Unassisted translocation of large polypeptide domains across phospholipid bilayers

Abstract

Although transmembrane proteins generally require membrane-embedded machinery for integration, a few can insert spontaneously into liposomes. Previously, we established that the tail-anchored (TA) protein cytochrome b(5) (b5) can posttranslationally translocate 28 residues downstream to its transmembrane domain (TMD) across protein-free bilayers (Brambillasca, S., M. Yabal, P. Soffientini, S. Stefanovic, M. Makarow, R.S. Hegde, and N. Borgese. 2005. EMBO J. 24:2533-2542). In the present study, we investigated the limits of this unassisted translocation and report that surprisingly long (85 residues) domains of different sequence and charge placed downstream of b5's TMD can posttranslationally translocate into mammalian microsomes and liposomes at nanomolar nucleotide concentrations. Furthermore, integration of these constructs occurred in vivo in translocon-defective yeast strains. Unassisted translocation was not unique to b5 but was also observed for another TA protein (protein tyrosine phosphatase 1B) whose TMD, like the one of b5, is only moderately hydrophobic. In contrast, more hydrophobic TMDs, like synaptobrevin's, were incapable of supporting unassisted integration, possibly because of their tendency to aggregate in aqueous solution. Our data resolve long-standing discrepancies on TA protein insertion and are relevant to membrane evolution, biogenesis, and physiology.

Figure 1.

Schematic representation of the constructs used in this study. The opsin tag (corresponding to the first 19 amino acids of the bovine protein) is shown as an open box, with the N-glycosylation site represented by a hexagon. The numbers in the name of each construct indicate the total length of the lumenal sequence. Sequences deriving from b5 are depicted in black, those from PTP1B are depicted as striped boxes, and those from the cytoplasmic tail of VSVG are in light gray. The shaded gray box in panel b indicates sequences of different lengths deriving from bovine opsin or yeast Hsp150. b5-ops-47 carries the entire N-terminal lumenal sequence of bovine opsin, which has been duplicated and triplicated in b5-ops-85 and -125, respectively; Hsp-elongated constructs contain one, two, or four copies of a 19-residue repeated sequence of the yeast protein Hsp150. The constructs illustrated in panels a–d all have the TMD of b5; the constructs illustrated in panel e have an altered TMD. In b5-HH-ops-28, b5's TMD has been mutated to be more hydrophobic, and in b5-scrambled-ops-28, the order of amino acids in b5's TMD has been changed. In b5-Syb2-ops-28 and b5-Syb2mut-ops-28, it has been replaced by the one of Syb2 or by a mutated, less hydrophobic version thereof. The TMDs of b5 (a–d), PTP1B (f), and Syb2 or mutated TMDs of Syb2 and b5 (e) are indicated with different symbols. Sequences of the TMDs and lumenal domains are given in Tables I and S1 (available at http://www.jcb.org/cgi/content/full/jcb.200608101/DC1), respectively.

Figure 2.

Comparison between co- and posttranslational translocation efficiencies for b5 constructs with extended lumenal domains. (A) Illustration of the different hypothetical modes of insertion into RMs for the constructs in this study. The thickened part of the polypeptide chain represents the hydrophobic membrane-anchoring segment (see Results for further explanation). (B) b5-based constructs were incubated with or without RMs after (post) in vitro translation or with RMs during translation (co) as indicated. An aliquot of each sample was directly subjected to SDS-PAGE (top; −PK), whereas the remaining was digested with PK (bottom; +PK). (C) Co- or posttranslational translocation reactions for VSVG-elongated constructs. (B and C) Asterisks indicate the glycosylated form of both the full-length proteins (−PK) and the PFs (+PK), whereas the boxes indicate the corresponding nonglycosylated forms. For each construct, the total number of charged amino acids and the net charge of the lumenal sequence are reported in parentheses above the lanes. (D) Comparison between efficiencies of post- versus cotranslational translocation for the constructs illustrated in B and C (see Materials and methods).

Figure 3.

Extended C-terminal domains of b5 constructs are translocated across the ER membrane of yeast mutants defective in translocon function. Wild-type yeast cells (lanes 1 and 2) or yeast cells harboring the sec61-3 mutation transformed with the indicated b5 construct were incubated in low glucose medium for 1 h at 38°C and were ³⁵S labeled for 5 min at the same temperature. The cells were lysed and subjected to immunoprecipitation with antiopsin antibody (top) or CPY antiserum (bottom). The immunoprecipitates were divided in two, and one part was digested with EndoH before SDS-PAGE analysis as indicated. Asterisks and boxes indicate the glycosylated and nonglycosylated products, respectively.

