Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Oct 26.
Published in final edited form as: Cell Rep. 2017 Sep 5;20(10):2287–2293. doi: 10.1016/j.celrep.2017.08.035

Tail-anchored protein insertion by a single Get1/2 heterodimer

Benjamin E Zalisko 1, Charlene Chan 2, Vladimir Denic 2,*, Ronald S Rock *, Robert J Keenan 1,3,*
PMCID: PMC5657598  NIHMSID: NIHMS900719  PMID: 28877464

SUMMARY

The Get1/2 transmembrane complex drives the insertion of tail-anchored proteins from the cytosolic chaperone Get3 into the endoplasmic reticulum membrane. Mechanistic insight into how Get1/2 coordinates this process is confounded by a lack of understanding of the basic architecture of the complex. Here we define the oligomeric state of full-length Get1/2 in reconstituted lipid bilayers by combining single-molecule and bulk fluorescence measurements with quantitative in vitro insertion analysis. We show that a single Get1/2 heterodimer is sufficient for insertion and demonstrate that the conserved cytosolic regions of Get1 and Get2 bind asymmetrically to opposing subunits of the Get3 homodimer. Taken together, our results define a simplified model for how Get1/2 and Get3 coordinate TA protein insertion.

eTOC blurb

Tail-anchored membrane proteins are inserted into the endoplasmic reticulum via the post-translational GET pathway. Zalisko et al. combine single-molecule and bulk fluorescence measurements with quantitative in vitro insertion assays to define the architecture of the heterodimeric Get1/2 insertase and its engagement with the soluble chaperone Get3.

INTRODUCTION

Tail-anchored (TA) proteins, defined by a single carboxy-terminal transmembrane domain (TMD) and a cytosolic-facing amino-terminal domain (Kutay et al., 1993), are post-translationally inserted into the ER membrane via the evolutionarily conserved GET pathway (Denic et al., 2013; Hegde and Keenan, 2011; Stefanovic and Hegde, 2007). In yeast, the membrane targeting factor is a homodimeric ATPase called Get3, which changes conformation in a nucleotide-dependent manner (Bozkurt et al., 2009; Hu et al., 2009; Mateja et al., 2009; Suloway et al., 2009; Yamagata et al., 2010) to bind TA proteins in the cytosol and release them at the ER membrane. The Get1/2 transmembrane complex (Schuldiner et al., 2008) recruits the Get3-TA targeting complex (Mateja et al., 2015) to the ER, coordinates TA protein release and insertion, and mediates ATP-driven recycling of Get3 to the cytosol (Mariappan et al., 2011; Stefer et al., 2011; Wang et al., 2014; Wang et al., 2011).

Get1 and Get2 are the only integral membrane components required for TA protein insertion (Mariappan et al., 2011; Wang et al., 2011), and specific functions for the individual subunits have been defined based on a series of functional and structural studies (Kubota et al., 2012; Mariappan et al., 2011; Stefer et al., 2011; Wang et al., 2011). The long N-terminal cytosolic domain of Get2 facilitates initial recruitment of the targeting complex, while the cytosolic coiled-coil of Get1 drives TA protein release. Following release, the TMDs of both Get1 and Get2 contact the TA protein as it inserts into the bilayer (Wang et al., 2014), and ATP binding enhances dissociation of Get3 from the Get1 coiled-coil, facilitating Get3 recycling to the cytosol.

Despite these mechanistic insights, how full-length Get1 and Get2 function together to coordinate events at the membrane remains unclear. Although Get1/2 complexes can be isolated with Get3 from yeast rough microsomes (Auld et al., 2006; Jonikas et al., 2009), the quaternary structure of the Get1/2/3 complex is undefined. Crystal structures of the cytosolic Get1 or Get2 fragments bound symmetrically to different nucleotide states of homodimeric Get3 have led to closely related models involving a heterotetrameric Get1/2 assembly of two Get1 and two Get2 subunits (Mariappan et al., 2011; Stefer et al., 2011) (Fig. 1a); such an assembly might exist constitutively or it might form dynamically in the presence of Get3 (Fig. 1b). Simpler models are also plausible, including a heterodimeric Get1/2 assembly with only one copy of each subunit (Fig. 1c).

Figure 1. Alternative models for the Get1/2 architecture.

