Sss1p, an essential component of the heterotrimeric Sec61 complex in the ER (endoplasmic reticulum), is a tail-anchored protein whose precise mechanism of action is largely unknown. Tail-anchored proteins are involved in many cellular processes and are characterized by a single transmembrane sequence at or near the C-terminus. The Sec61 complex is the molecular machine through which secretory and membrane proteins translocate into and across the ER membrane. To understand the function of the tail anchor of Sss1p, we introduced mutations into the tail-anchor sequence and analysed the resulting yeast phenotypes. Point mutations in the C-terminal hydrophobic core of the tail anchor of Sss1p were identified that allowed Sss1p assembly into Sec61 complexes, but resulted in diminished growth, defects in co- and post-translational translocation, inefficient ribosome binding to Sec61 complexes, reduction in the stability of both heterotrimeric Sec61 and heptameric Sec complexes and a complete breakdown of ER structure. The underlying defect caused by the mutations involves loss of a stabilizing function of the Sss1p tail-anchor sequence for both the heterotrimeric Sec61 and the heptameric Sec complexes. These results indicate that by stabilizing multiprotein membrane complexes, the hydrophobic core of a tail-anchor sequence can be more than a simple membrane anchor.
Transport across the ER (endoplasmic reticulum) is a vital first committed step in the biogenesis of secretory and integral membrane proteins. These proteins are targeted to the ER either co-translationally, as ribosome-bound nascent chains, or in a post-translational manner, as fully synthesized polypeptides.
The translocon is the molecular machine that forms the channel in the ER through which proteins are conducted. The core of the translocon is composed of a heterotrimeric membrane complex called the Sec61 complex [1–5]. The α- and γ-subunits of this complex (Sec61p and Sss1p in yeast, and SecY and SecE in bacteria) show significant sequence conservation and are essential for cell viability. In contrast, the β-subunits (Sbh1p/SecG) are dispensable and are similar in archaea and eukaryotes, but they show little homology with the corresponding subunit in eubacteria.
In yeast, the Sec61 complex is required for both co- and post-translational translocation. During co-translational translocation, the Sec61 complex, the ER membrane protein Sec63p and the ER luminal chaperone Kar2p [the yeast homologue of BiP (binding immunoglobulin protein)] are necessary for efficient protein import [6–9]. In reconstituted systems, the co-translational translocation of proteins synthesized from membrane-docked ribosomes requires only the Sec61 complex . However, post-translational translocation requires the Sec61 complex as well as another multiprotein complex that contains Sec62p, Sec63p, Sec71p and Sec72p. The resulting heptameric complex is known as the Sec complex [10,11]. The Sec complex consists of both essential (Sec62p and Sec63p) and non-essential (Sec71p and Sec72p) proteins [10,12,13]. In addition, Kar2p is required for post-translational transport; it binds to the J domain of Sec63p and facilitates translocation of the emerging polypeptide through successive rounds of binding that are driven by ATP hydrolysis [14,15]. Proteins using the post-translational pathway are maintained in a translocation-competent conformation by molecular chaperones, which also play a role in targeting them to the heptameric Sec complex [16,17].
The γ-subunit of the Sec61 complex, or Sss1p in the yeast Saccharomyces cerevisiae, is a 9 kDa tail-anchored protein. Depletion of Sss1p leads to a severe defect in co- and post-translational translocation . Tail-anchored proteins, such as Sss1p, are inserted into membranes post-translationally via a C-terminal hydrophobic region [19–21]. It is not completely clear how distinct tail-anchor sequences favour the association of a protein with one intracellular membrane versus another, even though the machinery that inserts tail-anchor proteins into the ER membrane has been better defined in recent years [22–24].
A number of models have been proposed to define why Sss1p is essential and why protein depletion leads to profound translocation defects. Sss1p may help to maintain the membrane permeability barrier by acting as a place holder for signal peptides within the Sec61 complex , it may facilitate the oligomerization of the heterotrimeric Sec61 complex  and/or it may clamp together the two halves of the Sec61α subunit . The cytoplasmic region of Sss1p has also been shown to interact with oligosaccharyl-transferase [28,29], suggesting that the protein helps assemble other components required for nascent secretory protein biogenesis. The hydrophobic core of tail-anchor sequences are generally considered to have limited function(s) other than localization of the protein at the correct subcellular membrane location. Consistent with this view, point mutations of the conserved glycine residues in the hydrophobic core of Sss1p were tolerated, but deletion of the 12 residues C-terminal of the hydrophobic sequence disrupted the translocation function of the Sec61 complex . The molecular cause of the functional defect in translocons with large deletions in Sss1p remains unclear because the mutant Sss1p proteins still targeted to, and interacted with, Sec61p .
To determine the molecular mechanism for which the tail-anchor sequence of Sss1p is important, we introduced specific mutations into this motif. Point mutations in conserved sequences in both the transmembrane and flanking regions were analysed phenotypically and by biochemical methods. The impact of single point mutations on cellular physiology was remarkably profound, including complete alteration of the ER structure and cessation of cell division. Results from this genetic analysis were then complemented by biochemical analyses of isolated Sec61 complexes and the larger Sec complexes that revealed the molecular defects in specific mutant proteins. Our collective results suggest that the specific sequence of the hydrophobic core of the tail anchor of Sss1p is critical for stabilizing the Sec61 complex and, therefore, Sec complexes. Surprisingly, Sss1 proteins with mutations in this region still assemble into heterotrimeric Sec61 complexes and heptameric Sec complexes, yet the oligomers are less stable and exhibit translocation defects that lead to profound defects in ER morphology. Even small perturbations, such as extending the length of the hydrophobic core by a single residue or transposing two adjacent residues, were sufficient to abolish translocon function. We conclude that residues at the C-terminal end of the Sss1p tail-anchor sequence are critical for translocon architectural integrity and function and that translocon function places severe constraints on the stability of the assembled complexes.
A low-copy YCp (yeast centromere plasmid) vector with a HIS3 marker (pRS313 ) was used to express WT (wild-type) and mutant Sss1p proteins. All mutant coding sequences were initially constructed in psPUTK (Stratagene) or a derivative thereof and later subcloned into a pRS313 derivative containing the endogenous promoter and 3′-UTR (untranslated region) of Sss1p. All plasmid sequences are available upon request.
