Subunit cooperation in the Get1/2 receptor promotes tail-anchored membrane protein insertion

Integral membrane proteins comprise >30% of the proteins encoded by the genome and are essential for the proper functioning of biological membranes. The high aggregation propensity of the transmembrane domains (TMDs) on membrane proteins requires sophisticated protein machineries for their efficient targeting and translocation. An important class of membrane proteins is tail-anchored proteins (TAs), defined by a single TMD near the C terminus. TAs comprise up to 5% of the eukaryotic membrane proteome and mediate key cellular functions including vesicle fusion, apoptosis, protein translocation, and quality control (Chartron et al., 2012; Hegde and Keenan, 2011). As their single C-terminal TMDs are buried inside the ribosome exit tunnel until translation termination, TAs cannot engage cotranslational protein targeting machineries, such as the signal recognition particle (SRP), and must instead use a post-translational mechanism of targeting and insertion (Kutay et al., 1993, 1995).

The best studied pathway for TA biogenesis is the guided entry of TA (GET) pathway in yeast and the homologous TMD recognition complex (TRC) pathway in mammals, which mediates the targeted delivery of an essential class of TAs with highly hydrophobic TMDs to the ER membrane (Chio et al., 2017a). Newly synthesized TAs are relayed, via a cascade of cytosolic chaperones and cochaperones, to the targeting factor Get3 (Cho and Shan, 2018; Wang et al., 2010; Rao et al., 2016; Chio et al., 2019; Mariappan et al., 2010). Get3•TA is captured by the cytosolic domains (CDs) of the Get1/2 receptor complex at the ER membrane, which further induces the release of TA from Get3 and mediates TA insertion into the ER (Mariappan et al., 2011; Stefer et al., 2011; Wang et al., 2011). The GET pathway is conserved across eukaryotic organisms, and deletion of Get3 homologues causes sensitivity to stress conditions in yeast and embryonic lethality in mice (Mukhopadhyay et al., 2006; Shen et al., 2003; Schuldiner et al., 2008).

Structural, biochemical, and biophysical studies showed that nucleotide, TA substrate, and interaction partners extensively regulate the conformation of Get3 during the targeting cycle. Get3 is an obligate homodimer, and nucleotide binding/release at its ATPase domain induce adjustments at the dimer interface that are amplified into larger movements of the helical domains (Mateja et al., 2009; Suloway et al., 2009). In the cytosol, ATP and Get4/5 induce Get3 into a “closed” conformation in which the helical domains of both Get3 subunits are brought into proximity and form a contiguous hydrophobic groove, which provides a binding site for the TA-TMD (Gristick et al., 2014; Mateja et al., 2015). The TA substrate triggers ATP hydrolysis in Get3 and induces it to sample open conformations, which drives the detachment of Get3•TA from Get4/5 and exposes the Get1/2 binding sites, allowing Get3•TA to interact with the membrane receptor complex (Rome et al., 2013; Chio et al., 2017b).

The membrane-associated events remain one of the least understood aspects of the GET pathway. Both Get1 and Get2 contain three TMDs, which mediate their heterodimerization and together form an “insertase” homologous to YidC that is necessary and sufficient for the insertion of TAs into the ER membrane (Wang et al., 2011; Mariappan et al., 2011; McDowell et al., 2020; Zalisko et al., 2017; Stefer et al., 2011). Both subunits also contain CDs that interact with Get3. Get2CD (residues 1–150) contains two conserved helices rich in basic residues (residues 1–40), which cocrystallized with closed, nucleotide-bound Get3 (Mariappan et al., 2011; Stefer et al., 2011; Kubota et al., 2012). In contrast, Get1CD (located between TM1 and TM2) forms a coiled-coil (cc) that inserts like a wedge at the dimer interface of apo-Get3 and induces Get3 into a wide-open conformation, leading to nucleotide release and disruption of the TA binding site. Get1CD is also sufficient to induce TA release from Get3 at high concentrations (Wang et al., 2011), while a large fragment of Get2CD (residues 1–106) did not induce TA release (Mariappan et al., 2011). These observations led to the current model in which the long, flexible Get2CD mediates the capture of Get3•TA using its N-terminal helix, whereas Get1CD is responsible for remodeling the conformation of Get3 to induce TA release.

This modular model is largely based on the results of studies using the individual Get1CD or Get2CD. However, the measured binding affinities of Get1CD and Get2CD for the Get3•TA complex are low, with K_ds in the micromolar range (Rome et al., 2014), whereas the in vivo concentration of Get1/2 is 0.17–0.2 µM (Kulak et al., 2014). This poses questions as to how the targeting complex is efficiently captured at the ER membrane. Using a mini-Get1/2 complex in which Get1CD and Get2CD are fused to oppositely charged α-helices that form a stable cc heterodimer (Wang et al., 2011), pull-down and TA release experiments suggested that the presence of both subunits can enhance the binding of the receptor complex to Get3 (Wang et al., 2011). However, the molecular basis for the potential cooperation between the CDs of the two subunits in the Get1/2 complex remains elusive.

In this work, we studied subunit cooperation in Get1/2 during the capture and remodeling of the Get3•TA targeting complex using mini-Get1/2, which has been shown to recapitulate the cytosolic functions of the Get1/2 complex (Wang et al., 2011). Quantitative binding assays and single-molecule (sm) fluorescence spectroscopy studies revealed extensive cooperation between Get1CD and Get2CD during the binding and conformational remodeling of Get3•TA. Unexpectedly, Get2CD also induces Get3 opening. This activity is mediated by two molecular recognition features (MoRFs) in the disordered linker of Get2CD, and their mutations attenuated TA insertion in vivo. Our results identify a mechanistically important step for the membrane-associated events in the GET pathway.