Figure 1

Open in a new tab

(a) In static heterotetramer models, based on structural and biochemical studies (Mariappan et al., 2011; Stefer et al., 2011), Get1/2 exists as a constitutive complex comprising two copies of each subunit. Accordingly, the Get3-TA protein complex is captured by two copies of Get2, which bind on opposite sides of the symmetric Get3 dimer. Subsequently, the ADP or apo form of the Get3-TA complex is handed off to Get1, which displaces each Get2 subunit such that two Get1 subunits bind on opposite sides of Get3. Alternatively, Get1 only partially displaces Get2, such that two Get1 and two Get2 subunits are bound to Get3. Importantly, all heterotetramer models predict that two copies of Get1 and/or Get2 bind simultaneously to Get3 at different steps in the pathway. In such models, conformational changes in Get3 could be coupled to the transmembrane domains of Get1/2 by rigid interactions mediated by the coiled-coil (not shown). (b) In a dynamic model, Get3 binding drives the transient assembly of two heterodimeric Get1/2 complexes into a single heterotetrameric complex. (c) In a static heterodimer model, the Get3-TA complex is initially captured by a single copy of Get2 bound to one side of the Get3 dimer; Get1 then engages the ADP-bound or apo form of the Get3-TA complex by partially displacing Get2, or by binding to the opposite side of the Get3 dimer.

Defining the oligomeric state of the functional Get1/2 complex is critical for understanding its molecular mechanism. For example, the number of subunits present in the functional complex likely dictates whether the Get1 and Get2 cytosolic domains bind competitively or simultaneously to the same or opposite sides of the Get3 homodimer at various stages along the pathway (Mariappan et al., 2011; Stefer et al., 2011) (Fig. 1). Likewise, the number of subunits has important implications for how the Get1/2 TMDs guide TA substrates into the bilayer (Wang et al., 2014) and whether conformational changes in Get3 can be coupled to these TMDs during insertion (Denic et al., 2013; Hegde and Keenan, 2011; Stefer et al., 2011). Thus, a rigorous description of how Get1/2 coordinates key steps at the membrane requires knowledge of its quaternary structure. Here we show that the minimal functional unit of the full-length Get1/2 complex is a heterodimer, which drives the insertion of TA proteins by binding to opposites sides of the Get3 homodimer.

RESULTS

To gain insight into the organization of the Get1/2 complex, we developed a bulk FRET assay in proteoliposomes that reports on changes in the proximity of Get1 and/or Get2 subunits upon binding to Get3. We introduced single cysteines at membrane-proximal (Get1-A95C; Get2-E220C) or cytosolic (Get1-S77C; Get2-S28C) positions within Get1 and Get2 (Fig. 2a). After purification (Fig. S1), the individual subunits were labeled with FRET donor (Cy3) or acceptor (Cy5) fluorophores. Cytosolic or membrane-proximal labeled Get1 and Get2 subunits were then reconstituted into proteoliposomes in different donor-acceptor combinations: Get1Cy3-Get1Cy5-Get2, Get1Cy3-Get2Cy5 (membrane proximal), Get1Cy5-Get2Cy3 (cytosolic) and Get1-Get2Cy3-Get2Cy5. Importantly, these proteoliposomes are reconstituted at a high protein-to-lipid ratio (12:10,000), such that they contain multiple copies of Get1/2. This allows for an unbiased analysis of Get3 binding to different oligomeric states of Get1/2. After verifying the insertion activity of the different proteoliposomes (Fig. S2a), we monitored changes in FRET as a function of binding to different nucleotide states of Get3.

Figure 2. Bulk FRET measurements of Get3 binding to Get1/2 complexes in proteoliposomes.

Figure 2

Open in a new tab

(a) Get1 and Get2 subunits were labeled with FRET donor Cy3 (green) or FRET acceptor Cy5 (red) fluorophores at membrane proximal or cytosolic positions and then reconstituted into proteoliposomes in different donor/acceptor combinations. (b) Proteoliposomes were reconstituted at 12 Get1/2 molecules per 10,000 lipids (‘High Get1/2-to-lipid’). Histogram shows the FRET change after addition of 50 nM Get3 (dark grey), Get3 + ADP (grey) or an ATPase-deficient Get3-D57N mutant + ATP (white) to each of the six different proteoliposomes (10 nM Get1/2). A significant Get3-dependent FRET increase is only observed when the donor/acceptor pairs are located on the cytosolic positions of Get1 and Get2 subunits. All samples show a non-specific FRET component in the absence of Get3 due to co-reconstitution of multiple donors and acceptors in the same proteoliposomes (Fig. S2b); we interpret Get3-dependent FRET decreases as an increase in the average distance between labeled Get1/2 heterodimers due to steric hindrance caused by Get3 binding. This decreased FRET is eliminated in proteoliposomes reconstituted at lower protein-to-lipid ratios (see (d) below and Fig. S2b). (c) At left, FRET-based titration of cytosolically-labeled Get1Cy5-Get2Cy3 proteoliposomes (10 nM Get1/2) with Get3-D57N. At right, disruption of the Get1-Get2-Get3 interaction in the same proteoliposomes, monitored by the change in FRET upon titration with ATP. (d) As in (b), but after reconstitution at lower Get1/2-to-lipid ratios (1.2 Get1/2 molecules per 10,000 lipids; ‘Low Get1/2-to-lipid’). See also Figures S1 and S2.

