First crystallographic models of human TBC domains in the context of a family‐wide structural analysis

RAB-family1 guanine nucleotide-binding proteins mediate membrane-associated protein transport2 via a mechanism tightly linked to their GTPase activity. RABs cycle between “inactive” GDP-bound and “active” GTP-bound states. “Activation,” or replacement of GDP by GTP, is assisted by guanine nucleotide exchange factors,3 and “inactivation” is effected by GTP hydrolysis. RABs exhibit poor inherent GTPase activity, and rely on RAB-specific GTPase-activating proteins (RAB-GAPs) for GTP hydrolysis.4 RAB-GAPs5 share sequence motifs characteristic of Tre-2/Bub2/Cdc16 (TBC) domains.6 In fact, with few exceptions,7 the terms RAB-GAP and TBC are now used interchangeably. The crystal structure of the GAP domain of Saccharomyces cerevisiae Gyp1p9 provided the first structural insight into RAB-GAP activity and led to the identification of the catalytically relevant R343 and Q378 residues. The central role of these residues was later confirmed by the crystal structure of the complex of the Gyp1p GAP domain with murine Rab33-GDP and AlF3.10 To date, no mammalian RAB-GAP structure has been reported; and Gyp1p has remained the only RAB-GAP with an available atomic model. The affinities of 40 TBC family members for different RABs have recently been examined in a large-scale study.11 Complementing this approach, we have targeted human TBC domain-containing proteins for structure determination by protein crystallography, and report the crystal structures of the first two mammalian TBCs, human TBC1 family members 22A (TBC1D22A) and 14 (TBC1D14). Gene constructs encoding amino acid residues 191–517 and 357–672 of TBC1D22A and TBC1D14, respectively, were amplified by PCR from Mammalian Gene Collection12 clones (Geneservice coordinates AT17-G1 and AT71-E11, respectively) and sub-cloned into the BseRI sites of a modified pET vector pET28-mhl (gi:134105571) using the In-Fusion PCR cloning system (Clontech). The constructs were transformed into Escherichia coli strain BL21-CodonPlus(DE3)-RIL (Stratagene) and the cells were grown at 37°C in Terrific Broth to an OD600 of ∼3.0 using the LEX bubbling system.13 The temperature of the water bath was lowered to 18°C for half-an-hour before induction by the addition of isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 1 mM. The cells continued to grow at 18°C overnight and were then harvested by centrifugation, flash frozen in liquid nitrogen, and stored at −80°C. For selenomethionine labeling of TBC1D14, the M9 SeMET growth media kit (Medicilon) was used. The protocol was slightly modified from manufacturer's instructions: batches of 50 mL LB medium were inoculated with single colonies from a freshly transformed LB/Agar plate with appropriate antibiotics and grown at 37°C overnight. Cells were separated from the overnight culture by centrifugation, re-suspended in 1.8 L M9 media and grown in the LEX system until OD600 reached 1.5. When the temperature was lowered to 18°C, protein overexpression was induced by the addition of IPTG to a final concentration of 1 mM and growth was continued overnight. The cell pellet was stored at −80°C. It was thawed and re-suspended in 50 mL of a binding buffer (20 mM HEPES pH 7.5, 300 mM NaCl, 1 mM β-mercaptoethanol, BME) supplemented with protease inhibitor cocktail (0.1 mM benzamidine-HCl and 0.1 mM phenylmethyl sulfonyl fluoride final concentrations), and 0.5% CHAPS. The cells were disrupted by a Microfluidizer fluid processor (Microfluidics Corporation). The lysate was clarified by centrifugation at 27,000g for 60 min and the supernatant was loaded onto a 5-mL HiTrap Chelating HP column (GE Healthcare) loaded with Ni2+, equilibrated with the binding buffer at 4°C. The column was washed with 25 mL washing buffer (20 mM HEPES pH 7.5, 300 mM NaCl, 30 mM imidazole, 1 mM BME) and the proteins were eluted with a linear gradient of imidazole from 30 to 500 mM in the HEPES binding buffer within 10 column volumes. The fractions containing the desired proteins were pooled and loaded onto a Superdex 75 column (16/60) equilibrated with a low salt buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 5 mM dithiothreitol). Fractions containing the proteins were collected and concentrated using Amicon Ultra-15 centrifugal filters to a final concentration of 20 mg/mL. The purity of the protein