Mini-Get1/2 was previously used to mimic the function of the CDs of the Get1/2 receptor complex (Fig. S1, A and B; Wang et al., 2011). Mini-Get1/2 coimmunoprecipitated with Get3 more strongly than variants bearing either the N72A/R73E mutation (NR) in Get1CD or the R14E/E15R/R16E/R17E mutation (RERR) in Get2CD that disrupt binding to Get3, suggesting that both subunits contribute to Get3 binding (Wang et al., 2011). To decipher potential subunit cooperation during capture of the Get3•TA targeting complex, we quantitatively evaluated the contribution of the individual CDs to the binding of Get3•TA. We assembled Get3•TA complexes by in vitro translation of Bos1, a model GET substrate, in Escherichia coli lysate in the presence of Get3 and affinity-purified Get3•TA complexes via the 3xStrep-tag on Bos1 (Rao et al., 2016). Get3•TA complexes generated by this procedure were kinetically stable and highly efficient in TA targeting and insertion into the ER membrane (Chio et al., 2017b; Rao et al., 2016). To quantitatively characterize the interaction of Get3 with Get1/2, we used Förster resonance energy transfer (FRET) between BODIPY-FL (BDP; donor dye) labeled on Get3 (Rao et al., 2016) and tetramethylrhodamine (TMR; acceptor dye) labeled on Get1CD or Get2CD in mini-Get1/2 (Fig. 1 A). Addition of Get3^BDP to mini-Get1^TMR/2 or mini-Get1/2^TMR led to a ∼50% reduction in donor fluorescence and a corresponding increase in acceptor fluorescence (Fig. 1, B and C; green versus black lines). This fluorescence change was competed away by excess unlabeled mini-Get1/2 (Fig. 1, B and C, black versus blue lines), confirming that the observed fluorescence changes arose from FRET between the dye pair upon interaction of mini-Get1/2 with Get3•TA.

Equilibrium titrations using this FRET assay showed that mini-Get1/2 binds Get3•TA tightly, with an equilibrium dissociation constant (K_d) of 0.5–1 nM (Fig. 1, D and F). Similar K_d values were observed regardless of whether Get1CD or Get2CD was labeled, indicating that the tight binding was unlikely to arise from artifacts of fluorescence labeling. To evaluate the contribution of the individual subunits to Get3•TA binding, we introduced the NR mutation in Get1CD or the RERR mutation in Get2CD. Each mutation weakened the binding of mini-Get1/2 to Get3•TA 10–30-fold (Fig. 1, E and F). Introduction of both mutations abolished the binding of mini-Get1/2 to Get3•TA (Fig. 1, E and F), ruling out the possibility that the individual mutations did not completely disrupt the Get3 interaction of the respective CDs, or that artificial stabilizing interactions were introduced by fluorescence labeling.

While the observed mutational effects are consistent with the model that Get1CD and Get2CD co-bind Get3 in the Get1/2 complex (Stefer et al., 2011; Zalisko et al., 2017; Wang et al., 2011), mutants mini-Get1(NR)/2^TMR and mini-Get1^TMR/2(RERR) retained high affinities for Get3•TA, with K_d values of 7.1 nM and 14.8 nM, respectively (Fig. 1 F). These affinities are ∼220- and ∼170-fold higher than those measured previously with isolated Get2CD and Get1CD, respectively (Fig. 1 F; Rome et al., 2014). These results suggest that, in addition to avidity effects, the assembly of Get1CD with Get2CD in mini-GET1/2 induced conformational changes in the CDs that greatly enhanced their individual affinities for the Get3•TA complex.

We next determined the contributions of the individual Get1/2 subunits to disassembly of the Get3•TA complex. To this end, we used a previously developed FRET assay based on a donor dye, coumarin (Cm), labeled on the TA-TMD, and an acceptor dye, BDP, labeled on Get3 (Rao et al., 2016). TA dissociation from Get3 was induced by chasing a preassembled Get3^BDP•TA^Cm complex (15 nM) with an excess amount (20 µM) of a plant-derived membrane protein chaperone, which traps any dissociated TAs (Fig. 2 A, TA trap). The dissociation kinetics was determined by real-time measurement of the recovery of donor (Cm) fluorescence due to FRET loss between dyes. In the absence of mini-Get1/2, the Get3•TA complex dissociated slowly, with a t_1/2 of ∼160 min (TA-trap only in Fig. 2 C and Chio et al., 2019). The additional presence of mini-Get1/2 accelerated TA release from Get3 fivefold to sixfold (Fig. 2, B and C; t_1/2 ∼30 min). The Get1(NR) or Get2(RERR) mutation each reduced the rate of TA release from Get3 (Fig. 2, B and C; t_1/2 ∼60 min), and introduction of both mutations resulted in a TA release rate similar to spontaneous Get3•TA dissociation (Fig. 2, B and C; t_1/2 ∼180 min). These experiments were performed at mini-Get1/2 concentrations (3 µM) far above their respective K_ds for Get3•TA (Fig. 1 F); thus, the observed defects of mini-Get1(NR)/2 and mini-Get1/2(RERR) in TA release could not be attributed to incomplete Get3•TA binding. These results suggest that both the Get1 and Get2 CDs play direct roles in the disassembly of Get3•TA complexes, and argue against models in which Get1CD is solely responsible for triggering TA release from Get3 (Mariappan et al., 2011; Stefer et al., 2011; Wang et al., 2011).

The observation that mini-Get1(NR)/2 significantly accelerated Get3•TA disassembly (Fig. 2) strongly suggests that Get2CD can also remodel Get3•TA to facilitate TA release. To directly test this hypothesis, we used an established solution-based smFRET assay that monitors the global conformation of individual Get3 dimers at equilibrium (Chio et al., 2017b). In this assay, Get3 dimers are doubly labeled with donor and acceptor fluorophores such that closed Get3 displays high FRET, whereas open Get3 displays low FRET (Fig. 3 A; and Fig. S2, A and B). Both the stoichiometry (S) and effective FRET efficiency (E*) of individual Get3 dimers can be detected and quantified as they diffuse through a femtoliter-scale observation volume, using confocal microscopy with alternating laser excitation with microsecond time resolution (µs-ALEX; Kapanidis et al., 2005). To better resolve the open conformations, we used a FRET pair with a longer Förster radius (ATTO 550/ATTO 647N; R₀ ∼65 Å) than the Cy3B/ATTO 647N pair used in the previous study (Chio et al., 2017b). Analysis on Get3 labeled with only the donor or acceptor dye showed that Get1CD and Get2CD do not alter the dye photophysics in a manner that will alter the observed FRET efficiency (Fig. S2 C). Therefore, the observed changes in FRET efficiency can be attributed to global conformational changes in Get3.

Apo-Get3 displayed a broad smFRET histogram that peaked at E* values of ∼0.7–0.9 (Fig. 3 B, black outline), as reported previously (Chio et al., 2017b). Addition of Get2CD shifted the FRET histogram of Get3 toward lower FRET efficiencies, with a new peak at E* ∼0.4 (Fig. 3 B, gray bars versus black outline), indicating that Get2CD induced Get3 opening. As predicted from previous structural and biochemical data (Kubota et al., 2012; Mariappan et al., 2011; Stefer et al., 2011), Get1CD induced extensive opening of Get3, such that the FRET histogram is dominated by a peak at E* ∼0.3 (Fig. 3 B, magenta outline). In addition to apo-Get3, Get2CD also induced Get3 opening in the presence of ADP (Fig. 3 C), but not in the presence of ATP (Fig. 3 D). The same results were observed using Get3 labeled with the Cy3B/ATTO 647N FRET pair, except that the FRET histograms are shifted toward lower E* values due to the shorter Förster radius of this dye pair (Fig. S2, D–G). Together, these results provide direct evidence that, in addition to Get1CD, Get2CD can also remodel Get3 and bias its conformation to more open states.

