Molecular Basis of the Acceleration of the GDP-GTP Exchange of Human Ras Homolog Enriched in Brain by Human Translationally Controlled Tumor Protein*

Ras homolog enriched in brain (Rheb), a small GTPase, positively regulates the mTORC1 pathway. The GDP-GTP exchange of Rheb has been suggested to be facilitated by translationally controlled tumor protein (TCTP). Here we demonstrate that human TCTP (hTCTP) interacts with human Rheb (hRheb) and accelerates its GDP release in vitro and that hTCTP activates the mTORC1 pathway in vivo. To investigate the underlying mechanism, we built structure models of GDP- and GTP-bound hRheb in complexes with hTCTP and performed molecular dynamics simulations of the models, which predict key residues involved in the interactions and region of hRheb undergoing conformational change during the GDP-GTP exchange. These results are verified with site-directed mutagenesis and in vitro biochemical and in vivo cell biological analyses. Furthermore, a crystal structure of the E12V mutant hTCTP, which lacks the guanine nucleotide exchange factor activity, shows that the deficiency appears to be caused by loss of a salt-bridging interaction with Lys-45 of hRheb. These data collectively provide insights into the molecular mechanisms of how hTCTP interacts with hRheb and activates the mTORC1 pathway.

The long searched for GEF for Rheb was proposed to be translationally controlled tumor protein (TCTP) based on genetic data (16), but this has become controversial (11,17), providing a strong motivation for our studies. As implicated in its name, TCTP is regulated at both translational and posttranslational levels in response to a wide range of extracellular signals and conditions (18) and exerts diverse functions in various cell processes such as cell cycle progression, cell growth, microtubule stabilization, and apoptosis (reviewed in Ref. 19). In Arabidopsis thaliana, TCTP has been shown to be an important regulator of growth, but whether and/or how the TOR activity is affected by plant TCTP was not studied (20). TCTP homozygous knock-out in mouse is embryonic-lethal, and the mutant embryos are smaller than the wild-type embryos and have reduced number of cells in epiblast (21). In Drosophila, down-regulation of TCTP results in phenotypes, including reduced cell size, cell number, and organ size (16), similar to those of Drosophila Rheb (dRheb) mutant (22). Genetic epistasis experiments suggested that Drosophila TCTP (dTCTP) is epistatic to dRheb but acts upstream of S6K (16). Although TCTP was discovered about two decades ago, its GEF-like activity was not realized until recently. TCTP structurally resembles a GEF, namely Mss4 (23). Indeed, in vitro experiments showed that dTCTP binds to nucleotide-free dRheb and specifically stimulates the GDP-GTP exchange of dRheb (16). Moreover, co-immunoprecipitate experiments showed that dTCTP interacts with dRheb, and reducing the TCTP level in Drosophila S2 cells results in lowered GTP-bound dRheb, further suggesting that Rheb is a downstream target of TCTP in vivo (16). However, the GEF activity of TCTP on Rheb and the role of TCTP in S6K regulation have been doubted in two recent papers. Rehmann et al. showed that the GEF activity of human TCTP (hTCTP) on human Rheb (hRheb) could not be detected using an in vitro real-time fluorescence-based in-solution assay (17). Wang et al. reported that overexpression of hTCTP had no significant effect on the phosphorylation level of S6, a substrate of S6K, in stressed HEK293 cells (11). In addition, TCTP knockdown by siRNA did not affect the phosphorylation level of S6K (17) or S6 (11).
In this work, we demonstrated with in vitro biochemical experiments that hTCTP can accelerate the GDP release of hRheb, and with in vivo cell biological experiments that overexpression of hTCTP can result in enhanced phosphorylation of S6K, whereas knockdown of hTCTP has an opposite effect, supporting the notion that hTCTP is a GEF of hRheb and a regulator of the mTORC1 pathway. To investigate the molecular mechanism of the GEF activity of hTCTP on hRheb, we carried out homology modeling studies of the hRheb⅐hTCTP complexes based on the crystal structures of GDP-and GTPbound hRheb (24) and hTCTP using the crystal structure of the Rab8⅐Mss4 complex as a template (25), and further employed molecular dynamics (MD) simulation to the modeled complexes. Our modeling and simulation results predicted the key residues that are involved in the hTCTP-hRheb interaction and the important region that undergoes conformational change during the GDP-GTP exchange of hRheb and reached a different conclusion than that by Rehmann et al. (17). These results were validated by using mutagenesis studies, in vitro biochem-ical and in vivo cell biological analyses. We also determined the crystal structure of the E12V mutant hTCTP, which lacks the GEF activity and cannot activate the mTORC1 pathway. The computational, structural, biochemical, and cell biological data together provide insights into the molecular mechanism of the GDP-GTP exchange of hRheb accelerated by hTCTP.

