Results

Crystallization of Get3.

We purified ScGet3 and AfGet3 from constructs expressed in E. coli using Ni-affinity and size exclusion chromatography under reducing conditions. The majority of the protein eluted as a dimer from both constructs and this was used in crystallization trials. The AfGet3-ADP crystals diffracted to 3.2-Å resolution in the space group P2₁2₁2₁ with a hexamer in the asymmetric unit. A seleno-methionine data set was collected to 4.5-Å resolution and phases were solved by multiwavelength anomalous dispersion and extended to 3.2-Å resolution using the 6-fold noncrystallographic symmetry. The final refined structure contained 292 of 348 residues with a Free-R factor of 25.1% (Fig. 1C and Fig. S1). The two Apo forms of Get3 were solved using molecular replacement of a truncated AfGet3-ADP monomer. The ScGet3-apo crystal form diffracted to 3.7-Å resolution in the space group H32 and contained a monomer in the asymmetric unit. The final model contained 260 of 369 residues and refined to a Free-R factor of 33.5% (Fig. S2 A and B). The AfGet3-apo crystal form diffracted to 7.5-Å resolution in the space group P4₂32 and contained a dimer in the asymmetric unit that we did not refine due to the low resolution (Fig. S2D). Crystallographic statistics are provided in Table S1.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Structures of Get3. (A) An AfGet3-ADP monomer with secondary structure elements numbered as in Fig. S1. (B) The NHD dimer of the AfGet3-ADP hexamer. One monomer is color ramped from N- (blue) to C- (red) and the other is colored relative to motifs described in the text: P-loop (green), Switch I (magenta), Switch II (blue), A-loop (red), SB1 (purple), and SB2 (brown). (C) The AfGet3-ADP hexamer of the asymmetric unit colored by monomers. Dimer interfaces and subunits are labeled. The 3-fold is indicated by a triangle and 2-folds are indicated by an oval. The modeled residues from SB1 of each monomer are transparent. Nucleotides and bridging disulfides are shown as spheres. (D) The arm dimer of the AfGet3-ADP hexamer. One monomer is color ramped as in A and the other is colored salmon. (E) The ScGet3 model colored purple and overlayed with an AfGet3-ADP monomer colored and oriented as in B.

Description of a Get3 Monomer.

The structure of a monomer of Get3 is a mixed alpha-beta fold containing a “P-loop” type NHD with two α-helical loops that extend outward from the structure, here designated substrate binding loop 1 (SB1) and 2 (SB2) (Fig. 1 A and B). The Get3 NHD fold falls into a more specific structural class [defined by SCOP (24)] that includes ArsA (25), GTPase domains of the signal recognition particle (SRP) (26), and SRP receptor (27) along with NifH (28). Get3 is the only eukaryotic example in this class that utilizes ATP.

The Three Crystal Forms of Get3.

The Af-ADP crystal form contains a hexamer in the asymmetric unit (Fig. 1C) with 3-fold symmetry in which the monomer can be assembled into two potential dimers formed by either SB1/2 (arm dimer) or by the interface between the NHD (Fig. 1 B and D). Although the arm dimer contains a more extensive interface, 1758Ų versus 1263Ų calculated by PISA (29), we believe that the NHD dimer is the most relevant to TA-protein binding and contains two disulfides formed across the interface by a conserved pair of cysteines (Fig. 1B).

The Sc-apo crystal form contains a monomer in the asymmetric unit (Fig. 1E and Fig. S2A), and there is no apparent NHD dimer interface despite the protein purifying as a dimer. The crystals are stabilized by the coordination of a zinc by Cys-285/Cys-288 and the His-tag which also coordinates a second metal at the crystallographic three fold in a square-planar geometry (Fig. S2A). The AfGet3-apo contains two copies in the asymmetric unit in an orientation similar to the NHD dimer, although rotated so that the SB1 and SB2 regions of the two monomers are slightly closer (Fig. S2E). Both Apo crystal forms contain symmetry related interfaces similar to the arm dimer (Figs. S2C and S3F).

Monomers in each of the three crystal forms show slight variations in SB1/2 demonstrating the general mobility of these regions (Fig. 1E and Fig. S2F). These loops in the hexamer are stabilized by a series of hydrophobic interactions in a highly flexible region that perhaps explains the difficulty in obtaining high-resolution crystals. SB1 and SB2 contain disordered regions in all of the crystal forms. The missing residues of SB1 have been modeled into the AfGet3-ADP hexamer to demonstrate the amount of disordered protein that could not be built (Fig. 1C).

Nucleotide Binding.