Viability assays, antibodies and immunoblotting
Saccharomyces cerevisiae haploid strain FKY173 (MATα sss1:: URA3 leu2-3,-112 ade2-1 ura3-1 his3-11,-15 trp1-1 can1-100) [pFKp106 (PGAL10-CYC1-SSS1-ADE2)] was a gift from François Képès (GENOPOLE, Evry, Ile de France, France) . Strains expressing WT and mutants of Sss1p were created by transforming FKY173 with the plasmids described above and grown as described in the Supplementary Experimental section at http://www.BiochemJ.org/bj/436/bj4360291add.htm. Viability assays were performed as follows. Yeast strains expressing Sss1p mutants were grown in 0.67% synthetic dropout medium (supplemented with 2% galactose and 0.2% amino acids) without adenine and histidine overnight in a rotating incubator at 30 °C. A total of one D600 unit of cell culture was centrifuged (2300 g, 5 min) and resuspended in 1 ml of distilled water. From this cell suspension, four 10-fold serial dilutions were prepared. A total of 5 μl from each of the five cell suspensions was then spotted on to selective solid medium plates (containing 2% dextrose as the carbon source) and incubated at 30 °C for 2–3 days.
Protein samples were separated by SDS/PAGE (10% Tricine gel) and transferred on to a nitrocellulose membrane. Immunodetection was performed using antibodies as described previously  and in the Supplementary Experimental section, visualized using the Western Lightning Chemiluminescence Reagent Plus kit (PerkinElmer) and the decorated blots were exposed to Bioflex Econofilm (Clonex). Exposed immunoblots within the linear range were quantified with ImageQuant 5.0 Software (Molecular Dynamics). Data were subjected to statistical analysis using the Student's t test.
Purification of post-nuclear membranes
Yeast strains were grown in selective liquid medium containing dextrose at 30 °C for 12 h from a starting D600 of 0.1–0.15 to a final D600 of between 0.5 and 2.0. Cells were centrifuged in a Beckman JA-10 rotor at 2700 g for 5 min, washed once with 200–300 ml of distilled water and re-centrifuged as before. Cells were suspended in 7 ml/g of wet mass of 100 mM Tris/HCl, pH 8.4, and 10 mM DTT (dithiothreitol) and pre-incubated at 30 °C in an orbital shaker at 150 rev./min for 10 min. Cells were centrifuged at 5000 g for 5 min in an Eppendorf model 5084 instrument, resuspended in 7ml/g of wet mass of zymolase buffer [10 mM Tris/HCl, pH 7.4, 0.5% dextrose, 0.7 M sorbitol and 1 mM DTT in 0.75× YP medium (1% yeast extract and 2% peptone)], and incubated with Zymolase-20T (Seikagaku) at 5 mg per g of wet mass of cells for 60–90 min at 30 °C in an orbital shaker at 150 rev./min. Spheroplasts were centrifuged at 1000 g for 10 min, resuspended in 2 ml/g of ice-cold SKEEM buffer [5 mM Mes/KOH, pH 5.5, 1 M sorbitol, 0.5 M EDTA, 1 mM KCl, 0.1% ethanol, 1 mM PMSF, (Sigma–Aldrich), 5× CPIC (Complete™ protease inhibitor cocktail; Roche) and 1 mM iodoacetamide (Sigma–Aldrich)], transferred to a pre-chilled stainless steel hand-held homogenizer and disrupted with 30 strokes. Homogenized spheroplasts were centrifuged for 10 min at 1000 g at 4 °C in a Beckman JA25.5 rotor to pellet intact cells and nuclei. This last step was repeated and the resulting supernatant was centrifuged at 35 000 rev./min in a Beckman Ti 50.2 rotor for 1 h at 4 °C. Pellets containing post-nuclear membranes were resuspended in 20 mM Tris/HCl, pH 7.4, 5 mM magnesium acetate, 250 mM sucrose and 1mM PMSF and stored at a concentration of 100–200 A280 units/ml at −80 °C.
Ribosome-associated membrane protein assay
The ribosome-associated membrane protein assay was performed as follows: 40 equivalents (1 equivalent=1 μl of a solution of 50 A280 units/ml) of post-nuclear membranes were solubilized in a final volume of 400 μl of TAG-M buffer [20 mM Tris/HCl, pH 7.4, 500 mM aminocaproic acid (Sigma–Aldrich), 5 mM magnesium acetate, 10% (v/v) glycerol, 1% Triton X-100 (BDH), 1× CPIC (EDTA-free) and 1 mM PMSF], and homogenized in a 2 ml Dounce Pestle B, (Kontes) for 20 strokes. The sample was then incubated on ice for 15 min; 20 μl was removed for total products, and the remaining sample was loaded on to a 10–30% linear sucrose gradient and centrifuged at 41000 rev./min for 5 h in a Beckman SW41Ti rotor at 4 °C. Linear gradients were prepared as described previously , with the following modifications: 3.35 ml of 10, 20 and 30% (w/v) sucrose solutions in TAG-M buffer were layered in a centrifuge tube and allowed to diffuse overnight at room temperature (22–25 °C). Twelve equal fractions were collected, and the proteins were precipitated with a final concentration of 15–20% trichloroacetic acid, subjected to electrophoresis by SDS/PAGE (10% Tricine gel) and transferred on to a nitrocellulose membrane. Immunoblots were decorated with antibodies against Sec61p, Sss1p and L3 (see above). Fractions containing the majority of L3 protein (5–10 inclusive) were quantified and expressed as a percentage of the total protein in all fractions except fraction 12, which consisted primarily of aggregated protein.