We first explored whether Get3 binding drives Get1/2 towards a higher oligomeric state, as would be expected in the dynamic model (Fig. 1b). When proteoliposomes containing different combinations of membrane-proximal labels were incubated with Get3, we observed no significant FRET increase, regardless of which subunits were labeled or the nucleotide state of Get3 (Fig. 2b and Fig. S2b). Thus, consistent with static models, Get3 binding does not drive assembly of a higher-order oligomer of Get1/2.

Next, we used proteoliposomes containing different combinations of cytosolically labeled Get1 and Get2 to explore how they engage Get3. The cytosolic coiled-coil of Get1 and the long (150-residue) unstructured N-terminus of Get2 do not interact with each other, but bind to overlapping sites present on each of two sides of the Get3 homodimer (Mariappan et al., 2011; Stefer et al., 2011). These cytosolic regions are expected to FRET most efficiently when brought into close proximity by simultaneous binding to Get3. Importantly, studies with the isolated cytosolic fragments show that the Get2-Get3 interaction is insensitive to nucleotide, but the Get1-Get3 interaction is weakened by ADP and completely disrupted by ATP (Mariappan et al., 2011; Rome et al., 2014; Stefer et al., 2011; Wang et al., 2011).

When Get3 was incubated with cytosolically-labeled Get1Cy3-Get1Cy5-Get2 or Get1-Get2Cy3-Get2Cy5 proteoliposomes, no significant FRET increase was observed, regardless of the nucleotide state (Fig 2b and Fig. S2b). Likewise, no significant FRET increase was observed when ATP-bound Get3 was incubated with Get1Cy5-Get2Cy3 proteoliposomes. However, when ADP-bound or nucleotide-free Get3 was incubated with Get1Cy5-Get2Cy3 proteoliposomes we observed a strong FRET increase (Fig 2b and Fig. S2b). Importantly, when cytosolic-labeled Get1Cy5-Get2Cy3 proteoliposomes (10 nM) were titrated with Get3, we observed a linear FRET increase that became saturated at one equivalent of Get3 (10 nM homodimer) (Fig. 1c), and could be reversed by ATP (Fig. 2c). These data are consistent with Get3 binding with sub-nanomolar affinity to a single Get1/2 heterodimer.

The quantitative and reversible Get3-dependent FRET increase observed with cytosolically-labeled Get1Cy5-Get2Cy3 proteoliposomes provides direct evidence for simultaneous binding of full-length Get1 and Get2 to the nucleotide-free and ADP-bound states of Get3. Moreover, the absence of Get3-dependent FRET increases with cytosolically-labeled Get1Cy3-Get1Cy5-Get2 or Get1-Get2Cy3-Get2Cy5 proteoliposomes suggests that Get3 does not bind concomitantly to two Get1 and/or two Get2 subunits in any nucleotide state, consistent with a heterodimeric Get1/2 complex.

To recapitulate the result with single Get1/2 heterodimers per liposome, we repeated the bulk FRET experiment with all donor/acceptor combinations reconstituted at a lower protein-to-lipid ratio (1.2:10,000). Consistent with the presence of ~1 Get1/2 heterodimer per liposome, we observed negligible FRET in Get1Cy3-Get1Cy5-Get2 and Get1-Get2Cy3-Get2Cy5 proteoliposomes, but clear FRET for both cytosolic labeled Get1Cy5-Get2Cy3 and membrane proximal Get1Cy3-Get2Cy5 proteoliposomes (Fig. S2b). Notably, Get3 binding did not significantly change FRET between membrane-proximal labeled Get1Cy3-Get2Cy5 (Fig. 2d), indicating that the Get3-dependent FRET decreases observed at higher protein-to-lipid ratios (Fig. 2c) result from the presence of multiple copies of labeled Get1/2 in the same proteoliposome. However, as with the higher protein-to-lipid ratio reconstitutions, the addition of ADP-bound or nucleotide-free Get3 increased FRET between cytosolically-labeled Get1Cy5-Get2Cy3 (Fig. 2d). While these bulk FRET data are consistent with a model in which Get3 binds to a heterodimeric Get1/2 complex comprising a single copy of each subunit, they do not formally exclude the possibility that Get1/2 functions as a heterotetrameric or higher order complex.