To better mimic the in vivo function of Get1/2, we tested how mini-Get1/2 regulates the conformation of the Get3•TA complex. The smFRET histogram of Get3•TA displays a major peak at E* ∼0.9, with a highly populated tail extending toward lower FRET values (Fig. 4, A and E). This confirms the results of previous work showing that the TA substrate induces Get3 to sample open conformations (Chio et al., 2017b). Addition of mini-Get1/2 shifted the smFRET histogram of Get3•TA to lower FRET values, with a major peak at E* ∼0.35 and a minor peak at E* ∼0.9 (Fig. 4, B and E). In comparison, the low FRET population was more dominant with mini-Get1/2(RERR) bound to Get3•TA (Fig. 4, C and E), consistent with previous crystallographic results showing that Get1CD induces a highly open Get3 (Kubota et al., 2012; Mariappan et al., 2011; Stefer et al., 2011). Mini-Get1(NR)/2 also shifted the smFRET histogram, reducing the peak at E* ∼0.9 and inducing a new peak at E* ∼0.35 (Fig. 4, D and E), although the low FRET state was less populated compared to that with WT mini-Get1/2 (Fig. 4, A and E). Thus, Get2CD also induces opening of Get3•TA, but less extensively compared with Get1CD.

To further determine whether the conformational heterogeneity of Get3•TA is dynamic or static in nature, we implemented burst variance analysis (BVA), which detects dynamic conformational sampling by comparing the SD of E* over time with the SD expected from shot noise (Fig. S2, H–K). If multiple conformations interconvert on the sub-millisecond or faster timescale, the SD of sub-bursts in each burst would be higher than the expected shot noise–limited SD (static limit; Fig. S2, H–K, red lines; Torella et al., 2011). Get3•TA exhibiting intermediate E* values (0.5–0.7) displayed higher SD than the static limit (Fig. S2 H), consistent with our previous report (Chio et al., 2017b). This indicates that TA-bound Get3 dynamically samples open conformations and that the observed intermediate E* values are due to averaging of Get3•TA complexes in lower and higher FRET states that interconvert on the sub-millisecond or faster timescale. Get3•TA remains highly dynamic when bound to all the mini-Get1/2 constructs, transitioning between low and high FRET states on the sub-millisecond timescale (Fig. S2, I–K).

Notably, mini-Get1/2(RERR) is less efficient than mini-Get1/2 in facilitating TA release from Get3 (Fig. 2, B and C) despite its ability to induce a more open Get3 (Fig. 4, C and E). This provides additional evidence that the defect of mini-Get1/2(RERR) in Get3•TA disassembly cannot be attributed to weaker binding of mini-Get1/2(RERR), and suggests that increased sampling of the open conformation does not necessarily lead to more efficient Get3•TA disassembly. Instead, our results suggest that Get1CD and Get2CD together induce an intermediary degree of Get3 opening, in which both the open and closed states are sampled, and that this conformational balance is conducive to trigger the most efficient TA release.

The ability of Get2CD to induce Get3 opening is unexpected, as the conserved N-terminal helix of Get2CD (residues 1–40) cocrystallized with closed Get3 and binds to one of the subunits in the Get3 dimer, in contrast to Get1CD, which inserts at the dimer interface of open Get3 (Stefer et al., 2011; Mariappan et al., 2011). Get2CD also contains an ∼110–amino acid disordered linker that connects its N-terminal helix to the TMDs. To test whether the N-terminal helix or the linker of Get2CD is responsible for Get3 opening, we purified Get2CD(1–40; Fig. S1 C) and tested its effect on Get3 conformation. Get2CD(1–40) did not significantly change the smFRET histogram of Get3 in both the apo and ADP states (Fig. 5 A), indicating that the disordered linker in Get2CD is required for remodeling Get3.

To search for sequences in the Get2CD linker required to induce Get3 opening, we generated a set of deletion mutants in which 10–20—amino acid segments in the linker sequence are replaced with (GS)₆ (GS1–GS6; Fig. S1 C and Fig. S3 A). smFRET measurements showed that mutants GS2, GS3, and GS4 displayed substantial defects in the remodeling of Get3, whereas GS1, GS5, and GS6 induced Get3 opening similarly to or slightly more efficiently than WT Get2CD (Figs. 5 G and Fig. S3). These results suggest that residues 60–120 in the Get2CD linker contain potential interaction motifs that mediate the conformational remodeling of Get3.

To identify interaction motifs in the Get2CD linker, we analyzed its sequence using multiple predictors for MoRFs, 10–70-residue segments in intrinsically disordered regions (IDRs) that mediate molecular interactions and undergo disorder-to-order transitions upon partner binding (Mohan et al., 2006; Mészáros et al., 2009; Disfani et al., 2012; Fang et al., 2013; Cumberworth et al., 2013; Fung et al., 2018; Oldfield et al., 2008). Three MoRF predictors, ANCHOR, MoRFpred, and MoRF_CHiBi__Web (Fang et al., 2013; Disfani et al., 2012; Malhis et al., 2016), converged on two predicted MoRFs in the Get2CD linker at residues 76–88 and 103–113 (termed H1 and H2, respectively; Fig. 5, B and C). The location of these MoRFs is consistent with observations with the linker deletion mutants (Fig. 5 G and Fig. S3). In addition, two protein secondary structure prediction programs, PSIPRED (Jones, 1999) and RaptorX (Källberg et al., 2012), predicted high helical propensity in H1 and H2 (Fig. S4, A and B). Finally, H1 and H2 contain multiple residues conserved in their hydrophobic nature despite the overall poor sequence conservation of the Get2CD linker (Notredame et al., 2000; Fig. 5 B and Fig. S4 C).