Sucrose-density-gradient separation of digitonin-soluble Sec complexes
Digitonin-soluble Sec complexes were prepared by solubilizing 50 equivalents of post-nuclear membranes in 100 μl of 50 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 2.5% (w/v) digitonin (Sigma–Aldrich), 10 mM EDTA, 5 mM 2-mercaptoethanol, 1 mM PMSF and 5× CPIC. Next, the mixture was homogenized with 20 strokes in a 2 ml Dounce Pestle B and incubated on ice for 15 min. After removal of 10 μl of the sample for a pre-centrifugation control, the remaining sample was layered on to a 25–35% linear sucrose density gradient and centrifuged for 16 h at 50000 rev./min in a Beckman TLS-55 rotor at 4 °C. Thirteen fractions of 158 μl each were collected and processed as described above except that immunoblots were also probed for Sec63p, Sec62p and Sec72p. Linear gradients were prepared as described above using 1 ml each of 25% and 35% (w/v) sucrose solutions prepared in 50 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 1 mM EDTA, 0.1% digitonin, 5 mM 2-mercaptoethanol, 1 mM PMSF and 1×CPIC, and allowed to form for 2 h at room temperature. Triton-soluble Sec61 complexes and Sec62p–Sec63p–Sec71p–Sec72p complexes were also resolved in this manner by substituting 1% Triton X-100 for 2.5% (w/v) digitonin in the solubilization buffer, and 0.1% Triton X-100 for 0.1% digitonin in the linear gradients.
Sucrose-density-gradient separation of Triton-soluble Sec61 complexes
Triton-soluble Sec61 complexes were prepared and assayed as follows: 50 equivalents of post-nuclear membranes were solubilized in 100 μl of 50 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 1% Triton X-100, 10 mM EDTA, 5 mM 2-mercaptoethanol, 1 mM PMSF and 5×CPIC, agitated by vortex-mixing at full speed for 15 s and incubated on ice for 15 min. A total of 10 μl was removed for a pre-centrifugation sample, and the remainder was layered on to a 0–15% linear sucrose density gradient and centrifuged for 16 h at 50000 rev./min in a Beckman TLS-55 rotor at 4 °C. Fractions were collected and processed as described above. Linear gradients were prepared as described above using 500 μl each of 0, 5, 10 and 15% (w/v) sucrose solutions prepared in 50 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 1 mM EDTA, 0.1% Triton X-100, 5 mM 2-mercaptoethanol, 1 mM PMSF and 1× CPIC.
Post-nuclear membranes from yeast strains expressing Myc-tagged Sss1pWT or Sss1pYG [where YG indicates inversion of glycine and tyrosine at the carboxy end of the TMS (transmembrane sequence)] and from strains expressing Sss1pWT, Sss1pTMSa and Sss1pCTSa (where CTS is C-terminal sequence) were first salt extracted at a concentration of 0.050 A280 units/μl in 25 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 10 mM EDTA, 1 mM iodoacetamide and 1× CPIC for 30 min on ice and then centrifuged at 60000 rev./min for 10 min in a Beckman TLA120.2 rotor. Membranes totalling 200–800 equivalents were resuspended in 400 μl of 50 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 2.5% (w/v) digitonin, 10% (v/v) glycerol, 10 mM EDTA, 5 mM 2-mercaptoethanol, 1 mM iodoacetamide and 5× CPIC, processed in a 2 ml Dounce Pestle B for 20 strokes, and allowed to incubate on ice for 30 min. The sample was pre-cleared at 72000 rev./min for 15 min in a Beckman TLA120.2 rotor to remove insoluble complexes. A total of 300 μl of the cleared sample was loaded directly on to a Superdex 200 HR 10/30 column (or Superose 6 HR 10/30 column for WT, CTSa and TMSa samples) (GE Healthcare) pre-equilibrated in 50 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 0.1% digitonin, 1 mM EDTA, 5 mM 2-mercaptoethanol and 1× CPIC and run at 0.2 ml/min on an ÄKTA FPLC System (GE Healthcare) at 4 °C. A total of 50 0.5 ml fractions were collected, and the proteins were precipitated with trichloroacetic acid, separated by SDS/PAGE (10% Tricine gel) and transferred on to nitrocellulose membranes. Immunoblots were decorated with anti-Myc antibodies and the bands were quantified as described above.
ConA (concanavalin A) binding assay
A total of 100 equivalents of post-nuclear membranes were recovered by centrifugation at 50000 rev./min for 5 min at 4 °C in a Beckman TLA100 rotor, resuspended in 400 μl of 25 mM Hepes/KOH, pH 7.5, 500 mM potassium acetate, 50 mM EDTA and 1×CPIC and incubated on ice for 15 min. Membranes were again recovered by centrifugation at 50000 rev./min for 5 min at 4 °C in a Beckman TLA120.2 rotor and this step was repeated two more times. After the final wash, membranes were resuspended in 200 μl of 50 mM Hepes/KOH, pH 7.5, 400 mM potassium acetate, 2.5% digitonin or 1% Triton X-100, 10% (v/v) glycerol, 5 mM 2-mercaptoethanol, 1 mM PMSF and 1× CPIC, homogenized in a 2 ml Dounce Pestle B for 20 strokes and placed on ice for 15 min. A 150 μl aliquot of the sample was pre-cleared by centrifugation at 65000 rev./min for 10 min at 4 °C in a Beckman TLA100 rotor to remove insoluble components, and 100 μl of the resulting supernatant was added to 450 μl of ConA buffer [50 mM Hepes/KOH, pH 7.5, 293 mM potassium acetate, 0.5 mM MgCl2, 0.5 mM MnCl2, 0.5 mM CaCl2, 5 mM 2-mercaptoethanol, 10% (v/v) glycerol, 1× CPIC and 1 mM PMSF] and 50 μl of packed beads of ConA–Sepharose-4B (Sigma–Aldrich) equilibrated in ConA buffer. The mixture was incubated for 1 h at 4 °C on an end-over-end rotator after which the beads were recovered by centrifugation at 2500 g for 2.5 min in a microcentrifuge. The supernatant (containing unbound Sec61 complexes) was transferred to a fresh tube and unbound proteins were precipitated with trichloroacetic acid. The recovered beads were washed three times in 1 ml of ConA buffer plus 1 mM PMSF, resuspended in 300 μl of 50 mM Hepes/KOH, pH 7.5, 500 mM NaCl, 1% Triton X-100, 10% (v/v) glycerol, 0.5 mM MgCl2, 0.5 mM MnCl2, 0.5 mM CaCl2, 5 mM 2-mercaptoethanol and 1 mM PMSF, and incubated for 30 min on an end-over-end rotator at 4 °C. After centrifugation at 2500 g for 2.5 min in a microcentrifuge, the supernatant (containing bound Sec61 complexes) was precipitated with trichloroacetic acid and the beads (containing bound Sec complexes) were washed two times in ConA buffer plus 1 mM PMSF and once in ConA buffer containing 100 mM NaCl (instead of potassium acetate) plus 1 mM PMSF. Precipitated samples and beads were resuspended in 30 μl of SDS/PAGE loading buffer (10% SDS, 0.1 M Tris/HCl, pH 8.9, 2 mM EDTA, 0.1% Bromophenol Blue, 20% glycerol and 0.5 M DTT) and analysed by SDS/PAGE (10% Tricine gel) and immunoblotting with the indicated antibodies.