To rigorously test this heterodimeric Get1/2 model, we sought to quantify the number of Get1/2 complexes required for TA protein insertion into membranes. If the minimal functional unit is a Get1/2 heterodimer, then proteoliposomes containing a single Get1/2 heterodimer would be expected to have the same specific activity as proteoliposomes containing multiple Get1/2 heterodimers.

To prevent dissociation of the Get1 and Get2 subunits during reconstitution, we took advantage of an engineered single-chain Get1/2 (Get2-1sc) construct shown previously to be functional in yeast (Wang et al., 2014). We expressed and purified Get2-1sc from E. coli and verified its activity in vitro. Like native Get1/2, Get2-1sc, but not its variants containing inactivating mutations in the cytosolic fragments of Get1 (R73E) or Get2 (R17E), is functional for TA protein insertion in proteoliposomes (Fig. 3a).

Figure 3. Single Get1/2 heterodimers mediate TA protein insertion.

Figure 3

Open in a new tab

(a) Yeast rough microsomes (‘yRMs’) or the indicated proteoliposomes were tested for insertion of radiolabeled TA protein, Sec61β (FL), by a proteinase K protection assay. The appearance of a protected fragment (PF), which is diagnostic for insertion, was quantified by SDS-PAGE and autoradiography. Co-reconstituted Get1 and Get2 subunits (‘Get1/2’) show equivalent specific activity to Get2-1sc proteoliposomes. Mixtures of Get1-only and Get2-only proteoliposomes (‘Get1, Get2’), and single-chain constructs containing inactivating point mutations in either the Get1 (‘Get2-1*sc’; R73E in Get1) or Get2 (‘Get2*-1sc’; R17E in Get2) cytosolic domains show no activity. The four ‘Get2-1sc-Cy5’ samples correspond to the four protein-to-lipid ratio reconstitutions in panel (b), at increasing protein-to-lipid ratios. All reconstitutions were diluted with empty liposomes to a final concentration of 10 nM Get1/2. Notably, the normalized insertion activity is independent of the number of Get2-1sc-Cy5 molecules in each proteoliposome. Coomassie-stained proteinase K (‘PK’) was used as a loading control. Before performing the insertion assay, proper Get2-1sc-Cy5 normalization was confirmed by Cy5-imaged SDS-PAGE. For completeness, the concentration of Get2-1sc-Cy5 samples was also compared with unlabeled Get1/2 samples by stain-free SDS-PAGE; no sample was run for the most dilute Get2-1sc-Cy5 sample (marked with an asterisk), because of high lipid content, low protein concentration, and the large amount of sample required for stain-free detection. (b) Single-molecule photobleaching analysis of proteoliposomes from panel (a), reconstituted with Cy5-labeled single-chain Get1/2 (‘Get2-1sc-Cy5’) at different protein-to-lipid ratios. The number of photobleaching steps per labeled proteoliposome is shown in red; ‘x’ represents discarded traces. SDS-solubilized Get2-1sc-Cy5 serves as a monomeric control. The inset (black) shows the calculated proportion of Get2-1sc-Cy5 found in different oligomeric states, as described in the Supplemental Experimental Procedures. (c) In parallel with the assay shown in panel (a), Get2-1sc-Cy5 proteoliposomes (reconstituted at a ratio of 12 proteins per 10,000 lipids) were diluted to the indicated final concentrations, and TA protein insertion was quantified by autoradiography. This control experiment demonstrates that the assay is linear up to a total Get2-1sc-Cy5 concentration of ~10 nM, and is not limited by active targeting complex. See also Figures S1 and S3.

Next, we purified and fluorescently labeled a Get2-1sc construct harboring the Get1-S77C mutation (Fig. S1). By varying the protein–to-lipid ratio during reconstitution, the average number of Get2-1sc-Cy5 molecules per liposome could be adjusted. This was directly quantified by single-molecule photobleaching using TIRF microscopy (Fig. 3b and Fig. S3a–d). At the lowest protein-to-lipid ratio tested, more than 80% of Get2-1sc-Cy5 was incorporated into liposomes containing only one Get2-1sc-Cy5 molecule. At the highest protein-to-lipid ratios, we observed a corresponding increase in the number of molecules per liposome, with as much as 80% of Get2-1sc-Cy5 incorporated into liposomes containing two or more Get2-1sc-Cy5 molecules. Thus, proteoliposomes reconstituted at the highest protein-to-lipid ratio contain about four times as many Get2-1sc-Cy5 molecules—that could, in principle, oligomerize—than proteoliposomes reconstituted at the lowest protein-to-lipid ratio.

