Insertion of short transmembrane helices by the Sec61 translocon

Abstract

The insertion efficiency of transmembrane (TM) helices by the Sec61 translocon depends on helix amino acid composition, the positions of the amino acids within the helix, and helix length. We have used an in vitro expression system to examine systematically the insertion efficiency of short polyleucine segments (L(n), n = 4 ... 12) flanked at either end by 4-residue sequences of the form XXPX-L(n)-XPXX with X = G, N, D, or K. Except for X = K, insertion efficiency (p) is <10% for n < 8, but rises steeply to 100% for n = 12. For X = K, p is already close to 100% for n = 10. A similar pattern is observed for synthetic peptides incorporated into oriented phospholipid bilayer arrays, consistent with the idea that recognition of TM segments by the translocon critically involves physical partitioning of nascent peptide chains into the lipid bilayer. Molecular dynamics simulations suggest that insertion efficiency is determined primarily by the energetic cost of distorting the bilayer in the vicinity of the TM helix. Very short lysine-flanked leucine segments can reduce the energetic cost by extensive hydrogen bonding with water and lipid phosphate groups (snorkeling) and by partial unfolding.

Fig. 1.

Integration of polyleucine H-segments into microsomal membranes. (A) Leader peptidase (Lep) model protein construct. Wild-type Lep, consisting of 2 TM helices (TM1, TM2) and a large luminal domain (P2), inserts into rough microsomes in an N_lum-C_lum orientation. H-segments (red) of the form XXPX-L_n-XPXX with n = 4 to n = 12 and X = G, N, D, or K were engineered into the P2 domain between residues 226 and 253. Glycosylation acceptor sites (G1 and G2) were placed in positions 96–98 and 258–260, flanking the H-segment. For H-segments that integrate into the membrane, only the G1 site is glycosylated (Left), whereas both the G1 and G2 sites are glycosylated for H-segments that do not integrate into the membrane (Right). (B) Membrane integration of H-segments for GGPG-L_n-GPGG constructs with n = 8, 10, and 12. Plasmids encoding the Lep/H-segment constructs were transcribed and translated in vitro in the presence (+) and absence (−) of dog pancreas rough microsomes (DRM). Translation products were analyzed by SDS/PAGE. Bands of unglycosylated protein are indicated by white dots; singly (1g) and doubly (2g) glycosylated proteins are indicated by 1 and 2 red dots, respectively. The data were quantitated by scanning the gels in a phosphoimager. The probability (efficiency) of membrane insertion is given by P = 1g/(1g + 2g). Mean values from 2 independent experiments were used for computing P values. On average, glycosylation levels vary by about ± 2% between repeat experiments.

Fig. 2.

Length dependence of various parameters. (A) Efficiency of insertion of XXPX-L_n-XPXX constructs in dog pancreas rough microsomes. Data for X = G (black squares) were collected for n = 5, 6, 8, 10, 12, and 19. The solid curve through the points is a best-fit Boltzmann distribution (see Results). Diameters of points correspond to the experimental uncertainty. (B) Wavelength-minima of the OCD spectra as a function of the number of leucines for X = G and X = K. Approximate representative positions of the minima are shown in Fig. 3 B and D. Although the interpretation of the OCD spectra for n ≤ 8 (Fig. 3) is uncertain, this plot reveals a structural change that approximates the shape of the probability-of-insertion curves shown here. All OCD spectra for n ≥ 10 are consistent with TM α-helices, as shown by the examples in Fig. 3. (C) Relative free-energy costs of deforming a POPC bilayer to accommodate XXPX-L_n-XPXX peptides of different lengths and compositions. Relative free energy is the computed free energy for a given n divided by the free energy computed for GL20. By using the fits of Eq. S1 to the d(r) curves of Fig. 5, we calculated the differences in deformation free energies of the lipid-protein system by using Eq. S2. We calculated the difference in free energy of all systems relative to data from a simulation of a GL20 peptide in a bilayer, ΔG_peptide/ΔG_GL20. GL20 was chosen as a reference because the hydrophobic mismatch is small. For this reason, the relative free energies for n = 20 are zero in this figure.

Fig. 3.

Oriented circular dichroism (OCD) spectra of synthetic XXPX-L_n-XPXX peptides in oriented multilamellar POPC bilayer arrays for X = G and K. Oriented multilamellar bilayer arrays were formed on 2.5-cm-diameter quartz plates, as described in detail elsewhere (9, 10, 27). Solid curves (labeled in A) represent theoretical spectra for transmembrane helices oriented normal and parallel to the membrane plane, as indicated. Red arrows in B and D indicate the positions of the minima in OCD curves. (A) Spectra for GL12 at various peptide/lipid mole ratios. The spectra indicate that the peptides are predominantly α-helical and normal to the bilayer plane. (B) Spectra for KL12 at various peptide/lipid mole ratios. As for GL12, the spectra are consistent with TM α-helices. (C and D) Spectra for GL6 and KL6, respectively. GL12: X = G, n = 12; GL6: X = G, n = 6; KL12: X = K, n = 12; KL6: X = K, n = 6.

Fig. 4.

Summary of all-atom molecular dynamics simulations of GL6, GL12, GK6, and GK12 peptides in POPC bilayers. (Left) Cut-away snapshots of the peptides spanning the lipid bilayer. The representative snapshots were taken during the simulation production runs (i.e., during the stable period after the simulation cell dimensions had reached steady values). Each POPC bilayer consisted of 280 POPC molecules (140 in each monolayer) and 8,400 water molecules. At the beginning of each simulation, the peptides were positioned at the bilayer center of mass with the leucines in an α-helical conformation and the XXPX…XPXX flanks in an extended conformation. Color code: gold, lipid phosphates; dark blue, waters associated with phosphates; white, leucine residues; green, glycine residues (A and B) or lysine residues (C and D); red, proline residues; orange, waters within 7 Å of the peptides. (Right) The production-run time-averaged transbilayer distributions of the waters and peptide amino acids for each simulation. The color code is the same as in Left. The atom densities are for heavy atoms, i.e., H atoms are not included. For reference, the oxygen atom number-density of bulk water is 33 × 10⁻³ atoms per ų. (A) Simulation of GL6 in a POPC bilayer. (B) Simulation of GL12 in a POPC bilayer. (C) Simulation of KL6 in a POPC bilayer. (D) Simulation of KL12 in a POPC bilayer.

Fig. 5.

Radial time-averaged profiles of the bilayers in the simulations. The average thickness z(r) of the bilayer is shown as a function of the radial distance r from the average peptide axis. (Left) The time-averaged positions of the lipid carbonyl and peptide leucine groups. We ignored data for r < 5 Å because the average position of leucine sidechains is ≈5 Å. Taking periodic boundary conditions into consideration, we selected carbonyl groups within 50 Å of the center-of-mass of the helix, represented here by the orange dots. We then calculated the average position of 2-Å-thick rings of carbonyl groups centered on the helix's center of mass, indicated here by the black dots. Finally, we calculated the average position of leucine sidechains every 1 Å along the z-axis, represented her by the red circles. (Right) The computed average carbonyl-to-carbonyl separations across the bilayer at 2-Å intervals away from the helices (black points with error bars indicating the standard deviations of the means). The solid orange line represents the best-fit of the data to Eq. S1. A–D summarize the results of the analyses for GL6, GL12, KL6, and KL12 simulations, respectively.