Yeast viability is compromised by the expression of Sss1p with mutations in the CTS and TMS
Mutants of Sss1p were constructed with deletions or substitutions in the three sub-regions of the tail anchor [NTS (N-terminal sequence), TMS and CTS] in order to determine the effect on cell viability (Figure 1A). Each Sss1p variant was expressed under the control of the SSS1 promoter and transformed into yeast strain FKY173, which lacks a chromosomal copy of the SSS1 gene. The viability of this strain is maintained by galactose-inducible expression of WT Sss1p; therefore, after transformation with an engineered Sss1p expression vector and growth on glucose, only the mutant Sss1p is expressed. Cell viability therefore, reflects the function of the mutant.
Mutations in the TMS and CTS of the tail anchor of Sss1p result in growth defects in yeast
The NTS of Sss1p is composed of an amphipathic helix with five lysine residues on one face. Strains expressing Sss1p with all five of the lysine residues mutated to threonine (Sss1pK5T5) showed no change in viability compared with the WT (Sss1pWT) strain (Figure 1B). In contrast, mutations in both the TMS and CTS resulted in a marked decrease in growth. Inversion of two amino acids (glycine and tyrosine) at the carboxy end of the TMS (Sss1pYG) dramatically inhibited growth of the yeast. The Sss1pYG mutant includes three other conservative mutations that in control experiments were shown not to affect growth (results not shown). Comparison of the growth of Sss1pYG with Sss1pTMSa and Sss1pCTSa revealed that inversion of the GY with a YG had the same effect on viability as complete replacement of the TMS or CTS respectively (Figure 1). Remarkably, insertion of a single phenylalanine residue at the carboxy end of the TMS (Sss1pF) resulted in even more severe growth defects similar to substitution of the entire tail-anchor region with one that normally targets mitochondria (Sss1pTOM5, where TOM is translocase of the mitochondrial outer membrane; Figure 1). Although yeast expressing Sss1pYG or Sss1pF failed to grow, the strains were not dead even 24 h after shutting off expression of Sss1p as induction of the WT gene with galactose completely rescued growth of the strains (results not shown).
Replacement of the entire CTS (Sss1pCTSa) or the transmembrane region (Sss1pTMSa) did not prevent protein integration into the ER membrane (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/436/bj4360291add.htm). We also examined the amount of each of the mutant proteins in each of the strains. Indeed, the mutants were expressed at slightly higher levels than the WT proteins. Additional control experiments for selected constructs demonstrated that the effects reported in the present paper were not related to the expression level of the different mutants (results not shown).
Mutations in the tail anchor of Sss1p result in translocation defects
It was reported previously that depletion of Sss1p or deletion of the last 12 amino acids of the protein results in the accumulation of secretory proteins in the cytoplasm of yeast . We obtained similar results in the strain FKY173 after the cells had been switched to glucose-containing medium (Figure 2A). Sss1p was no longer detectable by immunoblotting at 6 h and pKar2p (pre-Kar2p) and pPαFactor (pre-pro-α factor) began to accumulate at 6–8 h. Both Kar2p and pro-α factor are processed from the precursor to the mature form by signal peptidase. Accumulation of the precursor could result from faulty signal cleavage and retention of the precursors pKar2p and pPαFactor in the ER, and/or from a defect in translocation resulting in accumulation of the precursors in the cytosol. Our results strongly suggest that there are translocation defects, as precursors begin to accumulate before there is substantial reorganization of the ER. Moreover, there is a corresponding defect in ribosome binding to Sec61 (see below) that would result in the synthesis of membrane and secreted proteins in the cytoplasm. There was a slight increase in the expression level of Sec61p over this time course, whereas the expression level of mitochondrial porin [also called VDAC (voltage-dependent anion channel)] remained constant. Sec61p induction may reflect an attempt by the cells to compensate for reduced translocation, or it may reflect induction of the unfolded protein response . Consistent with this second hypothesis, Kar2p levels also increased significantly over the same time course.
Depletion of WT Sss1p and expression of mutant tail-anchor proteins result in translocation defects and altered ER morphology
In order to determine whether the tail-anchor variants could repair the translocation defect, the FKY173 strain was transformed with either an empty vector or plasmids expressing the WT or one of the Sss1p tail-anchor mutants. The resultant strains were then grown in glucose for 8 h in order to maximally repress the expression of WT Sss1p. We discovered that cells expressing each of the tail-anchor mutants, except Sss1pK5T5, accumulated both pKar2p and pPαFactor, as compared with WT Sss1p. To rule out that low expression of the Sss1pK5T5 mutant was responsible for the the lack of accumulation of pre-proteins, Myc-tagged Sss1pK5T5 was expressed from a high-copy-number vector in the same strain. In these cells, expression of Myc-tagged Sss1pK5T5 was higher than expression of WT Sss1p and roughly equal to that of the other mutants, did not result in growth defects and showed minimal accumulation of pre-proteins (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/436/bj4360291add.htm). It has been shown previously that expression of Myc-tagged WT Sss1p has no effect on yeast cell growth . Under the present conditions, Sec61p and Sbh1p levels remained relatively constant (Figure 2B), suggesting that translocation defects do not arise from a loss of the other components of the Sec61 complex. Interestingly, most of the Sss1p variants that were unable to repair the translocation defect were present at significantly higher levels than the WT and K5T5 proteins, indicating that the mutants are unable to compensate for the loss of Sss1p even when overexpressed. In addition, because pPαFactor has been shown to translocate post-translationally, whereas Kar2p can translocate in a co- and post-translational manner [35,36], these results suggest that translocation defects occur in both of these pathways (also see below).