To measure the specific insertion activity of the different proteoliposome reconstitutions, we normalized the total Get2-1sc-Cy5 concentration in each sample by adding the appropriate volume of empty liposomes. If oligomerized Get2-1sc-Cy5 is required for insertion, photobleaching analysis predicts that proteoliposomes reconstituted at the lowest protein-to-lipid ratio should show at least 4-fold lower insertion activity than Get2-1sc-Cy5 reconstituted at the highest protein-to-lipid ratio. Strikingly, we found that the specific insertion activity was essentially independent of the number of copies of Get2-1sc-Cy5 per liposome; indeed, robust insertion was observed in proteoliposomes containing the largest proportion of single Get2-1sc-Cy5 molecules (Fig. 3b,c and Fig. S3e–i). Thus, we conclude that a single Get1/2 heterodimer is minimally required for TA protein insertion.

Finally, we sought structural information on how the two subunits of Get1/2 simultaneously contact Get3. Previous NMR studies showed that isolated cytosolic fragments of Get1 and Get2 can bind simultaneously to the same side of the Get3 dimer (Stefer et al., 2011) (Fig.1c and 4a). In this configuration, the two subunits are in close proximity, with Get1 displacing the second helix of Get2 (Fig. 4a and Fig. S4e). Notably, since Get1 and Get2 are bound to the same subunit of Get3, structural modeling predicts that their relative position will remain fixed irrespective of the Get3 conformational state (Fig. 4a). Alternatively, Get1 and Get2 might bind on opposite sides of the Get3 homodimer. In this case, the two subunits should be further apart. Moreover, because Get1 and Get2 are bound to different subunits of Get3, the distance between them is expected to increase as the Get3 dimer changes conformation from its ‘open’ to a ‘semi-open’ state in response to ADP binding (Kubota et al., 2012; Mariappan et al., 2011; Stefer et al., 2011) (Fig. 4a).

Figure 4. Get1 and Get2 bind on opposite sides of Get3 in its post-hydrolysis states.

Figure 4

Open in a new tab

(a) Models of Get1 (magenta) and Get2 (yellow) bound to the same or opposite sides of the nucleotide-free (‘open’; PDB 3ZS8) and ADP-bound (‘semi-open’; PDB 3VLC) conformations of the Get3 homodimer (blue, green). The models are aligned on one Get3 subunit (blue) to highlight the pseudo-rigid-body ‘swivel’ (grey arrows) of the other subunit (green) that accompanies ADP binding. When bound to opposite sides of Get3, Get1 and Get2 move apart in response to ADP binding; thus, the distance between donor and acceptor fluorophores on Get1-S77C-Cy5 (red) and Get2-S28C-Cy3 (green) is expected to increase when ADP is added. In contrast, when bound to the same side of Get3, the relative positions of Get1 and Get2 are fixed, since the primary contacts are made to the same subunit (blue). Thus, the distance between fluorophores is expected to remain constant as Get3 changes conformation in response to ADP binding. (b) Distributions of single-molecule FRET efficiencies (grey histograms) in nanodiscs containing cytosolically labeled Get1-S77C-Cy5 and Get2-S28C-Cy3 heterodimers during incubation with different nucleotide states of Get3. The solid black curves are the sums of the individual Gaussian functions (red lines) used to fit the raw data. In contrast with the constitutively high FRET efficiencies (>99%) predicted for same-side binding (Fig. S4e), intermediate FRET efficiencies are observed in the nucleotide-free and ADP-bound states. Moreover, the FRET efficiency of the ADP-bound state is less (53 ± 4%) than in the nucleotide-free state (71 ± 2%), as expected for an increase in the distance between fluorophores as Get3 changes conformation from the ‘semi-open’ to an ‘open’ state. (c) Model of the conformation-specific interactions between Get3 and the heterodimeric Get1/2 complex. See text for details. See also Figure S4.

To distinguish between these two possibilities, we used single-molecule FRET as a qualitative measure of the distance between Get1 and Get2 bound to different conformational states of Get3. Here we reconstituted cytosolically labeled Get1Cy5-Get2Cy3 into nanodiscs, yielding a more homogeneous population of single Get1/2 heterodimers (Fig. S4), and providing a cleaner system for use in TIRF flow cells. After verifying the heterodimeric Get1/2 composition of the nanodiscs by single-molecule photobleaching, we showed that they are functional for TA protein release from Get3 (Fig. S4c,d).