To test the role of H1 and H2, we replaced the hydrophobic residues in each or both MoRFs with glycine, generating mutants Get2CDΔH1, Get2CDΔH2, and Get2CDΔH1ΔH2 (Fig. 5 B). Get2CDΔH1 and Get2CDΔH2 displayed substantially reduced ability to induce the opening of Get3 in both the apo and ADP states (Fig. 5, D, E, and G), and mutation of both H1 and H2 abolished Get3 opening (Fig. 5, F and G). To test whether these remodeling defects arise from weakened binding of the Get2CD MoRF mutants with Get3, we introduced the ΔH1ΔH2 mutation in mini-Get1/2. Using the FRET assay between Get3^BDP and mini-Get1/2^TMR (Fig. 1 A; and Fig. S2, A and B), equilibrium titrations showed that mini-Get1/2 and mini-Get1(NR)/2 bound Get3 with the same affinity, within error, regardless of the presence of the MoRFs (Fig. 6, A–C and E). This excludes the possibility that the remodeling defects of mutant Get2ΔH1ΔH2 are due to weakened binding. When the RERR motif in Get2 was mutated, though, mutation of H1 and H2 weakened the affinity of mini-Get1^TMR/2(RERR) twofold (Fig. 6, D and E), suggesting that the Get2 MoRFs could facilitate the interaction of the Get1 subunit with Get3. Together, these data indicate that high-affinity binding between Get2CD and Get3 is primarily mediated by the N-terminal helices in Get2, whereas the remodeling of Get3 requires both MoRFs in the disordered Get2CD linker.

A helical wheel analysis further suggested that both MoRFs form amphiphilic helices that present a contiguous hydrophobic surface (Fig. 7 A), through which direct interactions between the disordered Get2CD linker and Get3 could be formed. To test whether H1 and H2 directly interact with Get3 to induce conformational changes, we used amber suppression to incorporate a zero-length photo–cross-linker, Bpa, in the H1 (Ala85) or H2 (Ser111) motifs (Fig. 7 A and Fig. S1 C, H1^Bpa and H2^Bpa). We observed efficient cross-links of Get3 with both Get2CD_H1^Bpa and Get2CD_H2^Bpa that are dependent on UV and the Bpa cross-linker (Fig. 7 B). In addition, mutation of conserved hydrophobic residues in H1 and H2 abolished the cross-link (Fig. 7 C). These results provide strong evidence for direct interactions of H1 and H2 directly contact Get3.

Collectively, the results in this section show that the disordered linker of Get2CD contains two conserved MoRFs, which directly interact with Get3 and are responsible for mediating the conformational remodeling of Get3.

Finally, we asked whether the ability of Get2CD to mediate Get3 opening is important for TA insertion into the ER in vivo. To this end, we used an established model GET substrate, in which BirA is fused to the C-terminal targeting sequence of the SNARE protein Bos1 (Fig. 8 A, BirA-Bos1) and constitutively expressed from a low copy plasmid in yeast cells (Cho and Shan, 2018). An engineered C-terminal opsin tag allows efficient glycosylation of BirA-Bos1 upon insertion into the ER lumen, providing a quantitative readout for insertion efficiency. To test the role of the Get2 H1 and H2 motifs in TA insertion in vivo, we replaced genomic GET2 in Saccharomyces cerevisiae with GET2-FLAG, get2ΔH1-FLAG, get2ΔH2-FLAG, or get2ΔH1ΔH2-FLAG.

The steady-state insertion efficiency of BirA-Bos1 in yeast cells was unaffected by mutations in the H1 or H2 motifs of Get2 (Fig. S5 A). When the insertion kinetics of newly synthesized TAs was measured using a pulse-chase assay, a modest but reproducible delay in BirA-Bos1 insertion was detected at early time points (1 min) with Get2 mutants bearing the ΔH1 and/or ΔH2 mutation (Fig. 8, B–D; and Fig. S5 B). As controls, we also introduced the NR and RERR mutations into genomic GET1 and GET2, respectively, and verified that both mutations induced a 10–15% reduction in the steady-state TA insertion level and impaired the insertion kinetics of BirA-Bos1 (Fig. 8 B and Fig. S5 B). The delay in TA insertion cannot be explained by lower expression levels of the Get2 mutants compared with WT Get2 (Fig. S5 C). Thus, the H1 and H2 motifs on Get2CD, which specifically mediate Get3 opening, facilitate GET-dependent TA insertion in vivo.

To ensure efficient membrane protein biogenesis, receptor and/or translocase complexes at the target membrane must effectively capture the targeting complex in the cytosol and remodel their conformation to induce the release of nascent membrane proteins at the site of their insertion. In this work, biochemical, biophysical, and sm fluorescence experiments shed light on how the two subunits in the Get1/2 translocase complex coordinate to remodel the Get3 targeting factor for efficient substrate release during the post-translational targeting of tail-anchored membrane proteins to the ER. In contrast to previous models, in which the role of Get2 is restricted to the initial recruitment of Get3•TA and Get1 is solely responsible for Get3•TA disassembly (Mariappan et al., 2011; Stefer et al., 2011; Wang et al., 2011), our results uncover a new role of Get2 in regulating the conformation of Get3 and reveal extensive cooperation between the Get1 and Get2 subunits during the capture and remodeling of Get3•TA. These results identify a previously unappreciated and mechanistically important step in the membrane-associated events in the GET pathway and, together with other works, emphasize the role of IDRs in the function of receptor and translocase complexes during protein biogenesis.

Previous measurements using isolated CDs reported low binding affinities of Get1CD and Get2CD for the Get3•TA complex, with K_ds in the micromolar range (Rome et al., 2014). Coupled with the low in vivo concentration of Get1/2, this posed questions as to how the targeting complex is efficiently captured at the ER membrane. The measurements here show that, when assembled into the mini-Get1/2 complex, the individual Get1CD and Get2CD subunits display low nanomolar affinities for Get3•TA, two orders of magnitude higher than the affinities observed with the isolated CDs (Fig. 1 F; Rome et al., 2014). The precise mechanism underlying this affinity enhancement remains to be determined, but a few models can be excluded based on available data. The enhanced binding is unlikely to be due to a Get1- or Get2-induced conformational change of Get3, as high-affinity binding was still observed even when the Get3 binding site in the other subunit was mutated (Fig. 1 F, compare the K_d of mini-Get1[NR]/2^TMR versus Get2CD, or the K_d of mini-Get1^TMR/2[RERR] versus Get1CD). The enhanced binding is also unlikely to arise from oligomerization of mini-Get1/2. Size-exclusion chromatography coupled with multi-angle light scattering (SEC-MALS) measurements showed that mini-Get1/2 is a heterodimer (Fig. S1 B), as expected. Although a recent work strongly suggests that Get3 induces Get1/2 tetramerization (McDowell et al., 2020), tetramer formation is mediated by the Get1/2 TMDs and is unlikely to occur with mini-Get1/2. The absence of cooperativity during the binding of mini-Get1/2 to Get3•TA (Fig. 1) also provided no indication of Get3-induced oligomerization of mini-Get1/2. While Get1CD and Get2CD do not appear to display significant interactions with one another (Zalisko et al., 2017), it is possible that transient interactions occur between the Get1 and Get2 CDs when they are positioned in close proximity in a complex, which primes their conformations for more favorable Get3•TA interaction.