The expression of tail-anchor mutants of Sss1p alters ER morphology
Efficient trafficking between the ER and Golgi complex and ribosome binding to the Sec61 complex are important to maintain ER structure . To determine whether depletion of Sss1p and expression of tail-anchor mutants affected ER morphology, the yeast strain FKY173 transformed with an empty vector or transformed with vectors expressing WT Sss1p or Sss1pYG were co-transformed with a vector expressing two fluorescent reporters: GFP (green fluorescent protein) fused to cyb5TA (tail anchor of yeast cytochrome b5) (GFP–cyb5TA), and the SA (signal anchor) of the β-subunit of the canine SRP (signal recognition particle) receptor fused to the N-terminus of mRFP (monomeric red fluorescent protein) (SA–mRFP). Both proteins are targeted to the ER in WT yeast (Supplementary Figure S1B), but localization of GFP fused to the cytochrome b5 tail anchor does not require a functional Sec61p-based translocon (M. P. Henderson, unpublished work). Thus the distribution of these proteins reports on ER structural changes, which may occur upon Sss1p depletion or mutation. We observed that depletion of Sss1p resulted in considerable changes to both the perinuclear as well as the peripheral ER structure. Nevertheless, the SA–mRFP and the GFP–cyb5TA reporters continued to co-localize (compare Figure 2C with Supplementary Figure S3 at http://www.BiochemJ.org/bj/436/bj4360291add.htm) and co-fractionate with an ER marker (Supplementary Figure S1B), demonstrating that the alteration was due to a change in ER morphology. At 3 h, the expression of Sss1pYG had little effect on the ER structure, since at least 6 h of expression under the GAL promoter is required to shut down the expression of WT Sss1p (Figure 2C, YG at 3 h). At 16 and 24 h, when only the mutant Sss1p proteins are present, radical structural changes are evident in both the perinuclear and peripheral ER. At these later time points, both of these compartments coalesce into very dense structures in strains containing both the YG mutant and the empty vector (Figure 2C; compare YG with vector at 16 and 24 h), but not the WT. Time-lapse imaging revealed that, unlike WT yeast (see Supplementary Movie S1 at http://www.BiochemJ.org/bj/436/bj4360291add.htm), cells expressing Sss1pYG were unable to bud, and by 21 h all of the ER had collapsed (see Supplementary Movie S2 at http://www.BiochemJ.org/bj/436/bj4360291add.htm).
Tail-anchor mutations decrease ribosome binding to the Sec61 complex
Binding of heterotrimeric Sec61 complexes to the ribosome–nascent chain complex is required for co-translational translocation in eukaryotes . Therefore we used a quantitative assay to determine for selected Sss1p tail-anchor mutants whether the defect in co-translational translocation was due to impaired ribosome binding to Sec61 complexes (Figure 2B). To this end, membranes from strains expressing WT and the tail-anchor mutants were solubilized in a low-ionic-strength buffer in order to preserve ribosome interaction. The lysate was then resolved by sucrose-density-gradient centrifugation to determine the degree of Sec61 complex–ribosome binding. Complexes separated by this method are almost completely soluble (see Supplementary Figure S4A at http://www.BiochemJ.org/bj/436/bj4360291add.htm); a relatively minor amount of each protein is present at the bottom (last fraction) of the gradient. Immunoblot analysis of the gradient samples showed that both Sec61p and Sss1p co-fractionated with the yeast 60S ribosomal protein L3 in membranes expressing Sss1pWT (Figure 3A, lanes 5–10). In gradient samples from membranes expressing Sss1pYG and Sss1pF, however, a considerable amount of Sec61p and Sss1p failed to co-fractionate with the L3 peak fractions, but instead fractionated at the top of the gradient, indicating reduced, but not abolished, binding of Sec61 complexes to ribosomes. A similar effect was seen for the TMSa and CTSa mutants (Supplementary Figure S4B). In a previous study, replacement of the entire tail-anchor sequence of Sss1p with that of UBC6 (ubiquitin-conjugating enzyme 6) reported translocation defects, yet pre-proteins could still be cross-linked to Sec61p . The partial defect in ribosome binding observed in the present study is consistent with this result and suggests that the defect underlying both observations involves the structure and/or function of the Sec61 complex.
Sec61 complexes isolated from strains expressing mutant Sss1p proteins are defective for ribosome binding
Gradient samples from Sss1pYG, Sss1pF, Sss1pTMSa and Sss1pCTSa also showed partial proteolysis of Sss1p, particularly for Sss1p that did not co-migrate with ribosomes (Figure 3A and Supplementary Figures S4B and S4C; a faster-migrating form of Sss1p is denoted with an asterisk). Degradation appears to occur during gradient centrifugation, as the amount of this product is minimal at the time of solubilization (Supplementary Figure S4C). These results are consistent with studies indicating that the Sec61 complex is more susceptible to proteolysis when it is not protected by bound ribosomes . Consistent with this interpretation, WT Sss1p shows no susceptibility to degradation during gradient centrifugation. Quantitative analysis of ribosome binding for the Sss1pYG, Sss1pF, Sss1pTMSa and Sss1pCTSa mutants indicated that binding of Sec61 complexes to ribosomes was compromised by 20–50% relative to the WT control (Figure 3B).