Figure 4.

Posttranslational translocation of extended lumenal domains of b5 constructs occurs across ER microsomes and protein-free liposomes with equal efficiency. (A) Posttranslational translocation reactions were performed with PC or PC/phosphatidylethanolamine (PE; 4:1 ratio) liposomes, with RMs at equivalent phospholipid concentration (0.7 μg phospholipids/μl) or without added vesicles as indicated. The total number of charged amino acids and the net charge of the lumenal sequence are reported in parentheses above the lanes. Asterisks and boxes indicate the glycosylated and nonglycosylated products, respectively. (B) Time course of translocation into RMs of b5-ops-28 and -85 monitored by glycosylation.

Figure 5.

TMDs with increased hydrophobicity require a proteinaceous component of the ER for their insertion. (A and B) Co- or posttranslational translocations were performed with the indicated constructs, without added vesicles, or with RMs or PC liposomes (both at 0.7 μg phospholipids/μl). (A) Background bands generated after PK treatment in the absence of membranes are indicated with arrows (lanes 4 and 7). (B) Detergent controls for b5-Syb2mut-ops28 and b5-scrambled-ops-28 are shown (lanes 7 and 13). For the other constructs, detergent controls are shown in Fig. S1 A (available at http://www.jcb.org/cgi/content/full/jcb.200608101/DC1). (C) Analysis of RMs treated with different amounts of trypsin by immunoblotting. Digestion of the cytosolic region of ribophorin I results in a lower M_r band (arrows), which is revealed by the antiribophorin antibody raised against a lumenal epitope. (D) After digestion of RMs with trypsin at the indicated concentrations, equal aliquots of the treated vesicles were incubated with the indicated in vitro–synthesized proteins. Background bands generated in the absence of vesicles are indicated with arrows. The percentage of translocation efficiency, corrected for background and normalized to the efficiency obtained in mock-treated RMs, is given below the lanes. Asterisks and boxes indicate the glycosylated and nonglycosylated products, respectively.

Figure 6.

Unassisted insertion of the TA protein PTP1B. (A) PTP1B engineered with the opsin tag (Fig. 1 f) was tested for its ability to insert into RMs co- and posttranslationally or into PC liposomes (1.1 μg phospholipids/μl in all cases). No PF was recovered in the absence of membranes or when detergent was present during PK digestion (right; lanes 4 and 8). The translocation behavior of b5-ops-28 was analyzed in parallel (left). (B) Effect of cholesterol on PTP1B translocation. Protein-free liposomes were prepared from PC–cholesterol mixtures as indicated and were tested for PTP1B-ops-35 insertion (1.3 μg phospholipids/μl) as in A. Asterisks and boxes indicate the glycosylated and nonglycosylated products, respectively.

Figure 7.

Energy requirements for the transmembrane integration of b5 constructs. (A) In vitro–synthesized b5-ops-28 and -85 were gel filtered and diluted in TB to obtain a final ATP concentration of 3 nM (lanes 1 and 7). The diluted samples were incubated with RMs or PC liposomes (two top and two bottom panels, respectively), both at 0.15 μg phospholipids/μl, and were tested for insertion by protease protection. Where indicated, 0.5 mM ATP, 0.1 mM GTP, or both were added (lanes 2–4 and 8–10). Samples not subjected to nucleoside triphosphate depletion were analyzed in parallel. (B) In vitro–synthesized b5-Syb2-ops-28 was subjected to gel filtration and diluted in TB to reach a final ATP concentration of 5 nM during incubation with RMs (0.67 μg phospholipids/μl). ATP, GTP, or both were added where indicated as described for A. Samples not subjected to gel filtration were analyzed in parallel, and background signals were generated after PK digestion in the absence of membranes, as indicated with arrows (lane 5). The percentage of translocation efficiency normalized to the efficiency obtained in mock-treated RMs and corrected for background and for the different input of gel-filtered and untreated samples is given below the lanes. Asterisks and boxes indicate the glycosylated and nonglycosylated products, respectively.