Although previous biochemical and crystallographic studies found no evidence for a role of Get2CD in regulating the conformation of Get3 (Mariappan et al., 2011; Stefer et al., 2011), the smFRET measurements directly visualized Get3 opening by Get2CD (Fig. 3). Consistent with previous observations in crystallographic analyses, the highly conserved N-terminal helices of Get2CD that bind Get3 (Stefer et al., 2011; Mariappan et al., 2011) do not contribute to the remodeling of Get3 (Fig. 5 A). In contrast, the intrinsically disordered linker in Get2, which connects the N-terminal helices to its TMDs, contains two semi-conserved MoRF elements, H1 and H2, that are responsible for mediating Get3 opening. MoRFs are elements embedded within the IDRs of a protein that undergo disorder-to-order transitions upon binding with interaction partners (Malhis and Gsponer, 2015; Malhis et al., 2016). The role of H1 and H2 as MoRFs is supported by the conservation of multiple hydrophobic residues that could provide protein interaction sites and by secondary structure analyses that predicted the formation of amphiphilic helices in H1 and H2 (Figs. 5, 7, and S4). The photo–cross-linking of Get3 to Bpa probes engineered in H1 and H2 provided further evidence for direct contacts of these MoRFs with Get3. Although the precise interaction sites of H1/H2 on Get3 remain to be determined, an intriguing possibility is that the hydrophobic face of the H1/H2 helices could invade the TMD binding groove of Get3 to open Get3 and displace the bound TA.

Introduction of the ΔH1ΔH2 mutations did not substantially affect the affinity of mini-Get1/2 for Get3•TA. This indicates that the N-terminal helices of Get2 provide the anchoring contact that enables high-affinity binding of Get2 with Get3•TA, whereas the H1 and H2 MoRFs mediate transient interactions that do not contribute to equilibrium stability. The transient interaction of the Get2 MoRFs with Get3 may also explain the dynamic sampling of Get3 between open and closed conformations when bound to Get2CD and mini-Get1(NR)/2. When the Get3 interaction sites in the N-terminal helices of Get2 are removed in mini-Get1/2(RERR), however, the additional ΔH1ΔH2 mutations in Get2 modestly weakened the binding affinity to Get3, suggesting an ancillary role of H1 and H2 in assisting the interaction of Get1 with Get3. In the context of the Get1/2 complex at the ER, the additional interaction of the Get2 H1/H2 motifs with Get3•TA could bring the targeting complex closer to the membrane-proximal Get1CD and help initiate the interaction with and further remodeling of Get3 by Get1.

The combined effects of Get1CD and Get2CD in mini-Get1/2 generate a broad conformational landscape for Get3•TA, wherein both the open and closed conformations are sampled (Fig. 4). Unexpectedly, although the open state is more dominant in the mini-Get1/2(RERR)–bound Get3•TA complex (Fig. 4), TA release from this complex is significantly slower compared with that for mini-Get1/2–bound Get3•TA (Fig. 2). As mini-Get1/2(RERR) binds Get3•TA with a K_d value of 14.8 nM, the lower efficiency of Get3•TA disassembly cannot be attributed to incomplete binding. These results indicate that the Get1-induced opening of Get3•TA is necessary but not sufficient for optimal Get3•TA disassembly. It is plausible that the dynamic sampling of Get3 over a broad conformational landscape is important for optimal TA release. Alternatively or in addition, the hydrophobic face of the H1/H2 helices in Get2 could invade the TMD binding groove of Get3 and help displace the bound TA, as discussed earlier. The observed Get3•TA disassembly rate of ∼0.022 min⁻¹ by mini-Get1/2 is still slower than the rates of TA insertion into yeast rough microsomes from Get3•TA (0.045–0.14 min⁻¹; Rao et al., 2016), strongly suggesting that the TMDs of Get1/2 coordinate with the CDs to provide a more facile conduit for TA release compared with the external TA trap used here. Notably, mutations of H1 and/or H2 delayed TA insertion into the ER in vivo (Fig. 8, 1 min). These effects, though modest, support a role of the MoRFs in the optimal functioning of the Get1/2 receptor in vivo.

Our findings adds a mechanistically important new step for the membrane-associated molecular events in the GET pathway (Fig. 9). After ATP hydrolysis, the Get3•TA complex in the cytosol is initially captured by the N-terminal helices of Get2 (step 1). Get2 forms additional interactions with Get3•TA via the H1 and H2 MoRFs, which initiates the opening of Get3 and brings the targeting complex closer to the ER membrane and to Get1CD (step 2). The interaction of Get1CD with Get3•TA induces further opening of Get3, which together with the Get2 H1 and H2 MoRFs generates a conformational distribution of Get3 (step 3) that is optimal for triggering the release of TA and its subsequent insertion into the lipid bilayer (step 4).

Emerging data highlight key roles of IDRs in the function of receptors and translocase complexes during membrane protein biogenesis. Get1/2 and its mammalian homologue, WRB/CAML, share structural homology with the insertase YidC (McDowell et al., 2020), a member of the Alb3/Oxa1/YidC membrane protein insertase family. Oxa1 and YidC use positively charged residues in their disordered C-terminal CDs to mediate interaction with ribosomes during cotranslational targeting (Geng et al., 2015; Jia et al., 2003; Szyrach et al., 2003). The disordered C-terminal domain of Alb3 contains multiple conserved motifs rich in basic residues, which mediate the recruitment and remodeling of the chaperone cpSRP43 to facilitate post-translational membrane protein insertion into the chloroplast thylakoid membrane (Falk et al., 2010; Liang et al., 2016). Analogously, this and previous work identified multiple protein interaction motifs in the disordered Get2CD of the Get1/2 complex that play crucial roles in recruiting and remodeling the targeting factor Get3 during TA insertion (Mariappan et al., 2011; Stefer et al., 2011). A recent study identified a MoRF in the disordered linker of the mammalian SRP receptor, which plays an essential role in the ribosome-induced SRP receptor assembly during cotranslational protein targeting (Hwang Fu et al., 2019). In general, MoRF elements within IDRs can provide binding sites through transient interactions with binding partners to mediate dynamic cellular processes, such as signaling and complex assembly (van der Lee et al., 2014). The dynamic nature of IDRs also allows MoRFs to recruit binding partners over a longer radius than structured protein domains with the same amino acid length. Our work here further uncovered a new role of MoRFs in the conformational remodeling of protein complexes, which could be envisioned in a variety of protein translocation pathways. As such, interaction motifs embedded within IDRs could serve as interaction hubs to aid in the function of a broad spectrum of receptors and translocase complexes.