Heptameric Sec complexes containing tail-anchor mutants remain intact, but are unstable
The heptameric Sec complex, consisting of the heterotrimeric Sec61 complex and the tetrameric Sec62p–Sec63p–Sec71p–Sec72p complex, is required for post-translational translocation in yeast [10,40]. Thus these complexes can function without binding ribosomes. It has been shown that solubilization of yeast microsomes in digitonin preserves the interaction between these sub-complexes, whereas they are dissociated after solubilization in Triton X-100 . To investigate whether the post-translational translocation defect in strains expressing the Sss1p tail-anchor mutants resulted from impaired assembly of the heptameric Sec complex into its sub-complexes, membranes from strains expressing WT and tail-anchor mutants were solubilized in digitonin or Triton X-100 and centrifuged through a 25–35% sucrose density gradient. Immunoblots were probed for Sss1p and Sec61p, which form part of the Sec61 complex, and for Sec62p, Sec63p and Sec72p. The amount of each product was quantified, expressed as a percentage of the total, and the results were plotted against the fraction number to generate gradient profiles for each protein. In control samples solubilized with Triton X-100, we found that profiles of complexes derived from yeast expressing Sss1pWT resulted in two major peaks. The first peak (Figure 4A, fraction 2), contained Sss1p and Sec61p and migrated at the size expected for the heterotrimeric Sec61 complex (~70 kDa). The second peak, enriched for Sec62p, Sec63p and Sec72p, is consistent with the size of the Sec62p–Sec63p–Sec71p–Sec72p sub-complex (~160 kDa) (Figure 4A, fraction 3). This second peak also contained a significant fraction of the total amount of Sec61p.
Sec complexes containing tail-anchor mutants are intact, but altered, when assayed by sucrose-density-gradient centrifugation
As anticipated, WT complexes solubilized in digitonin fractionated as one major peak at approximately 230 kDa, consistent with a heptameric complex (Figure 4B, fraction 7). This peak contained all of the examined members of the Sec complex. These results confirm previous observations suggesting that the integrity of the Sec complex is preserved in digitonin, but not in Triton X-100 . Digitonin-solubilized complexes isolated from yeast expressing the mutant proteins Sss1pYG and Sss1pF resulted in a distribution similar to the larger Sec complexes, but spread out over more fractions (Figures 4C and 4D). Moreover, individual proteins peaked in different fractions (Figures 4C and 4D), and a large portion of some of the proteins was found at the bottom of the gradient (compare fraction 13 in Figures 4B–4D, and see Supplementary Figure S5 at http://www.BiochemJ.org/bj/436/bj4360291add.htm). Gradient profiles for complexes derived from yeast expressing Sss1pTMSa, Sss1pCTSa and Sss1pTOM5 tail-anchor mutants showed similar results to those of the Sss1pYG and Sss1pF mutants (Supplementary Figure S5). The heterogeneous nature of the samples on density gradients suggests that the compositions of the Sec complexes solubilized in digitonin were not uniform, consistent with either impaired assembly or reduced stability of complexes containing the Sss1p mutants.
The oligomeric state of digitonin-soluble Sec complexes was also investigated by gel-filtration chromatography using Myc-tagged versions of Sss1pWT and Sss1pYG (Figure 4E). Similar to the results obtained with sucrose density gradients, WT complexes eluted primarily as a single peak, whereas complexes from the YG mutant (Figure 4E) and the other mutants except for Sss1pK5T5 (see Supplementary Figure S6A at http://www.BiochemJ.org/bj/436/bj4360291add.htm), had a much broader size distribution. Not surprisingly, the elution profile for the K5T5 mutant was indistinguishable from that of Sss1pWT. To directly compare the sizes of complexes eluting from the gel-filtration column for each mutant, the Stokes radii at 50% peak height (at the upper and lower ends of elution) were calculated for the Sss1pK5T5, Sss1pTMSa, Sss1pCTSa and Sss1pTOM5 mutants (Supplementary Figure S6B). The results reveal that complexes containing mutant Sss1p proteins had a greater variation in size and shape than those formed from WT protein and that the Sec complexes containing Sss1p mutants were less soluble in the extraction buffer (Supplementary Figure S6C).
There is a discrepancy between the molecular masses of the major Sss1p-containing complexes estimated by the sucrose gradient and gel-filtration analyses: 200 kDa for density-gradient data compared with 400 kDa for gel-filtration data. This may be due to differences in the protein composition of the complexes, or the effects of osmolytes on detergent binding since the density gradient contained increasing concentrations of sucrose, whereas the gel-filtration buffer contained 10% glycerol. Another explanation for the difference is the possible existence of dimers of Sec61 heterotrimers that may have been stable during gel filtration, but that dissociated during density-gradient centrifugation. Consistent with this interpretation, there are several studies indicating that the Sec61 complex may exist as oligomers of the Sec61 heterotrimer [41–43]. Nonetheless, the results of both techniques provide evidence to suggest that detergent-soluble heptameric Sec complexes derived from yeast expressing Sss1p tail-anchor mutants are more heterogeneous in mass and shape than those derived from yeast expressing WT Sss1p. We propose that the amino acid sequence of the tail-anchor sequences affects the final assembly of the Sec61 heterotrimer into Sec complexes.
The stability of Sec complexes containing mutant tail-anchor proteins was further investigated by affinity separation with ConA. In WT membranes, some of the Sec61 complexes form part of the heptameric Sec complex, which remains intact in digitonin and binds to ConA due to the presence of oligosaccharides on the Sec71p component [10,11]. In agreement with previously published results , when WT salt-extracted post-nuclear membranes were solubilized in digitonin and incubated with ConA–Sepharose, approximately half of the Sec61 complexes (represented by Sec61p and Sss1p) were found in the supernatant fraction (Figure 5A, lane 4, ‘free’). Washing the ConA beads with a high-salt buffer disrupted the interaction between the Sec61 complex and the Sec62p–Sec63p–Sec71p–Sec72p sub-complex. As a result, the Sec61 complexes were released into the high-salt wash (Figure 5A, lane 5, ‘bound’). This fraction of Sec61p represents the Sec61 complexes that were part of the heptameric Sec complex. In contrast, Sec63p and Sec72p formed part of the Sec62p–Sec63p–Sec71p–Sec72p complex and remained bound to ConA even in high salt (Figure 5A, lane 6). In Triton X-100, most of the Sec61 complex dissociateed from the Sec62p–Sec63p–Sec7p–Sec72p sub-complex, leaving the latter bound to ConA. This resulted in only a very small amount of Sec61p in the salt wash (Figure 5A, lane 2). Therefore exposure to Triton X-100 mimics the result expected if mutation of Sss1p destabilizes the Sec complex.