Next, we measured the distribution of FRET efficiencies between the Get1 and Get2 cytosolic domains bound to different nucleotide states of Get3, and fit each dataset to two Gaussian distributions (Fig. 4b). In the absence of Get3, we observed a broad distribution of low FRET states corresponding to long inter-dye distances, as expected for non-interacting cytosolic domains. Similarly, in the presence of ATP-bound Get3, we observed a broad distribution of low FRET states, consistent with the inability of Get1 to interact with ATP-bound Get3. However, when ADP-bound Get3 was added, an intermediate FRET state was observed at 53 ± 4%, and this shifted to a higher FRET state of 71 ± 2% in the presence of nucleotide-free Get3.

The magnitude of the observed FRET efficiencies is inconsistent with simultaneous Get1 and Get2 binding on the same side of Get3, which would be expected to give rise to high FRET (~99%) in the nucleotide-free and ADP-bound Get3 samples. Moreover, the observed shift from intermediate FRET in the ADP-bound state to higher FRET in the nucleotide-free state is consistent with the expected changes if Get1 and Get2 bind on opposite sides of the Get3 homodimer. Indeed, given the Förster distance of the Cy3 and Cy5 FRET pair (60 Å) (Murphy et al., 2004) and assuming that the fluorophores freely rotate at the labeling site (i.e., κ2 = 2/3), the relative change in distance between fluorophores in the nucleotide-free and ADP-bound states (~7 Å) is in qualitative agreement with the expected distance change based on structural modeling (~10 Å) (Fig. 4a and Fig. S4e). These data are most consistent with a model in which Get1 and Get2 bind simultaneously to opposite sides of the Get3 homodimer.

DISCUSSION

The previously undefined nature of the oligomeric state of the Get1/2 transmembrane complex has limited our understanding of how Get1, Get2, and Get3 coordinate TA protein insertion. Here, using full-length Get1 and Get2 in lipid bilayers, we show that the minimal functional unit of Get1/2 is a heterodimer comprising a single copy of each subunit. Notably, even when presented with multiple complexes in the same membrane, Get3 engages only a single Get1/2 heterodimer. We also show that the cytosolic domains of Get1 and Get2 bind simultaneously to opposites sides of the post-hydrolysis Get3 homodimer.

The bulk and single-molecule FRET studies described here were performed in the absence of TA protein, since we lack an experimental means to trap a post-hydrolysis Get3-TA targeting complex that does not release TA protein to Get1/2. Nevertheless, we qualitatively extend our conclusions regarding the architecture of the Get1/2/3 complex to the targeting complex because previous structural and biophysical analyses have shown that TA protein binding stabilizes ‘closed’ conformations resembling those sampled by ATP-and ADP-bound Get3 (Mateja et al., 2015; Mateja et al., 2009). Thus, we propose a simplified model for how docking, TA protein release and insertion, and Get3 recycling are coordinated by a Get1/2 heterodimer (Fig. 4c).

Following release from Get4/5, the Get3-TA complex arrives at the membrane in a ‘closed’ conformation. Because the Get1 binding site is only partially accessible in this conformation (Mateja et al., 2015; Mateja et al., 2009; Stefer et al., 2011), the targeting complex is captured first by Get2, bringing it into close proximity with Get1 (Mariappan et al., 2011; Rome et al., 2014; Stefer et al., 2011; Wang et al., 2011). Once ATP has been hydrolyzed, Get1 binds to the opposite side of ADP-bound or nucleotide-free Get3-TA, driving it from a partially destabilized closed conformation into an ‘open’ conformation that disrupts the hydrophobic TA protein binding site (Mariappan et al., 2011; Stefer et al., 2011). This large conformational change in Get3 is likely decoupled from the TMDs of the Get1/2 heterodimer by the long, flexible cytosolic domain of Get2. Following release, the TA protein is guided into the membrane by the TMDs of Get1 and Get2 (Wang et al., 2014). Finally, Get3 is recycled by ATP binding, which disrupts the Get1-Get3 interaction (Mariappan et al., 2011; Stefer et al., 2011); subsequent dissociation from Get2 may be facilitated by Get4/5 (Rome et al., 2014). Future studies are needed to obtain high-resolution structural information for each step along the pathway.

EXPERIMENTAL PROCEDURES

Protein Expression, Purification and Labeling

Full-length, 6xHis-tagged Get1, Get2, Get3, and Get2-1sc constructs were expressed in E. coli and purified via Ni-NTA affinity and size-exclusion chromatography. Single-cysteine containing variants were labeled with maleimide fluorophores Cy3 or Cy5 (GE Healthcare) and free dye was removed by size exclusion chromatography.