CRISPR/Cas9-mediated genome editing was used to mutate genomic GET1 and GET2 in the BY4741 strain to generate strains harboring GET2FLAG, get2^RERRFLAG, get1^NRGET2FLAG, get2^ΔH1FLAG, get2^ΔH2FLAG, and get2^ΔH1ΔΗ2FLAG (Ryan et al., 2016). Genomic GET2 was first replaced by URA3 by cotransforming 1 µg pCAS plasmid encoding Streptococcus pyogenes Cas9, an sgRNA with a guide sequence (5′-CAA​ATG​ACT​GCT​TGC​CTG​CC-3′) that targets the codon for GET2(Q36), and 5 µg double-stranded linear repair DNA encoding URA3 flanked with sequences from GET2 (5′-CCA​TGT​TTG​TAG​CAT​CAG​CAA​CGT​AGC​TCT​AGG​AAA​TAT​GTC-3′ and 5′-AAA​TTC​TAG​ATA​GTT​CAG​TAA​TAT​AAT​ACA​TTG​TT-3′). Cells were selected on synthetic defined (SD)-uracil (URA)+G418 (200 mg/l) agar plates. The second round of CRISPR-Cas9 editing replaces URA3 (pCAS URA3) with GET2FLAG or its mutants using a double-stranded repair DNA fragment that encodes the desired Get2 variant and is flanked by sequences 400 bp upstream and 250 bp downstream of genomic GET2. Positive colonies were doubly selected by growth on SD media and no growth on SD-URA media during the second round of CRISPR editing. The final strains were verified by sequencing.

Get3, Get1CD, Get2CD, intein-cpSRP43, and their variants were expressed and purified as previously described (Rome et al., 2014; Liang et al., 2016) with slight modifications. In brief, Get3 or its variants were fused to a N-terminal His-SUMO tag. The fusion protein was expressed in E. coli BL21 Star (DE3) cells and purified through Ni Sepharose High Performance (GE Healthcare) affinity chromatography followed by the His₆-SUMO protease digestion to remove the His₆-SUMO tag. The untagged Get3 was further purified through gel filtration using a Superdex 200 10/300 GL (GE Healthcare) and buffer exchanged into GET storage buffer (50 mM K-Hepes, pH 7.4, 150 mM KOAc, 5 mM Mg[OAc]₂, 20% glycerol, and 2 mM β-mercaptoethanol). Get1CD and Get2CD and their variants were fused to a N-terminal His₆ tag followed by a PreScission protease cut site. His₆-PreScission-Get1CD and His₆-PreScission-Get2CD were purified through nickel-coupled nitrilotriacetic acid (Ni-NTA; GoldBio) affinity chromatography followed by His₆-PreScission protease (a gift of André Hoelz, California Institute of Technology, Pasadena, CA) digestion to remove the His₆ tag. His₆-PreScission protease was further removed by Ni affinity chromatography. The untagged protein was further dialyzed against the GET storage buffer. His₆-cpSRP43 was purified through Ni affinity chromatography followed by anion exchange using a MonoQ 10/100 GL column. The protein was then dialyzed against storage buffer (50 mM K-Hepes, pH 7.4, 200 mM NaCl, and 20% glycerol). T7 RNA polymerase was fused to an N-terminal His₆-PreScission tag. After Ni affinity chromatography, the His₆ tag was removed by PreScission protease, and the untagged T7 RNA polymerase was further purified using a Superdex 200 10/300 GL column (GE Healthcare) and buffer exchanged into the storage buffer (50 mM K-Hepes, pH 7.4, 200 mM NaCl, and 20% glycerol). The purified T7 RNA polymerase was supplemented with 50% glycerol and stored at −30°C. Cm-tRNA synthetase (CmRS) was fused to an N-terminal His₆-tag followed by a PreScission protease cut site (Rao et al., 2016). His₆-PreScission-CmRS was first purified using Ni Sepharose High Performance (GE Healthcare) chromatography, and then the His₆-tag was removed using His₆-PreScission protease. The reaction was next passed through Ni Sepharose High Performance to remove the His₆-PreScission protease and undigested CmRS. Purified CmRS was dialyzed into buffer (50 mM K-Hepes, pH 7.4, 200 mM NaCl, and 20% glycerol), supplemented with 50% glycerol and stored at −30°C. Sfp phosphopantetheinyl transferase (Sfp) was fused to a C-terminal His₆-tag (Yin et al., 2006). Sfp-His₆ was first purified using Ni-NTA (Qiagen), and then the elution was dialyzed against storage buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 10% glycerol).

Mini-Get1/2 was coexpressed using the pETDuet-1 vector (kind gift from Vladimir Denic, Harvard University, Cambridge, MA) as previously described. Briefly, Get1CD(20–102) and Get2CD(1–150) were fused to the engineered C-terminal sequence deriving from pActPL-Gal4AD (Addgene plasmid #15303; EE12RR345L) and pActPL-Gal4DBD (Addgene plasmid #15304; RR12EE345L) that form a stable cc structure (Moll et al., 2001; Wang et al., 2011; Luan et al., 2006). Get1CD contains a His₆ tag at its extreme C terminus that allows for affinity purification, while Get2CD-cc is untagged and was copurified by forming a heterodimer with Get1CD-cc. Mini-Get1/2 was purified using Ni Sepharose High Performance resin (Ni-HP; Cytiva) followed by size exclusion chromatography on a Superdex 75 10/300 GL (GE Healthcare) in GET storage buffer.

The photo–cross-linker p-benzoylphenylalanine (Bpa) was incorporated into residue 85 or 111 in Get2 using amber suppression (Young et al., 2010). The plasmid pEVOL-BpaRS/tRNA_CUA^Opt, which encodes for optimized amber suppressor tRNA (tRNA_CUA^Opt) and evolved tRNA synthetase for Bpa (BpaRS), was cotransformed with pET33b-Get2CD^Amb-FLAG-PreScission-His₆ into BL21 Star (DE3) cells (Thermo Fisher Scientific). Cells were grown in LB Miller media at 37°C to OD₆₀₀ of 0.4, at which expression of tRNA_CUA^Opt and BpaRS was induced by addition of 0.2% arabinose and 1 mM Bpa. The expression of Get2CD^Amb-FLAG-PreScission-His₆ was induced at OD₆₀₀ of 0.6 with 0.1 mM IPTG, and cell growth was continued for 90 min before harvesting. Full-length Get2CD^Bpa-FLAG-PreScission-His₆ was purified using Ni-NTA affinity chromatography. The elution was mixed with His₆-tagged PreScission protease and dialyzed overnight against buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10% glycerol, and 10 mM imidazole at 4°C. Cleaved Get2CD^Bpa-FLAG was further purified by passing through Ni-NTA to remove the His₆ tag and His₆-PreScission protease. The purified Get2CD^Bpa-FLAG was collected in flowthrough and dialyzed in GET storage buffer (50 mM K-Hepes, pH 7.4, 150 mM KOAc, 5 mM Mg[OAc]₂, 20% glycerol, and 2 mM β-mercaptoethanol) at 4°C overnight and stored at −80°C.