Digitonin-soluble Sec complexes prepared from membranes from mutant Sss1p strains are unstable when assayed by binding to ConA–Sepharose
On the basis of the results presented above, we used the digitonin-solubilization/ConA assay to determine the stability of the interaction between Sec61 complexes and the Sec62p–Sec63p–Sec71p–Sec72p sub-complex in strains expressing the Sss1p mutants. As expected from the results presented above, digitonin-solubilized complexes from strains expressing the Sss1pTMSa, Sss1pCTSa and Sss1pTOM5 mutants showed reduced solubility compared with complexes from strains expressing WT Sss1p (see Supplementary Figure S7A at http://www.BiochemJ.org/bj/436/bj4360291add.htm). To our surprise, ConA-binding experiments using complexes containing the Sss1pYG and Sss1pF mutant proteins produced results in which Sec61p was present in the salt wash, similar to the WT control, but Sss1p was detected in the unbound fractions. This result suggests that some fraction of Sec61p remains bound to the Sec62p–Sec63p–Sec71p–Sec72p complex in digitonin (Figure 5A; compare the amounts of Sec61p in lanes 8 and 11 with lane 2). However, the Sss1p mutants are no longer part of the Sec complexes under these conditions (Figure 5A; compare the amounts of Sss1p in lanes 8 and 11 with lane 5). Quantification of the results of multiple experiments showed that the amount of Sec61p bound to ConA in digitonin lysates was significantly reduced for the Sss1pF, Sss1pTMSa, Sss1pCTSa and Sss1pTOM5 mutants, but not for the YG mutant (Figure 5B and Supplementary Figure S7B). These results indicate that Sec complexes containing these mutant Sss1ps are less stable than WT complexes under the assay conditions. Moreover, it appears that the mutant Sss1ps are not as stably bound within the Sec complex as Sec61p and that the presence of the mutant Sss1p protein affects the stable association of the Sec61 complex with the Sec62p–Sec63p–Sec71p–Sec72p complex. We also noticed that complexes isolated from yeast expressing Sss1pF contained substantially less Sec63p (Figure 5), which is also consistent with destabilization of Sec complexes by the Sss1pF mutant and may explain why the effect of the phenylalanine insertion on cell growth is even more severe than the YG mutation.
Tail-anchor mutants of Sss1p form unstable heterotrimeric Sec61 complexes
To determine whether the reduced binding of the Sss1p mutants to the Sec complexes was due to an altered composition or conformation of the heterotrimeric Sec61 complex, membranes were solubilized in Triton X-100 and resolved by centrifugation through a 0–15% linear sucrose density gradient. After centrifugation, gradient samples were probed for each of the three components of the Sec61 complex: Sec61p, Sss1p and Sbh1p. Unlike heptameric Sec complexes, heterotrimeric Sec61 complexes are stable in Triton X-100 . Thus, as expected, the three proteins co-migrated when the complex prepared from WT membranes was examined (Figure 6, lanes 5–9). In samples derived from yeast expressing either the Sss1pYG or Sss1pF mutant, both Sss1p and Sbh1p exhibited decreased sedimentation and were found near the top of the gradient, whereas Sec61p sedimented towards the bottom (Figure 6, lanes 2–4), suggesting protein misfolding and aggregation. Similar results were obtained for complexes derived from strains expressing the Sss1pTMSa, Sss1pCTSa and Sss1pTOM5 mutants (see Supplementary Figure S8A at http://www.BiochemJ.org/bj/436/bj4360291add.htm). Quantification of both Sec61p and Sss1p in all of the mutant strains exhibiting translocation defects showed significant disruption of heterotrimers by Triton X-100 (Supplementary Figure S8C). Taken together, these results indicate that heterotrimeric Sec61 complexes are less stable when they contain mutant Sss1p proteins. This hypothesis is also consistent with the noted proteolytic processing of the Sss1p mutants, which occurs during gradient centrifugation (denoted by the asterisk in Figure 6; see also Supplementary Figure S8A). Because the Sss1p mutants were intact when they were first solubilized (Supplementary Figure S8B), proteolysis must occur during gradient centrifugation.
Tail-anchor mutants of Sss1p form Sec61 complexes unstable in Triton X-100 when assayed by low-percentage sucrose-density-gradient centrifugation
Taken together, our results indicate that even subtle mutations in the tail-anchor sequence of Sss1p affect the integrity and function of the translocon. More generally, these results indicate that even the hydrophobic core of tail-anchor sequences can affect the assembly and stability of integral membrane protein partners. This result is noteworthy because most attention has been paid to the role of the tail-anchor sequence in protein targeting [20,45,46]. Although the mutations we created did not reduce the hydrophobicity of the transmembrane core of Sss1p and therefore did not alter ER targeting, the stability of the Sec and Sec61 complexes was still compromised. Moreover, this change in stability results in a severe defect in the translocation function of the Sec61 complex and possibly other defects that lead to complete reorganization of ER structure (Figure 2C and Supplementary Figure S3). This strongly suggests that the primary sequence of this region is involved in complex integrity.
It is surprising that a small change to the primary sequence of the transmembrane domain of Sss1p could have such a dramatic effect on ER function. With the exception of binding sites within membrane transporters, most transmembrane segments tend to be relatively insensitive to modest changes in primary sequence; the main role of these segments is to anchor the protein within the lipid bilayer. This is especially true when one considers the tail-anchor protein family. As a result, membrane-spanning segments are viewed as requiring a specific length and hydrophobicity, but are often considered rather generic in terms of primary amino acid sequence . Our results instead suggest that there can be very strict sequence requirements for the hydrophobic sequence of a tail-anchored protein. In contrast, even non-conservative mutations of the sequence N-terminal to the transmembrane domain, which is soluble, had no effect on translocon function.