EXPERIMENTAL PROCEDURES
Cloning, Expression, and Purification-The hTCTP gene was amplified by PCR from the cDNA of human HEK293T cells with the sense primer 5Ј-ATGATTATCTACCGGGACCTC-3Ј and the antisense primer 5Ј-ACATTTTTCCATTTCTAAAC-3Ј, and this gene fragment was further amplified with the sense primer 5Ј-AACATATGATTATCTACCGGGACCTC-3Ј (with the NdeI restriction site underlined) and the antisense primer 5Ј-AACTCGAGACATTTTTCCATTTCTAAAC-3Ј (with the XhoI restriction site underlined) to incorporate the restriction sites. The gene fragment was inserted into the NdeI and XhoI restriction sites of the pET-22b(ϩ)-His expression vector, and the plasmid was used as a template to generate an E12V mutant hTCTP with QuikChange site-directed mutagenesis kit (Stratagene). The plasmid encoding the mutant was transformed into Escherichia coli BL21(DE3) strain (Novagen). When the culture of the transformed cells reached an A 600 of 0.6ϳ0.8, protein expression was induced by 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside at 20°C for 18 h. The cells were lysed by sonication in the lysis buffer (20 mM Tris-HCl, pH 7.4, 500 mM NaCl, 10% glycerol, and 2 mM dithiothreitol). Protein purification was carried out by affinity chromatography using a nickel-nitrilotriacetic acid column (Qiagen) with the lysis buffer supplemented with 20 mM and 200 mM imidazole serving as washing buffer and elution buffer, respectively. The elution fractions containing the target protein were further purified by gel filtration using a Superdex G-75 Hiload 16/60 column (Amersham Biosciences) in buffer A (10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 10% glycerol, and 2 mM dithiothreitol). The target protein fractions were pooled and concentrated to ϳ30 mg/ml for crystallization. The hTCTP gene fragment was also amplified with the sense primer 5Ј-AGGATCCATGATTATC-TACCGGGAC-3Ј (with the BamHI restriction site underlined) and the antisense primer 5Ј-ACTCGAGTTAACATTTTTCC-ATTTC-3Ј (with the XhoI restriction site underlined) and cloned into the BamHI and XhoI restriction sites of the pEGX-4T1 vector, and the site-directed mutagenesis was performed to construct various GST-hTCTP mutants. The GST-tagged wild-type and mutant hTCTP proteins were expressed in E. coli BL21(DE3) cells as described above. The cells were lysed by sonication in phosphate-buffered saline buffer, and the protein was purified with glutathione-Sepharose beads (Amersham Biosciences). All of the purification processes were carried out at 4°C (except for gel filtration at 16°C) to reduce potential proteolysis and denaturation of the target protein. The quality of the purified protein was assessed by SDS-PAGE and Coomassie Blue staining. His-tagged hRheb was expressed and purified as described previously (24).
In Vitro GST Pulldown Assay-To investigate the effects of mutations of the key residues involved in the hRheb-hTCTP interaction, we performed in vitro protein-protein binding assay. Previous studies have shown that the purified hRheb was bound with GDP and Mg 2ϩ (24). To remove the bound GDP and Mg 2ϩ , 0.5 ml of the purified protein was incubated with 10-fold volume of buffer A supplemented with 10 mM EDTA and subsequently ultracentrifuged to 0.5 ml. Such process was repeated five times. GTP loading of hRheb was achieved by incubating the nucleotide-free hRheb with 25 mM GTP and 10 mM MgCl 2 . For experiments presented in Fig. 1A, ϳ150 g of the wild-type GST-hTCTP protein was immobilized onto the glutathione-Sepharose beads and then incubated with 300 g of the nucleotide-free or nucleotide-bound hRheb protein in binding buffer (10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 10% glycerol, and 10 mM EDTA for nucleotide-free hRheb or 10 mM MgCl 2 for nucleotide-bound hRheb, respectively) at 4°C for 2 h. The mixture was then washed ten times with chilled washing buffer (10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 10% glycerol, and 10 mM EDTA for nucleotide-free hRheb or 10 mM MgCl 2 for nucleotide-bound hRheb). The bound proteins were finally eluted by elution buffer (10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 10% glycerol, and 100 mM GSH). For experiments presented in Fig. 4 (A and B), the general procedure was similar except that the mutant proteins were used and the mixture was washed five times with chilled washing buffer supplemented with 0.5% Tween 20. The eluted proteins were resolved by 12% SDS-PAGE, transferred to polyvinylidene difluoride, and then blotted with anti-His antibodies (Tiangen). The GST protein served as a negative control.