The model TA substrate used for generating the Get3^{ATTO550/ATTO647N}•TA complex was 3xStrep-SUMOnc-Bos1-opsin (Rao et al., 2016; Chio et al., 2019), which contains three tandem N-terminal Strep tags, a mutant yeast Smt3 with the Ulp1 cleavage site removed, residues 203–244 of the SNARE protein Bos1 encompassing its TMD, followed by a C-terminal opsin tag (GSMRMNGTEGPNMYMPMSNKTVD) to monitor the successful insertion into the ER via glycosylation. For fluorescence labeling of TA, the fourth residue (Ala) upstream of Bos1 was replaced by the amber codon to allow amber suppression with the fluorescent amino acid 7-hydroxycoumarin (described in the next section).

The model TA substrate used in the in vivo pulse-chase assay was 3xHA-BirA-Bos1TMD-opsin (abbreviated as BirA-Bos1; Cho and Shan, 2018), which contains an N-terminal 3xHA, BirA, residues 203–244 of Bos1, a GSGGS linker, and a C-terminal opsin tag. Expression of BirA-Bos1 was controlled by the GPD promotor and PGK terminator that were cloned into pRS316 (Cho and Shan, 2018).

Get3 with a ybbR tag (DSLEFIASKLA) inserted between residues S110 and D111 was labeled with BDP-maleimide (Thermo Fisher Scientific) or with a 1:1 mixture of ATTO550-maleimide (GE Healthcare) and ATTO 647N-maleimide (ATTO-TEC) via Sfp-catalyzed incorporation of dye-CoA conjugates (Rao et al., 2016; Chio et al., 2019, 2017b). The C-terminal His₆-tagged Sfp was removed through Ni-NTA affinity chromatography. The excess dye–CoA conjugates were further removed through gel filtration using a Sephadex G-25 (Sigma-Aldrich) column. Get3^{ATTO550/ATTO647N}-TA complex and Get3^BDP-TA^Cm were generated by in vitro translation of TA substrate 3xStrep-SUMOnc-Bos1-opsin in E. coli S30 cell extract (Rao et al., 2016) supplemented with Get3 variants and untagged T7 RNA polymerase and/or untagged CmRS (for generating Get3^BDP-TA^Cm). The resulting Get3-TA complex was purified using Strep-Tactin Sepharose (IBA Life Sciences; Rao et al., 2016; Chio et al., 2019).

Mini-Get1/2 was labeled with TMR-5-maleimide (Invitrogen) via thiol-maleimide reaction on an engineered single cysteine at residue S77 of Get1 or residue T34 of Get2. The labeling reaction was quenched with 1 mM DTT, and excess dye was removed using a PD-10 column packed with Sephadex G-25 resin (GE Healthcare).

The cross-linked His-Get3×Get2CD^Bpa-FLAG protein band was detected by Western blot against His-Get3 and Get2CD^Bpa-FLAG separately. Purified His-Get3 and Get2CD^Bpa-FLAG variants were incubated in buffer (50 mM K-Hepes, pH 7.5, 150 mM KOAc, 5 mM Mg[OAc]₂, and 2 mM β-mercaptoethanol) at 0.1 µM and 20 µM (for anti-His Western blot) or 10 µM and 1 µM (for anti-FLAG Western blot) concentrations and cross-linked with UV (UVP B-100AP lamp; UVP LLC) on dry ice for 2 h or 15 min (with new lamp). The sample tubes were placed 1 cm away from the lamp source. After cross-linking, samples were separated using a 12% SDS-PAGE and analyzed by Western blot against His-Get3 using the anti-His primary antibody (Cat#A00186; GenScript) or against Get2CD^Bpa-FLAG using the anti-FLAG (anti-DYKDDDDK) primary antibody (Cat#A00187; GenScript) separately. IRDye 800CW Goat anti-Mouse polyclonal IgG (H + L; Cat#925-32210; LI-COR Biosciences) was used as the secondary antibody to visualize the protein bands using the Odyssey imaging system.

Yeast cells were first treated with 0.3 M NaOH at room temperature for 3 min. After discarding the NaOH, cells were rinsed with water once and then resuspended in lysis buffer (50 mM Hepes, pH 7.4, 150 mM NaCl, 2% SDS, and 2 mM β-mercaptoethanol). Samples were then heated at 95°C for 5 min, and the supernatant was used for the further Western blot. Samples were separated using a 12% SDS-PAGE. The abundance of Get2 in yeast strains was detected by Western blot against the C-terminal FLAG tag on Get2. Model TA substrate translocation levels at steady-state in WT and Get2 variant strains were detected by anti-HA (Cat#A01244; GenScript) Western blot. Anti-PGK1 (Cat#ab197960; Abcam) Western blot was used as a loading control.

All proteins and protein complexes were ultracentrifuged (TLA100; Beckman Coulter Inc.) at 100,000 rpm for 30 min at 4°C to remove aggregates before use in in vitro assays.

All fluorescence measurements were performed on a Fluorolog 3–22 spectrophotometer (HORIBA Instruments) at 25°C in GET assay buffer (50 mM K-Hepes, pH 7.4, 150 mM KOAc, 5 mM Mg[OAc]₂, 10% glycerol, and 1 mM DTT).

Observed E_FRET values were normalized by dividing with E_max, the FRET efficiency at saturating titrant concentrations. Normalized E_FRET was plotted against mini-Get1/2 concentration and fit to Eq. 2:

in which K_d is the equilibrium K_d between Get3•TA and mini-Get1/2. The Get3•TA concentration is 10 nM.

The S of purified mini-Get1/2 was characterized by SEC-MALS. 200 µl of 50 µM WT mini-Get1/2 was buffer-exchanged into 20 mM Tris, pH 8.0, 150 mM NaCl, and 0.02% NaN₃. The sample was injected onto a Superdex 75 10/300 GL column (Cytiva, formerly GE Healthcare) equilibrated in the same buffer. The chromatography column was connected with an 18-angle light-scattering detector (DAWN HELEOS II; Wyatt Technology) and a refractive index detector (Optilab t-rEX; Wyatt Technology). Data were collected every second at 25°C at a flow rate of 0.5 ml/min. The molar mass of mini-Get1/2 was then calculated using the ASTRA 6 software (Wyatt Technology).