Proteoliposome Reconstitutions

Liposomes were prepared at 20 mg/mL via 100 nm extrusion in 50 mM Hepes, pH 7.5, 15% glycerol, 2 mM DTT. Get1/2 was reconstituted at 4° C by diluting protein to 90 μL in DBC buffer on ice. Then, 10 μL liposomes was added and incubated 15 minutes. DBC was removed by overnight incubation with biobeads.

Insertion Assay

Radiolabeled Get3-TA protein ‘targeting’ complexes were prepared by cell-free synthesis of TwinStrep-tagged Sec61β in the presence of 35S-methionine and 25 μM Get3, followed by Streptactin affinity purification. Proteoliposomes were normalized with empty liposomes, diluted to 10 nM Get1/2 and incubated with ~200 nM targeting complex for 30 minutes at 32° C. After 2 hr incubation on ice with 2 mg/mL proteinase K, the reaction was quenched with PMSF followed by hot SDS. Samples were separated by 12% Tris-Tricine SDS-PAGE and visualized by coomassie staining and autoradiography.

Fluorescence Measurements

Bulk fluorescence was recorded using a Synergy Neo plate reader with excitation filters at 540/25 and emission filters at 590/35 and 680/30. Single molecule measurements were made with a custom TIRF microscope employing 532 nm and 633 nm lasers; emissions were filtered at 585/40 and 692/40. Single molecule TIRF videos were recorded with a 200 ms exposure over 500 frames. Proteoliposomes for single-molecule photobleaching were adhered to freshly plasma-cleaned glass coverslips. Get1/2 nanodiscs with biotinylated MSP were attached to glass coverslips via glass-adhered neutravidin and casein blocking.

Get1/2 Nanodisc Reconstitutions

Purified MSP1E3D1 was biotinylated with NHS-PEG4-Biotin (Thermo). Mixed micelles were prepared by sonication of lipid and excess UM detergent until clear. Get2 and 6xHis-tagged Get1 were diluted into UM buffer on ice, followed by addition of mixed micelles and MSP. After 1 hr incubation, detergent was removed by overnight incubation with biobeads. Nanodiscs were then purified by Ni-NTA affinity-and size exclusion chromatography.

Statistical Methods

Errors are reported ± SEM. The bulk FRET experiments in Figure 2 and S2 combine data from three independent proteoliposome reconstitutions for each Get1/2 fluorophore combination. The single-molecule photobleaching and insertion data in Figure 3 are from the same reconstitution, one Get2-1sc-Cy5 proteoliposome reconstitution for each condition, and are representative of multiple experiments (>3). Figure S3 shows variance of the insertion assay by splitting each reconstitution into triplicate. The single-molecule FRET distributions in Figure 4 were fit to two Gaussian curves using the mle package in R.

Supplementary Material

supplement

Highlights.

  • -

    The minimum functional unit for TA protein insertion is a Get1/2 heterodimer

  • -

    Get3 binds a single Get1/2 complex, even when multiple complexes are present

  • -

    Get1 and Get2 bind simultaneously to opposite sides of post-hydrolysis Get3 dimers

Acknowledgments

We thank members of the Denic, Rock, and Keenan laboratories for technical assistance and helpful discussions, and we thank E. Perozo and F. Bezanilla for sharing equipment. This work was supported by NIH R01 GM086487 (to R.J.K.), GM078450 and GM109863 (to R.S.R.) and GM099943 (to V.D.); B.E.Z. was supported by an NIH Chemistry and Biology Interface training grant (T32 GM008720).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

AUTHOR CONTRIBUTIONS

B.E.Z., V.D., R.S.R, and R.J.K. designed research; B.E.Z. and C.C. performed research; B.E.Z. and R.J.K. wrote the initial draft; all authors edited the manuscript.