In Vitro GDP Release Assay-In vitro GDP release assay was performed as described by Hsu et al. with minor modifications (16). Briefly, 1 M nucleotide-free His-hRheb was incubated with 1 M [ 3 H]GDP and 1.5 M wild-type or mutant GST-hTCTP in buffer B (50 mM HEPES, pH 7.6, 100 mM NaCl, 2.5 mM Mg 2ϩ , and 1 mM dithiothreitol) at 25°C for 60 min. The GDP/GTP exchange reaction of hRheb was initiated by addition of excessive GTP (ϳ100 M). At different time points 5 l of the mixture was loaded to nitrocellulose membrane. After the membrane was dried, it was washed four times with 1 ml of ice-cold buffer B. The amount of the radiolabeled GDP bound to hRheb was quantified by scintillation counting. To determine the dose effect of hTCTP on hRheb, the GDP-GTP exchange assay was carried out at different concentration (0 -1.5 M) of the wild-type GST-hTCTP at 15 min after initiation of the reaction.
Cell Culture, Transfection, and Immunoblotting-For in vivo functional assay, the wild-type and mutant hTCTP and human S6K were cloned to the XhoI and BamHI restriction sites of the pHA-N3 vector (modified from the pGFP-N3) with the primers 5Ј-CCGCTCGAGATGATTATCTACCGGGACCTC-3Ј (with the XhoI site underlined) and 5Ј-CGCGGATCCTTAACATT-TTTCCATTTTTAAACCAT-3Ј (with the BamHI site underlined). HEK293T cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin. For cotransfection of the plasmids, 5 ϫ 10 5 HEK293T cells were seeded in 6-well plate and 16 h later were transfected with 2 g of the pHA-S6K plasmid and 2 g of the pHA-hTCTP plasmid or the empty vector with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Forty hours after transfection, the cells were treated by replacing the medium with D-phosphate-buffered saline to remove the amino acids and serum. The cells were then assayed at 45, 60, 75, and 90 min after the treatment for analysis of the wild-type hTCTP, and at 75 min for analysis of the mutant hTCTP.
For co-transfection of the plasmids and siRNA for hTCTP, 2.5 ϫ 10 5 HEK293T cells per well were seeded in 6-well plates and 12 h later were transfected with 2 g of the pHA-S6K plasmid and 100 pmol of hTCTP siRNA or control siRNA with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Sixty hours after transfection, the cells were assayed at 0, 10, 20, and 30 min after removal of amino acids. The oligoribonucleotide sequences of hTCTP siRNA are 5Ј-AGGGAAACUUGAAGAACAGTT-3Ј (sense) and 5Ј-CUGU-UCUUCAAGUUUCCCUTT-3Ј (antisense) (Genepharma).