The Get3 nucleotide-binding pocket contains all of the features generally found in G-type hydrolases. The completely conserved Asn in S7 (Sc/Af272) forms hydrogen-bonds that specifically select for adenine. Additional interactions with the A-loop complete adenosine recognition (Fig. 2A). The P-loop, as is typical, makes extensive contacts to the α- and β-phosphate; however, the second lysine, completely conserved in P-loops, is in an orientation that points away from the β-phosphate. This is caused by an interaction in the arm dimer that leads to an Arg from SB2 (Af200) moving into the active site occupying a similar position near where one would expect Mg²⁺ to be bound (Fig. 2A). It is clear that AfR200 displaces the Mg²⁺ and generally disrupts the interactions of Switch I and II. Based on the resolution, we cannot be certain that there is no Mg²⁺; however, if present it would be in a unique position. AfR200 forms a salt bridge to the β-phosphate; but it is not conserved making the extent of these interactions surprising.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

The nucleotide-binding pocket and comparison of Get3 to other hydrolases. (A) The nucleotide-binding pocket of AfGet3-ADP with residues shown as sticks. Density is a 2Fo-Fc omit-map contoured at 1.5σ. (B) A ribbons diagram of the ADP form of EcArsA (1f48) with ADP, Mg²⁺ (green) and coordinated Sb (purple) as spheres. (C) A ribbon diagram of the apo form of NifH (2nip) with the Fe/S cluster (orange/yellow) as spheres. To the right in B and C are overlays of monomers the AfGet3-ADP monomer (gray) on the respective left subunit. Important residues and motifs are labeled. All residues in nucleotide binding motifs are colored as in 1B.

Comparison to ArsA and NifH.

Despite distinct functions, Get3 shares a similar topology to ArsA with an RMSD of 1.9Å in their NHD (Fig. 2B) (PDBID 1f48) (25). In contrast to Get3, ArsA SB1/2 bend in across the NHD dimer interface forming a coordination site for heavy metals (Fig. 2A); however, these coordinating residues are not conserved in Get3. It is thought that motions of these loops are coupled to ATP hydrolysis regulating metal release via the Switch II motif (22). The dimer interface is very similar to Get3 except that the interface is rotated moving the P-loop from 9.1-Å (G17/G336) separation in ArsA to 14.1 Å (AfG35) in Get3. An early homology model of Get3, based on ArsA, predicted the occurrence of the disulfide bridges between the subunits at the dimer interface. Based on the model they found that mutation of the two cysteines in Get3 was unable to rescue a metal sensitivity phenotype in a Get3 knockout (30). ArsA is a pseudodimer with a disordered linker peptide between the two subunits that may be required to stabilize the dimer interface.

The best understood deviant P-loop protein is NifH as its structure has been solved in Apo, ADP, and ADP·AlF₄⁻ forms. The structure closest to AfGet3-ADP is the NifH-Apo form (19) and the NHD domains have an RMSD of 2.78 Å. As noted, ATP stimulates a large conformational shift that moves the deviant P-loop (A. vinelandii NifH G11) from 10.1 Å to 4.0 Å apart. To move the Get3 dimer into a similar orientation would require an extensive conformational change across the dimer interface.

Tail-Anchor Binding Pocket.

In a search for the TA-protein binding pocket, the positions of SB1 and SB2 are clearly provocative. We analyzed the NHD dimer by displaying conserved and hydrophobic residues on a molecular accessibility surface (Fig. 3). The interface, formed by the NHD dimers, is highly conserved, as is expected for a common fold (Fig. 3A). The other concentration of conserved residues is found at the base and groove formed by SB1/SB2 (Fig. 3C). The overall alignment of SB1/SB2 is difficult; however, there is general conservation of hydrophobic residues and glycines (Fig. S1). Additionally, SB1 contains a disordered stretch that could also provide surface area in this region. The only hydrophobic patch on the surface of the dimer is that formed between SB1 and SB2 (Fig. 3 B and D). This putative TA-binding region would be analogous to the location of the metal binding sites of ArsA or the Fe-S cluster in NifH (Fig. 2 B and C) and one presumes that changes in the ATP binding pocket would be transmitted to this region.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Conserved and hydrophobic surfaces. (A) Front and back views showing one monomer as a ribbon colored as in 1A and the other as an accessible surface showing conservation colored as a gradient from 100% (purple) to 50% conserved (gray). Conservation is based on the Get3 alignment from Fig. S1 (B) Similar to A showing hydrophobicity based on the Kyte and Doolittle scale with most hydrophobic in dark yellow. (C) Conserved surface viewed from the top. (D) Hydrophobic surface viewed from the top.

Phenotypic Rescue.