References

  1. Auld KL, Hitchcock AL, Doherty HK, Frietze S, Huang LS, Silver PA. The conserved ATPase Get3/Arr4 modulates the activity of membrane-associated proteins in Saccharomyces cerevisiae. Genetics. 2006;174:215–227. doi: 10.1534/genetics.106.058362. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bozkurt G, Stjepanovic G, Vilardi F, Amlacher S, Wild K, Bange G, Favaloro V, Rippe K, Hurt E, Dobberstein B, et al. Structural insights into tail-anchored protein binding and membrane insertion by Get3. Proc Natl Acad Sci USA. 2009;106:21131–21136. doi: 10.1073/pnas.0910223106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Denic V, Dotsch V, Sinning I. Endoplasmic reticulum targeting and insertion of tail-anchored membrane proteins by the GET pathway. Cold Spring Harb Perspect Biol. 2013;5:a013334. doi: 10.1101/cshperspect.a013334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Hegde RS, Keenan RJ. Tail-anchored membrane protein insertion into the endoplasmic reticulum. Nat Rev Mol Cell Biol. 2011;12:787–798. doi: 10.1038/nrm3226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Hu J, Li J, Qian X, Denic V, Sha B. The crystal structures of yeast Get3 suggest a mechanism for tail-anchored protein membrane insertion. PLoS ONE. 2009;4:e8061. doi: 10.1371/journal.pone.0008061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Jonikas MC, Collins SR, Denic V, Oh E, Quan EM, Schmid V, Weibezahn J, Schwappach B, Walter P, Weissman JS, et al. Comprehensive characterization of genes required for protein folding in the endoplasmic reticulum. Science. 2009;323:1693–1697. doi: 10.1126/science.1167983. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Kubota K, Yamagata A, Sato Y, Goto-Ito S, Fukai S. Get1 stabilizes an open dimer conformation of get3 ATPase by binding two distinct interfaces. J Mol Biol. 2012;422:366–375. doi: 10.1016/j.jmb.2012.05.045. [DOI] [PubMed] [Google Scholar]
  8. Kutay U, Hartmann E, Rapoport TA. A class of membrane proteins with a C-terminal anchor. Trends Cell Biol. 1993;3:72–75. doi: 10.1016/0962-8924(93)90066-a. [DOI] [PubMed] [Google Scholar]
  9. Mariappan M, Mateja A, Dobosz M, Bove E, Hegde RS, Keenan RJ. The mechanism of membrane-associated steps in tail-anchored protein insertion. Nature. 2011;477:61–66. doi: 10.1038/nature10362. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Mateja A, Paduch M, Chang HY, Szydlowska A, Kossiakoff AA, Hegde RS, Keenan RJ. Protein targeting. Structure of the Get3 targeting factor in complex with its membrane protein cargo. Science. 2015;347:1152–1155. doi: 10.1126/science.1261671. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Mateja A, Szlachcic A, Downing ME, Dobosz M, Mariappan M, Hegde RS, Keenan RJ. The structural basis of tail-anchored membrane protein recognition by Get3. Nature. 2009;461:361–366. doi: 10.1038/nature08319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Murphy MC, Rasnik I, Cheng W, Lohman TM, Ha T. Probing single-stranded DNA conformational flexibility using fluorescence spectroscopy. Biophys J. 2004;86:2530–2537. doi: 10.1016/S0006-3495(04)74308-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Rome ME, Chio US, Rao M, Gristick H, Shan SO. Differential gradients of interaction affinities drive efficient targeting and recycling in the GET pathway. Proc Natl Acad Sci U S A. 2014;111:E4929–4935. doi: 10.1073/pnas.1411284111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Schuldiner M, Metz J, Schmid V, Denic V, Rakwalska M, Schmitt HD, Schwappach B, Weissman JS. The GET complex mediates insertion of tail-anchored proteins into the ER membrane. Cell. 2008;134:634–645. doi: 10.1016/j.cell.2008.06.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Stefanovic S, Hegde RS. Identification of a targeting factor for posttranslational membrane protein insertion into the ER. Cell. 2007;128:1147–1159. doi: 10.1016/j.cell.2007.01.036. [DOI] [PubMed] [Google Scholar]
  16. Stefer S, Reitz S, Wang F, Wild K, Pang YY, Schwarz D, Bomke J, Hein C, Lohr F, Bernhard F, et al. Structural basis for tail-anchored membrane protein biogenesis by the Get3-receptor complex. Science. 2011;333:758–762. doi: 10.1126/science.1207125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Suloway CJ, Chartron JW, Zaslaver M, Clemons WM., Jr Model for eukaryotic tail-anchored protein binding based on the structure of Get3. Proc Natl Acad Sci U S A. 2009;106:14849–14854. doi: 10.1073/pnas.0907522106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Wang F, Chan C, Weir NR, Denic V. The Get1/2 transmembrane complex is an endoplasmic-reticulum membrane protein insertase. Nature. 2014;512:441–444. doi: 10.1038/nature13471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Wang F, Whynot A, Tung M, Denic V. The mechanism of tail-anchored protein insertion into the ER membrane. Mol Cell. 2011;43:738–750. doi: 10.1016/j.molcel.2011.07.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Yamagata A, Mimura H, Sato Y, Yamashita M, Yoshikawa A, Fukai S. Structural insight into the membrane insertion of tail-anchored proteins by Get3. Genes Cells. 2010;15:29–41. doi: 10.1111/j.1365-2443.2009.01362.x. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement
NIHMS900719-supplement.pdf (10.6MB, pdf)

RESOURCES