was checked by SDS-PAGE to be >95% and the molecular weight was measured by MALDI-TOF mass spectroscopy (Agilent). Purified proteins were crystallized at 18°C using the sitting drop vapor diffusion method by mixing 0.5 μL of the protein solution with 0.5 μL of the reservoir solution (100 μL, see Table I). Single crystals were harvested from the mother liquor and dragged through a 1:1 mixture of mineral and paratone (Hampton Research) oils to remove excess mother liquor from the crystal surface. Crystals were frozen in cryo-loops17 (Hampton Research) by rapid immersion into liquid nitrogen and stored in liquid nitrogen until the time of the diffraction experiment. Diffraction data were collected in continuous ranges of oscillation images as described in Table I. During the experiments, crystal cooling was maintained by a stream of cold nitrogen gas at ∼100K. Diffraction data were processed with the HKL software suite.19 Molecular replacement for TBC1D22A was performed with the program PHASER20 using polymer chain A from PDB entry 2G7710 as a search model. Single wavelength anomalous diffraction21 substructure solution and initial phasing for a selenomethionyl derivative of TBC1D14 were carried out with the programs SHELXD22 and SHELXE,23 respectively. Automated model rebuilding in ARP/wARP24 was followed by several iterations of restrained coordinate and temperature factor refinement in REFMAC,25 geometry validation on the MOLPROBITY26 server and manual model adjustments in COOT.27 SHELXD/E, ARP/wARP, and REFMAC were run through the CCP4I interface28 of the CCP4 suite.29 PDB_EXTRACT30 was used for data harvesting prior to Protein Data Bank31 deposition. Alignment of atomic coordinate pairs was accomplished using “secondary structure matching”32 as implemented in COOT, unless stated otherwise. Molecular figures were prepared with PYMOL.33 Diffraction data for a crystal of recombinant TBC1D22A were collected and scaled in space group P21 including reflections to 2.1 Å resolution (Table I). Molecular replacement search using Gyp1p coordinates, with which TBC1D22A shares 48% amino acid sequence identity,34 located one homodimer of the protein in the asymmetric unit. Upon refinement, the Cα traces of the two TBC1D22A monomers differed by an RMSD of 0.4 Å. The refined model of TBC1D22A adopts an all-helical fold [Fig. 1(a)]. Given very few gaps in the sequence alignment (see Fig. 2) and nearly complete conservation of secondary structure features, TBC1D22A can be superimposed to unbound Gyp1p (PDB entry 1FKM)9 and the Gyp1p·Rab33 complex (2G77) with RMSDs of 1.2 Å and 1.1 Å, respectively [Fig. 1(a)]. Upon Cα superposition of TBC1D22A on Gyp1p, key catalytic residues of TBC1D22A are closely aligned with their Gyp1p counterparts. In TBC1D22A, R286, and Q320 are located at the boundary between solvent exposed areas and the buried noncrystallographic dimer interface. Electron density for the side chains of R286 and Q320 is weak, indicating the presence of multiple rotamers of these residues. (a) Stereographic cartoon model of TBC1D22A, based on chain A of PDB entry 2QFZ. Rainbow coloring is applied according to position in the peptide sequence, from blue amino terminus to red carboxy terminus. As a reference, an aligned model of the Gyp1p·Rab33 complex, based on PDB entry 2G7710 (outline of Gyp1p in magenta, cartoon representation of Rab33 in gray), is also shown. Helix numbering corresponds to the one reported for the Gyp1p·Rab33 complex.10 (b) Stereographic cartoon model of TBC1D14, based on PDB entry 2QQ8. Colors and outline of Gyp1p·Rab33 complex are as described for Figure 1(a). (c) Expanded view of the putative active site of TBC1D14 based on Figure 1(b). 2Fo-Fc model-phased electron density is contoured at 1.5 σ around the side chains of residues R457 and Q493. Coordinates for magnesium, aluminum fluoride, and GDP as well as side chains of R343 and Q378 from PDB entry 2G7710 are shown for reference. (d) Alternative mode of alignment of active site residues in Gyp1p and TBC1D14. LSQKAB35 was used to simultaneously align catalytic arginyl, glutamyl, and immediately adjacent residues only. This mode of superimposition causes significant distortion of the overall alignment of the TBC domains. CLUSTALW36 sequence alignment of Gyp1p, TBC1D22A, and TBC1D14. Helices are highlighted by colored background, approximately matching colors used in