Studies of the prokaryotic SecYEG–SecA complex reveal interesting interpretations of the results described in the present paper. Mutagenic studies of the SecE protein, which is the bacterial homologue of Sss1p, support the hypothesis that the membrane domain of Sss1p may have functions beyond that of a simple membrane anchor. Although the Escherichia coli protein contains three transmembrane domains, whereas archaea and eukaryotes have only one, the greatest extent of conservation between them lies in the third transmembrane domain and its flanking regions. However, even in this region, the consensus sequence for the SecE region includes a 28 residue hydrophobic region containing only a single charged residue. In eukaryotes, the hydrophobic core of Sss1p is only 21 residues flanked by hydrophilic sequences (Supplementary Figure S8). Deletion analysis of E. coli SecE shows that only the third transmembrane domain and a portion of the conserved second cytoplasmic region are necessary for its function during protein transport . PrlG (pre-protein translocase SecE subunit) mutations that are located in the third transmembrane domain of SecE and are close to the mutations made in the present study have been shown to weaken the interactions between the SecY, SecE and SecG subunits. This loosened association enhances translocation rates and is thought to increase the insertion of SecA due to increased conformational flexibility of the SecYEG complex .
Sequence alignments of Sec61γ from a number of different species show that the G residue mutated in the YG variant is absolutely conserved (Figure 7). Comparison of the sequence of Sss1p with that of SecE proteins from bacteria and archaea (Figure 7 and Supplementary Figure S9B at http://www.BiochemJ.org/bj/436/bj4360291add.htm) shows that the GY residues mutated in the present study are close to the positions of two of the four PrlG mutations in SecE. However replacement of the conserved glycine residues with leucine in the same region of Sss1p (including the glycine in the YG substitution in the present study) had no effect on Sss1p function . In contrast, addition of the phenylalanine residue to Sss1p, which had an even more dramatic effect on Sec61 complex stability than the YG inversion, is at a non-conserved position that is frequently a phenylalanine in other organisms. Moreover, the number of residues within this region of the tail anchor is variable. Although changes to the primary sequence of the transmembrane domain of SecE have been proposed to change the positions of the α- and γ-subunits with respect to one another, ultimately altering the stability and function of the entire complex [49,50], mutations in the TMS of yeast Sss1p serve to disrupt translocation rather than enhance it, as is the case in bacteria. This may be due to the differences between the SecYEG and Sec61 complexes. Crystal structures of SecA-bound SecYEG complexes superimposed on to ribosome nascent chain-bound Sec61 complexes show differing conformations; neither lateral gate opening nor plug movement was observed when SecA was bound. Also, comparison of the crystal structures of the two complexes shows that SecE is less structured than Sss1p and not as tightly associated with the complex . These observations suggest that, although homologous, Sss1p and SecE play unique roles in their interactions with the translocon.
Sec61γ is highly conserved
Nevertheless, there is evidence from both eukaryotic and bacterial studies that the transmembrane region of the Sec61 γ-subunit is in close proximity to several membrane helices of the Sec61 α-subunit, and the γ-subunit is required for complex stability. A representation of the yeast Sec61 complex , as modelled on the crystal structure of the Methanocaldococcus jannaschii SecYEβ translocon , shows that the transmembrane helix of Sss1p spans the membrane plane at a ~35° angle , and one side appears to be in contact with Sec61p membrane helices. Studies in yeast showed that strains expressing combinations of Sec61p polypeptides that were unable to form stable complexes were partially rescued by overexpression of Sss1p, providing evidence that Sss1p is required for complex stability . Perhaps the close proximity of the membrane helix of the γ-subunit to various regions of the α-subunit confers complex stability and thus constrains amino acid composition, rendering the protein sensitive to mutations within this region.
In our present study, substitution of the entire transmembrane domain of Sss1p results in unstable Sec61 and Sec complexes; however, these strains are not significantly more defective than strains expressing Sss1pYG or Sss1pF. This suggests that the C-terminal end of the hydrophobic sequence in Sss1p plays a role in the proper assembly of these complexes. Transmembrane helix rotational angle calculations  for the mutant proteins, Sss1pYG and Sss1pF, predict an increase in the α-helical angle when compared with the predicted angle of the WT helix. Analysis of the crystal structure of the yeast Sec61 channel  suggests that the region of Sss1p containing these highly conserved glycine residues may be in close proximity to the plug. Molecular-dynamics simulations of SecYEβ indicate that the translocon is a highly dynamic structure, accommodating both secretory proteins, which requires plug movement, as well as transmembrane proteins, which requires opening of the lateral gates . Therefore it remains somewhat surprising that replacement of GY with YG or the introduction of a phenylalanine residue some distance away would have such a profound effect on Sec-complex assembly. It may be that as a partner of Sec61p, Sss1p contributes to translocon dynamics. Thus small changes in amino-acid sequence may have profound effects by interfering with the required dynamics of the complexes. We speculate that small flexible residues allow the appropriate ‘fit’ between Sss1p and the neighboring helices of Sec61p. As the smallest side group, glycine is often found where main chains closely approach each other; it is more flexible than other residues, contributing to parts of the protein that need to move or act as hinges . It may also be that these residues contribute to an Sss1p interaction not seen in prokaryotes that requires the appropriate positioning of the charged region C-terminal of the hydrophobic region sequence. Consistent with this view, this charged region is not conserved in SecE proteins. Clearly, high-resolution structural data that display protein–protein contacts within this region will be particularly informative.
Complete™ protease inhibitor cocktail
tail anchor of yeast cytochrome b5
green fluorescent protein
monomeric red fluorescent protein
pre-protein translocase SecE subunit
translocase of the mitochondrial outer membrane
Domina Falcone, Matthew Henderson and Hendrik Nieuwland optimized the protocols and performed the experiments. Domina Falcone, Hendrik Nieuwland, Christine Coughlan, Jeffrey Brodsky and David Andrews designed the experiments and analysed the data. Domina Falcone, Jeffrey Brodsky and David Andrews prepared the figures and wrote the manuscript. David Andrews directed the project.
We thank Vicki Pierre for providing the data in Supplementary Figure S2(B) and acknowledge the undergraduate thesis students N. Usmani, J.M. Tkach and M. Campbell whose efforts contributed to the genesis of this project.
This work was supported by a grant from the Canadian Institute of Health Research [grant number FRN 10490] to D.W.A., who also holds a Canada Research Chair in Membrane Biogenesis, and by the National Institutes of Health [grant number GM75061 (to J.L.B.)].
Present address: Department of Biological Sciences, University of Denver, 2190 E. Iliff Avenue, Denver, CO 80208, U.S.A.