To probe the functional parts of Get3 we chose a series of mutants based on surface conservation or putative function and tested for their ability to rescue known knockout phenotypes (30). The knockout (Δget3) showed no obvious impairment on synthetic complete media at 30C but was unable to fully rescue on media containing Cu²⁺ or hydroxyurea or growth at elevated temperature. Replacing the Get3 gene on a plasmid (GET3) with the wild-type promoter rescued the Δget3 growth to near wild-type levels. We also inserted the AfGet3 gene on the same plasmid and this also rescued the yeast knockout demonstrating that functional aspects of the protein are conserved across species (Fig. 4A and Fig. S3). In all, we generated 69 Sc mutants and two Af mutants and scored their general loss-of-function (LOF) phenotype as strong, moderate, weak or none (Fig. 4 A and B, data for all mutants in Fig. S3 and Table S2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Phenotypic rescue of various Get3 mutants. (A) Spot plate assays of various yeast mutants on a plasmid with a wild-type promoter screened on SC-Ura plates at 30C and 40C and supplemented with 2 mM CuSO₄ at both 30C and 37C. Δget3 is the knockout transformed with a plasmid containing only the promoter. Mutants based on yeast numbering are indicated colored relative to their phenotype: strong (red), moderate (orange), weak (yellow), and none (cyan). (B) Two views with one monomer as accessible surface and the other in ribbon colored according to phenotype as in A.

In agreement with previous results, a mutation in either the P-loop (ScG30R or AfG38R) or the pair of Cys that form the disulfide bridges (C285T/C288T) had strong LOF phenotypes (Fig. 4A) (3, 30). The G30R mutation is thought to disrupt ATP binding. The effect of the Cys mutants is less clear. Presumably, this interface is somewhat unstable and requires the disulfides to stabilize the dimer, similar to the linked dimer of ArsA. As the cytoplasm is a reducing environment, it would be curious that the disulfides could form in vivo; however, we included reducing agent in all of our buffers and the disulfides formed in that context. Another possibility is that these residues coordinate a metal or are regulated by a redox pathway (30) as the reduced form in the ScGet3-apo crystals is a monomer and coordinates a zinc (Fig. S2 A and B).

The largest cluster of LOF mutants occurs at the NHD dimer interface found mostly on H8 and H9. The interface is a mixture of hydrophobic and charged groups that would be intimately involved in a re-arrangement of the dimer (Fig. S4B, Figs. S1 and S4A). Only a few of the conserved surface mutations that did not make contacts in this crystal form conferred LOF phenotypes (R75A, D265A, and Y338A) and probably do not affect the conformational changes in substrate binding (Fig. S4B). It is possible that these surface residues play a role in recognition of other proteins in the Get pathway.

Changes in switch helices are normally coupled to functional changes. Although there was little conformational change in ArsA nucleotide structures it was postulated that binding of ATP would cause a conformational change in Switch II that would be transmitted to a His involved in metal coordination (22). This residue (ScH172) is the only coordinating residue from ArsA that is found in Get3 and in our structure this is in a position to interact with a network of salt bridges that appear to stabilize the base of SB1/2 (Fig. S4C). Mutations of these residues had LOF phenotypes; however, they were not strong and it may be possible that coupling of Switch II changes is not essential for TA binding.

Get3 binds a variety of TA-protein substrates via hydrophobic interactions (4) and it is difficult to decide what mutations might interfere with binding. Based on the predicted pocket, we generated extensive mutations in SB1 and SB2. As expected, mutations in the predicted TA-protein binding pocket (I136S, D137A, L140S, S141A, M143S, and L219S) had LOF; however, the majority of the residues had no phenotype including those disordered in our structure (Fig. 4B and Fig. S4D). This hydrophobic interaction may require multiple mutations in the binding pocket to see significant disruptions.

The lack of conservation of the SB2 residue AfR200 made a comparable mutation in Sc impossible. Due to its location in the nucleotide-binding pocket in the AfGet3-ADP structure we decided to see if its mutation would have an effect on rescue by AfGet3. An AfR200A mutation was a clear LOF phenotype (Figs. 2A and 4A). This is in contrast to a number of other mutations, including some disordered residues, in this region of SB2 that showed no phenotypes. It is difficult to envision an effect of the AfR200A mutation in the absence of the hexamer.

A mutation of the conserved deviant P-loop lysine is expected to completely abolish function and should be a strong phenotype. In the AfGet3-ADP structure this residue makes no contacts (Fig. 4 and Fig. S4A), still, a mutation of this residue to Ala (ScK26A) was the strongest phenotype of all. In the Ras/RasGAP case any mutation of the Arg-finger leads to a total LOF even the seemingly benign mutation to Lys (31). We did the same type of mutation, ScK26R, and found that this mutant is a strong LOF phenotype, although not as strong as ScK26A.