Yeast cells harboring the indicated Get1 or Get2 variants were transformed with the plasmid pRS316-GPD-3xHA-BirA-Bos1TMD-opsin and grown in 30 ml SD-Ura media to mid-log phase (OD₆₀₀ ∼0.6) at 30°C. Cells were harvested, washed with 10 ml SD-Ura-Met-Cys media twice, resuspended in SD-Ura-Met-Cys media to a final OD₆₀₀ of 12, and incubated in a water bath at 30°C for 25 min. The remainder of the pulse-chase experiments were performed at 25°C for 5 min. Cells were pulse-labeled with 100 µCi/ml EasyTag EXPRESS35S Protein Labeling Mix (Perkin Elmer) for 2 min and chased with an equal volume of SD-Ura supplemented with 10 mM cold methionine and 1 mM cold cysteine, respectively. 450-µl aliquots of cells were removed at indicated times during chase and flash-frozen in liquid nitrogen. The 3xHA-tagged TA substrate was immunoprecipitated using anti-HA magnetic beads (Thermo Fisher Scientific) as previously described (Cho and Shan, 2018). Samples were analyzed by SDS-PAGE on 15% Tris-glycine gels and autoradiography. Successful TA insertion will lead to glycosylation at the C-terminal opsin tag in ER lumen. TA insertion efficiency was calculated using equation I_glycosylated/(I_{glycosylated +} I_{nonglycosylated})%, in which “I” denotes the band intensity quantified using ImageJ from autoradiography.

All protein samples were ultracentrifuged in a TLA 100 rotor (Beckman Coulter) at 100,000 rpm for 30 min at 4°C to remove aggregates before all µs-ALEX measurements. Proteins were diluted and incubated in GET assay buffer (50 mM K-Hepes, pH 7.4, 150 mM KOAc, 5 mM Mg[OAc]₂, 10% glycerol, and 1 mM DTT) supplemented with 0.3 mg/ml BSA during the assay.

For each measurement, fluorescently labeled Get3^{ATTO550/ATTO647N} or Get3 ^{ATTO550/ATTO647N}•TA was diluted to ∼200 pM final concentration and mixed with the indicated binding partners. Reactions contained 2 mM ATP or ADP, 20 µM Get1CD or Get2CD variants, or 20 µM mini-Get1/2 variants where indicated. Samples were mixed and placed either on a coverslip (for short-time measurements, i.e., up to 15 min) or in a closed chamber made by sandwiching a perforated silicone sheet (Grace Bio-Labs) with two coverslips to prevent evaporation (for long-time measurements, i.e., 30–120 min). Data were collected using an alternating-laser excitation fluorescence-aided molecule sorting setup (Kapanidis et al., 2005) with two single-photon Avalanche photodiodes (PerkinElmer) and 532-nm and 635-nm continuous wave lasers (Opto Engine LLC) operating at 135 µW and 80 µW, respectively.

All µs-ALEX data analyses were performed using FRETBursts, a Python-based open-source burst analysis toolkit for freely diffusing based smFRET (Ingargiola et al., 2016). All-photon burst search was performed to identify each burst as a minimum of 10 consecutive photons with a photon count rate at least 15-fold higher than the background rate during both the donor and acceptor excitation periods. Due to the fluctuating background rate within a measurement, the background rate was computed for every 50-s interval according to maximum likelihood fitting of the interphoton delay distribution. Dual-channel burst search (Nir et al., 2006) was performed to separate the burst species containing FRET pairs from background noise and species containing donor or acceptor only. The identified bursts were further filtered according to the following criteria: (1) $nDD+nDA ≥15$ (exclude acceptor only species) and (2) $nAA≥15$ (exclude donor only species), where $nDD$ is the number of photons detected from the donor channel during donor excitation, $nDA$ is the number of photons detected from the acceptor channel during donor excitation, and $nAA$ is the number of photons detected from the acceptor channel during acceptor excitation.

To investigate the effects of Get1CD and Get2CD on dye photophysics, we analyzed the peak photon rate of donor- or acceptor-labeled Get3 in the presence of the respective ligands (Fig. S2 C). To identify signals for Donor-only species, we searched for bursts with a minimum of 10 consecutive photons detected from the donor channel at a photon count rate that is at least 15-fold higher than the background rate during the donor excitation periods. The same burst search parameters were applied to photons detected from the acceptor channel during the acceptor excitation periods to identify bursts for acceptor-only species. The bursts were further screened according to the following: for donor-only species, (1) $nDD≥50$ and (2) $S≥0.85, E≤0.15$⁠; and for acceptor-only species, (1) $nAA≥50$ and (2) $S≤0.2$⁠.

The mean peak count rates (MPRs, the means of individual bursts’ peak photon count rates) were computed using FRETBursts and were used to present dye brightness.

Fig. S1 shows purification and characterization of mini-Get1/2 and Get2CD variants. Fig. S2 shows additional smFRET analysis performed in this study. Fig. S3 shows smFRET measurements of Get3 bound to the Get2CD linker deletion mutants. Fig. S4 shows Get2CD (1–150) structure predictions. Fig. S5 shows replicates for the pulse-chase assay to measure the kinetics of TA insertion into the ER in vivo.

We thank Vladimir Denic for the mini-Get1/2 construct, Jennifer Keefe for help with SEC-MALS, and members of the Shan laboratory for general discussion and comments on the manuscript.

This work was supported by Dean Willard Chair funds to S. Weiss and National Institutes of Health grants R01 GM107368 and R35 GM136321 and the Gordon and Betty Moore Foundation grant GBMF2939 to S.-o. Shan.

The authors declare no competing financial interests.

Author contributions: U.S. Chio, Y. Liu, and S.-o. Shan designed research; U.S. Chio, Y. Liu, W.J. Shim, and S. Chandrasekar performed biochemical experiments; U.S. Chio and Y. Liu analyzed biochemical data with input from S.-o. Shan; U.S. Chio, Y. Liu, and SY. Chung performed sm fluorescence experiments; U.S. Chio, Y. Liu, and SY. Chung analyzed biophysics data with input from S. Weiss and S.-o. Shan; U.S. Chio, Y. Liu, and S.-o. Shan wrote the manuscript; SY. Chung and S. Weiss revised the manuscript. All authors approved the final manuscript.