Lipid Scrambling Pathways in the Sec61 Translocon Complex

Abstract

Cellular homeostasis depends on the rapid, ATP-independent translocation of newly synthesized lipids across the endoplasmic reticulum (ER) membrane. Lipid translocation is facilitated by membrane proteins known as scramblases, a few of which have recently been identified in the ER. Our previous structure of the translocon-associated protein (TRAP) bound to the Sec61 translocation channel revealed local membrane thinning, suggesting that the Sec61/TRAP complex might be involved in lipid scrambling. Using complementary fluorescence spectroscopy assays, we detected nonselective scrambling by reconstituted translocon complexes. This activity was unaffected by Sec61 inhibitors that block its lateral gate, suggesting a second lipid scrambling pathway within the complex. Molecular dynamics simulations indicate that the trimeric TRAP subunit forms this alternative route, facilitating lipid translocation via a "credit card" mechanism, using a crevice lined with polar residues to shield lipid head groups from the hydrophobic membrane interior. Kinetic and thermodynamic analyses confirmed that local membrane thinning enhances scrambling efficiency and that both Sec61 and TRAP scramble phosphatidylcholine faster than phosphatidylethanolamine and phosphatidylserine, reflecting the intrinsic lipid flip-flop tendencies of these lipid species. As the Sec61 scrambling site lies in the lateral gate region, it is likely inaccessible during protein translocation, in line with our experiments on Sec61-inhibited samples. Hence, our findings suggest that the metazoan-specific trimeric TRAP bundle is a viable candidate for lipid scrambling activity that is insensitive to the functional state of the translocon.

Figure 1

Translocon reconstitution and fluorescence assays. A) Sucrose gradient purification of detergent-solubilized ribosome/Sec61 translocon complexes. A₂₆₀ UV trace (top), total protein staining (middle), and Western blot analysis (bottom) of the sucrose gradient fractions. HEK293T and sheep whole pancreatic lysates are included as antibody controls. The fraction used in the reconstitution (red) is highlighted together with the composition of the LUVs (blue). The uncropped gels are available in Figure S1 in the SI. B) Chemical structures of the lipids used in experiments. The LUVs are made of POPC with 0.4% (mol) of sn-2 acyl chain-NBD-labeled PC (NBD PC) or PS (NBD PS). C) The sodium dithionite, added to the supernatant, chemically reduces the fluorescent NBD PC or NBD PS to nonfluorescent ABD PC or ABD PS in the outer leaflet. This leads to a decrease of fluorescence intensity by ≈50% of its initial value. In the presence of a scramblase, all NBD-labeled lipids eventually reach the outer leaflet resulting in an almost complete loss of NBD fluorescence. D) The addition of bovine serum albumin (BSA) to the sample leads to partial quenching (≈50%) of NBD fluorescence. Hence, in the absence (presence) of a scramblase, a decrease of ≈25% (≈50%) is expected. E) Results from the dithionite assay. Fluorescence before dithionite addition is normalized to 1, and the curves show a mean of N = 2 repeats. Dashed line shows the expected result in the case of no scrambling (0.50), and further decrease is likely due to the dimming of NBD over time. In the absence of proteins, the NBD PC and NBD PS in the outer leaflet is chemically reduced, resulting in a fluorescence intensity decrease to ≈50% of the original value. In the presence of proteins, the value further decreases to ≈15%, i.e. the majority of the NBD-labeled lipids originally in the inner leaflet are scrambled to the outer leaflet and reduced therein. The presence of Sec61 inhibitors (here either Ipomoeassin F, “IpomF” or Apratoxin A, “AprA”) do not affect scrambling. Overall, the interaction between NBD PS and dithionite seems slower, likely due to the electrostatic repulsion between anionic PS head groups and dithionite. F) Results from BSA assay, confirming the findings of panel E). The dashed lines shows the expected results in the case there is (0.50) or is not (0.75) scrambling. The curves show mean of N = 2 repeats. The presence of Sec61 inhibitors (IpomF or AprA) had no effect on scrambling. Some graphical elements in panels C and D were created with BioRender.com, and they are available online: Šimek, J. (2025)https://BioRender.com/sx9c95m.