Figure 1(a,b). Helix boundaries were determined by the program DSSP37 for PDB entries 2G77, 2QFZ, and 2QQ8 on the DALI server.38 Structurally significant sequence insertions i and ii are shown in bold italic. Amino acid residues that are missing from the models are represented by lower case letters. For a crystal of recombinant TBC1D14, diffraction data were collected and scaled to 2.0 Å resolution in space group P41212/P43212. However, in contrast to TBC1D22A, all attempts at molecular replacement failed, partly due to the low sequence similarity of TBC1D14 to available search models. For example, TBC1D14 and TBC1D22A have an amino acid sequence identity of only 21% over 284 residues of the crystallized TBC1D14 construct. For experimental phasing, a selenomethionyl derivative of TBC1D14 was prepared and data were collected near the selenium absorption edge (Table I). Single wavelength anomalous diffraction combined with density modification provided an interpretable electron density map in space group P43212. Model building and refinement produced a model that shares its overall topology with both Gyp1p and TBCD1D22A. A structural superposition of TBC1D14 on the Gyp1p·Rab33 complex with an RMSD of 2.7 Å is shown in Figure 1(b). The all-helical fold is clearly conserved and equivalent secondary structure elements are easily identified. Consistent with large sequence alignment gaps, however, there are notable differences between Gyp1p and TBC1D22A. Helix h4, corresponding to αG4 in the Gyp1p·Rab33 complex,10 is shorter than that observed in both Gyp1p and TBC1D22A by two turns. Similarly, the adjacent connection between helices h7 and h8 is formed by only two residues in TBC1D14, whereas much longer loops connect the equivalent helices in Gyp1p and TBC1D22A (see also the sequence alignment in Fig. 2). In TBC1D22A, this loop is disordered. In Gyp1p, it acts as a lid on top of a helix bundle formed by helices αG5 through αG8. The presence of TBC-specific sequence motifs supports a putative role of TBC1D14 in RAB GTPase activation. In the crystal structure, the side chains of putative catalytic residues R457 and Q493 are well ordered and represented by unambiguous electron density [Fig. 1(c)]. Structural superposition of TBC1D14 and Gyp1p shows that R457 is displaced by 2.5 Å from the Cα position of R343 in Gyp1p, a significant deviation considering an estimated overall coordinate error of ∼0.2 Å. Two possible mechanisms could cause R457 and Q493 to assume spatial arrangements similar to those seen for the equivalent residues in Gyp1p and TBC1D22A. Firstly, Rab binding might cause the catalytic residues of TBC1D14 to adopt the conformation required for stabilization of the transition state during GTP hydrolysis, although this is not supported by precedence in Gyp1p, where the RAB-bound and free forms are similar to each other with an overall RMSD of 0.7 Å, and Cα shifts of ∼0.4 Å for catalytic residues R343 and Q378. Secondly, instead of superimposing the overall TBC domains, one could simultaneously align only R457 and Q493, including immediately adjacent residues, to Gyp1p's R343 and Q378, respectively [Fig. 1(d)], while treating the entire TBC domain as a rigid body. Following such alignment, RAB-GAP activity appears feasible for TBC1D14, even though the overall alignment of the TBC domains would be severely distorted (not shown). Considering these two possible mechanisms, we find that the structural evidence strengthens the case that TBC1D14 is a RAB-GAP. Crystal structures TBC1D22A and TBC1D14 provide the first atomic models of human TBC domain family members. With a sequence identity of less than 20% and a coordinate RMSD of 2.6 Å between them, these structures reflect the diversity of TBC domains in the human genome. Both structures are in agreement with their putative role in RAB GTPase activation. However, to achieve an arrangement of the catalytic arginyl and glutamyl residues that is consistent with the current model of RAB GTPase activation,10 a significant overall re-alignment is required in the case of TBC1D14. Additional experiments are required to confirm the role of TBCD14A in RAB activation. The authors thank Mr. Aiping Dong for the collection of TBC1D22A diffraction data. Dr. Marianne Cuff provided expert assistance during data collection at the Structural Biology Center.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.