The Get1/2 insertase forms a channel to mediate the insertion of tail-anchored proteins into the ER

Abstract

Tail-anchored (TA) proteins contain a single C-terminal transmembrane domain (TMD) that is captured by the cytosolic Get3 in yeast (TRC40 in humans). Get3 delivers TA proteins to the Get1/2 complex for insertion into the endoplasmic reticulum (ER) membrane. How Get1/2 mediates insertion of TMDs of TA proteins into the membrane is poorly understood. Using bulk fluorescence and microfluidics assays, we show that Get1/2 forms an aqueous channel in reconstituted bilayers. We estimate the channel diameter to be ∼2.5 nm wide, corresponding to the circumference of two Get1/2 complexes. We find that the Get3 binding can seal the Get1/2 channel, which dynamically opens and closes. Our mutation analysis further shows that the Get1/2 channel activity is required to release TA proteins from Get3 for insertion into the membrane. Hence, we propose that the Get1/2 channel functions as an insertase for insertion of TMDs and as a translocase for translocation of C-terminal hydrophilic segments.

Keywords: CP: Cell biology; CP: Molecular biology; channel; membrane protein insertion; protein translocation; tail-anchored proteins.

Figure 1.. The Get1/2 complex makes membranes permeable to solutes

(A) Description of the 6 SUV samples used and the expected outcome of the bulk fluorescence quenching experiment. (B) NBD fluorescence was monitored over 60 min for the indicated samples. 1 mM sodium dithionite is added to SUVs at t = 0. The lipid-to-protein ratio was defined when they were mixed for setting up reconstitution reactions. Data represent an average of three independent experiments (except for the H2000 sample in which there were two independent experiments). Error bars indicate mean ± SD. See also Figures S1.

Figure 2.. Get1/2 channel formation in a free-floating membrane

(A) A schematic of the experimental setup. A free-floating membrane is suspended between two microfluidics channels, as described in the text and STAR Methods. (B) No current is observed in protein-free membranes until their rupture. (C) In Get1/2-decorated membranes, transient currents are typically observed in sequential increases and decreases. These currents attest to the presence of channels in the membrane through which ions can pass upon the action of the voltage applied on the two sides of the membrane. This step increase and decrease of currents shows that the channels are very dynamic and can open and close. The data were collected from four independent experiments. (D) Magnification of traces. The left panel shows a channel that opens and reseals in less than 0.5 s. The right panel shows a channel that is intermittently opened and closed for 100 s (the duration of an acquisition trace is 1,000 s; hence, this channel did not reseal by the end of the trace). (E) Examples of 20-pA currents. Left panel: a channel corresponding to a 20-pA current is opened and closed twice in less than 10 s. Right panel: a 20-pA channel remains open for almost a minute, and a transient 60- to 70-pA channel briefly opens during the process on the current trace. See also Figures S2 and S3.

Figure 3.. Conductance of a single Get1/2 channel

(A) Current step increases and decreases from all four independent experiments are pooled and presented in a histogram with 10-pA bins from 0–350 pA. The positions of the peaks in the distribution are highlighted by dashed circles. Peak 1 is highlighted in red because it corresponds to an intermediate incomplete Get1/2 channel (see text for explanation). Peaks 2–6 correspond, respectively, to current steps in which 1, 2, 3, 4, and 5 Get1/2 channels are opening or closing simultaneously (increase or decrease as observed in Figures 2C-2E). (B) The same distribution presented between 0 pA and 150 pA with 5 pA bins. The histogram is fitted with Origin software by 3 Gaussian peaks that are used to determine the position of peaks 1–3. The positions of peaks 4–6 in A are estimated by averaging the current step values in the corresponding current range. (C) Variation of the conductance with the peak number was calculated from four independent experiments. Error bars indicate mean ± SD. The conductance is the ratio between the peak current and 100-mV voltage applied between the two channels. Peaks 2–6 are perfectly aligned. The corresponding fit is presented in the figure. Peak 1 is not aligned with the others. This shows that each peak from 2–6 corresponds to addition of an identical Get1/2 channel. (D) The mean diameter of the additional channel observed between two consecutive peaks is calculated from the conductance variation data collected from four independent experiments. Error bars indicate mean ± SD. As expected, there is no significant difference between the various increases, as indicated in the text. See also Figures S4.