Figure 2

Scrambling pathways of the Sec61/TRAP complex. A and B) Snapshots of the Sec61/TRAP complex from A) the side of the lateral gate, “front” and B) from the opposite side, “back”. The prospective scrambling paths of Sec61 along the lateral gate and of TRAP along the groove between the TRAPβ and TRAPγ are highlighted by bidirectional arrows. The structure is embedded in a lipid bilayer shown in gray (phosphorus atoms) and white (rest) to highlight the transmembrane regions. C and D) The polar (green), anionic (red), and cationic (blue) residues located near the potential scrambling pathways are highlighted for C) Sec61 and D) TRAP.

Figure 3

Mechanism of lipid scrambling by Sec61/TRAP. A and B) Volumetric density maps of the lipid headgroup phosphate beads (“PO4”) within A) the bundle of transmembrane helices of the TRAPβ, TRAPγ, and TRAPδ subunits or B) Sec61. Averaged from five 20 μs-long simulations of the Sec61/TRAP complex in the multicomponent membrane. TRAP and Sec61 subunits are colored as in Figure 2. The densities of the different lipid types are provided in Figures S2 (Sec61) and S3 (TRAP) in the SI. For visualization, atomistic protein structures are employed in the rendering. C and D) Snapshots of scrambling mechanism by C) the bundle of transmembrane helices of the TRAPβ, TRAPγ, and TRAPδ subunits or D) Sec61. The lipid headgroup (black bead) partitions to the polar crevices highlighted in Figure 2C and 2D, whereas the acyl chains (gray) remain in the membrane environment. Coloring of subunits as in Figure 2 and panels A and B. For visualization, atomistic protein structures are employed in the rendering. E) Volumetric density maps of water from atomistic simulations of the Sec61/TRAP complex. Water density is shown in blue, protein in transparent surface with TRAPβ, TRAPγ, and TRAPδ on the left and Sec61 with TRAPα on the right. Lipid headgroup phosphorus atoms are shown as brown spheres to highlight membrane thickness and curvature perturbations. F) Effect of membrane thickness on the free energy barrier for lipid flip–flop. Black markers show thickness/barrier pairs calculated from a set of bilayers comprised of lipids with saturated chains of varying length, whereas the gray dashed line is provided as a guide to the eye. The colored lines show the thickness values observed in the vicinity of Sec61, in the vicinity of the bundle of TRAPβ, TRAPγ, and TRAPδ subunits, and far away from the proteins in the “bulk” membrane. The thinning induced by the proteins lowers the free energy barrier by an estimated ≈4.1 kJ/mol (TRAP) or ≈6.8 kJ/mol (Sec61). These values are extracted as the differences of the intersections of the colored lines and the gray dashed fit along the y axis.

Figure 4

Energetics of lipid scrambling by Sec61/TRAP. A) The free energy profiles for a POPC flip–flop in the single-component membrane in the presence of Sec61, TRAP, or the Sec61/TRAP complex as well as in the protein-free membrane at 310 K. The profiles are calculated from density profiles and using biased AWH simulations, respectively (see Methods). Error bars are calculated from the difference of the profiles in the two membrane leaflets (protein-free system) or as the standard deviation of the five replica simulations (protein-containing systems). B) Temperature dependence of the free energy profile for lipid scrambling by Sec61/TRAP in the single-component membrane (Set 4 in Table 1). The profiles are calculated from density profiles (see Methods). The error bars show the standard deviation of the five replica simulations. C) Decomposition of the free energy profile into entropic and enthalpic components performed by assuming a constant enthalpy in the simulated temperature range (see Methods). The error bars show the difference between two estimates calculated from two pairs of temperatures (300 K and 320 K as well as 290 K and 330 K, Set 4 in Table 1). The location of the barrier at 310 K is marked with a dashed gray line, and the corresponding entropic and enthalpic values are also indicated. D) Arrhenius analysis of lipid scrambling rate by Sec61/TRAP in the POPC membranes (Set 4 in Table 1). The scrambling rates observed at different temperatures are fitted with , from which the activation energy E_A is obtained. The error bars show relative error, yet they are not visible for most of the data points. E) Lipid headgroup selectivity of the scrambling activity of Sec61, TRAP, and Sec61/TRAP complexes. Bars show the mean and standard error for the number of total scrambled lipids extracted from the five 20 μs-long replica simulations of the multicomponent membrane (Set 2 in Table 1). The membranes contain equal amounts of PC, PE, and PS. F) Scrambling of POPC by Sec61, TRAP, and Sec61/TRAP complexes in the single-component membrane (Set 1 in Table 1). Bars show the mean and standard error for the number of total scrambled lipids extracted from the five 20 μs-long replica simulations.