, before pore opening and plug movement occur. Because of this, translocation of the positive end of the signal does not happen (crossed arrow). Mutations that destabilize the closed state cause premature pore opening, potentially allowing translocation of the positive end for a fraction of the products.

prl Mutations in Yeast Sec61p Affect Topogenesis
shows the corresponding class 1 effect on signal orientation. The absence of a prl phenotype for this mutation confirms that it does not exert its effect by promoting pore opening as for example the plug mutations. Other exceptions to the correlation between prl and class 1 phenotypes may be explained by the type of mutation at a particular position. K284E and E460K both destroy the stabilizing salt bridge and thereby cause the prl phenotype. The large lysine in place of Glu 460 , however, appears to produce an additional disturbance responsible for the different topology effects of the two mutations. In this light, it is also doubtful whether class 2 mutations represent a homogeneous group. The present list of mutations will serve as a starting point for further directed mutagenesis.