Figure 4.. Get1/2 forms a heterotetrameric channel with a defined pore size

(A) Schematic showing GFP or SRB encapsulated in Get1/2 SUVs. (B and C) GFP or SRB was encapsulated in protein-free SUVS or Get1/2 SUVs with the indicated lipid-to-protein ratio. The retained fluorescence of GFP or SRB was quantified after removal of exterior free dyes. Data represent an average of two independent experiments. Error bars represent mean ± SD. (D) Diagram illustrating the topology of the Get2-1sc. K150 and K157 in Get2 TMD1 were mutated to alanine residues. The dotted line indicates the linker sequence that connects the C terminus of Get2 to the N terminus of Get1. (E) The indicated SUVs were analyzed by Coomassie staining. (F) NBD fluorescence was monitored for 40 min for the indicated samples. 1 mM sodium dithionite is added to SUVs at t = 0. Data represent an average of two independent experiments. Error bars represent mean ± SD. See also Figures S5 and S6.

Figure 5.. Get3 seals the Get1/2 channel

(A) Example of an NBD quenching assay similar to that presented in Figure 1. 2.6 μM or 26 μM Get3 was incubated with Get1/2 SUVs in the presence or absence of 1 mM ATPγS/MgCl₂. As a negative control, 5 μM BSA was incubated with Get1/2 SUVs. NBD fluorescence quenching in the inner leaflet is reduced as the concentration of Get3 is increased. The dotted line indicates the guideline for Get1/2 + Get3 (26 μM). The experiment was repeated independently three times with less than 5% variation. (B) Mean relative increase of the fluorescence plateau in (A) as the ratio between the plateau of the considered conditions to that of the reference quenching obtained with Get1/2 alone (pink curve). Data represent an average of three independent experiments. Error bars represent mean ± SD. (C) Comparison of the frequency of the Get1/2 channel series in microfluidics experiments similar to that presented in Figure 2 but performed with Get3 or BSA flew in the top channel. This frequency is decreased 10-fold with Get3 but not with BSA. (D) Comparison of the fraction of the time a Get1/2 channel series in microfluidics experiments similar to that presented in Figure 2, but Get3 or BSA was flowed in the top channel. This fraction is decreased ~4-fold with Get3 but not with BSA. The experiments shown in Figures 3C and 3D were repeated three times with similar results.

Figure 6.. Get1/2 channel mediates TA protein insertion

(A) Protein-free, Get2-1sc, or Get2-1sc (K150A/K157A) liposomes were incubated with the purified complex of Get3 and VAMP2 with a 13-amino-acid C-terminal tail. The proteinase K (PK)-protected fragment (PF) represents successful insertion. Proteoliposomes used for the TA protein insertion assay were analyzed by immunoblotting with anti-Get2 antibodies. Bottom: TA protein insertion was quantified by autoradiography. Data represent an average of two independent experiments. Error bars represent mean ± SD. (B) TA protein insertion and quantification were conducted as in (A) but using the purified complex of Get3 and VAMP2 bearing a 26-amino-acid C-terminal hydrophilic tail. Data represent an average of two independent experiments. Error bars represent mean ± SD. (C) Schematic of the TA protein release assay. (D) The Get3-VAMP2 complex was incubated with liposomes or proteoliposomes containing the indicated concentration of proteins and subjected to chemical crosslinking and immunoprecipitation with anti-Get3 antibodies. Aliquots of total crosslinking reactions (input) and immunoprecipitation products of the crosslinking reactions were analyzed by autoradiography. The right panel shows the quantification result of TA protein release from Get3. Data represent an average of five independent experiments. Error bars represent mean ± SD. Statistical analysis of the difference between WT and K150A/K157A for each concentration was performed using an unpaired t test. ns, not significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. #, a background crosslinked band. See also Figure S7.

Figure 7.. Model of TA protein insertion by the Get1/2 channel

Get1/2 forms a heterotetrameric channel that transiently opens and closes. The long and flexible cytosolic domain of Get2 recruits the Get3-TA protein complex closer to the cytosolic domain of Get1. Get1/2 channel opening and closing in the membrane accelerate the movement of the cytosolic Get1 domain to bind and release TA substrate from Get3, allowing the substrate to land directly on the channel for insertion into the ER membrane. K150A/K157A mutations in Get2 close the channel, immobilizing the Get1 coiled-coil domain, inhibiting release of TA substrate from Get3. When the TA protein is inserted into the membrane, Get3 binding to Get1/2 seals the channel until it is released by binding to ATP.