Homology Modeling and Molecular Dynamics Simulation-The initial models of the hRheb⅐hTCTP complexes were built by superimposing the GDP-bound hRheb (PDB code 1XTQ) or GTP-bound hRheb (PDB code 1XTS) onto Rab8, and hTCTP (PDB code 1YZ1) onto Mss4, respectively, in the crystal structure of the Rab8⅐Mss4 complex (PDB code 2FU5) using the program VMD (version 1.8.4) (26). The models were subjected to molecular mechanics refinement using software package Groningen Machine for Chemical Simulations (GROMACS) 3.3 (27) with the GROMOS 53a6 force field (28). Conformations of the homology models of the complexes and the crystal structures of hRheb-GDP, hRheb-GTP, and hTCTP were sampled by molecular dynamics simulations. All starting structures were first subjected to energy minimization and subsequently placed in the center of a box that extended a minimum of 10 Å from the protein surface and solvated with simple point charge water molecules (29). In addition to the proteins and nucleotides, the system contains 16,131 water molecules, 36 Na ϩ ions, and 29 Cl Ϫ ions with a final salt concentration of 100 mM NaCl to neutralize charges in the system and to emulate physiological conditions. A short equilibration run of 100 ps was then performed with the protein atoms restrained. The production runs were 15 ns long. Electrostatic interactions were evaluated using the Particle-Mesh Ewald method (30, 31) with a real space cutoff of 1.0 nm. Van der Waals interactions were modeled by a Lennard-Jones potential with a 1.4 nm cut-off. All of the bonds were constrained for the protein using the LINCS algorithm (32). During the simulations, temperature was maintained at 300 K, and the system pressure was coupled using the Berendsen method (33). Simulations of each set-up were repeated three times with different random initial velocities. The average structure was calculated from the last 10-ns time period of the 15-ns trajectory (path of the movements of the simulated model) using snapshots at intervals of 10 ps. Overall root mean square deviation (r.m.s.d.) values of the proteins were calculated using the backbone atoms only. r.m.s.d. values of the switch I region and the bound nucleotide were calculated using all atoms except hydrogen. The figures were prepared using PyMOL and VMD molecular visualization programs (26).
Crystallization, Diffraction Data Collection, and Structure Determination-Crystallization was carried out at 4°C using the hanging drop, vapor-diffusion method. Crystals of the E12V mutant hTCTP were grown in drops containing equal volumes of the protein solution (10 mg/ml) and the reservoir solution (0.1 M Tris-HCl, pH 8.2, and 22% polyethylene glycol 6000). Diffraction data were collected from a flash-cooled crystal at beamline NW12 of Photon Factory, Japan, and processed using the program HKL2000 (34). Statistics of the diffraction data are summarized in Table 1.
The structure of the E12V mutant hTCTP was solved with the molecular replacement method implemented in the program PHASER of the CCP4 suite (35) using the structure of the wild-type hTCTP (PDB code 1YZ1) as the search model. The initial structure refinement was carried out with the program CNS (36) following the standard protocols, and the final structure refinement was performed with the maximum likelihood algorithm implemented in the program REFMAC5 (37). A free R-factor monitor calculated with 5% of randomly chosen reflections and a bulk solvent correction were applied throughout the refinement. There are four monomers in the asymmetric unit, which were refined independently. Model building was performed with the program COOT (38) and guided by SIGMAAweighted 2F o Ϫ F c and F o Ϫ F c maps. A summary of the structure refinement statistics is given in Table 1.

RESULTS AND DISCUSSION
hTCTP Can Activate the mTORC1 Pathway-Previously, Hsu et al. showed that dTCTP can facilitate the GDP-GTP exchange of dRheb and regulate the TOR pathway (16). However, two research groups recently reported that hTCTP has no GEF activity toward hRheb and suggested that hTCTP is not involved in the mTORC1 pathway (11,17). To resolve this issue, we first repeated the in vitro GST pulldown assay and GDP release experiments by Hsu et al. (16). Our results show that hTCTP is able to bind nucleotide-free hRheb weakly and to bind GDP-and GTP-bound hRheb very weakly (Fig. 1A) and that hTCTP can accelerate the GDP release of hRheb (Fig. 1B), which are consistent with those reported by Hsu et al. (16). To investigate whether hTCTP is able to activate hRheb and thus the mTORC1 pathway in vivo in stressed cells, we examined the effects of overexpression and down-regulation of hTCTP on S6K activation in amino acid-depleted HEK293T cells, which is a different protocol from that by Rehmann et al. in which the S6K activity was maintained at a basic level or induced by insulin (17). As shown in Fig. 1C, the mTORC1 pathway was inactivated when the cells were depleted of amino acids. However, in the cells overexpressing hTCTP, the mTORC1 pathway remained activated with apparently higher phosphorylation levels of S6K after removal of amino acids. Moreover, the elevated phosphorylation level of S6K was sustained for ϳ75 min after the amino acid depletion treatment and then decreased to a low level similar to that in the control cells at 90 min. These results clearly indicate that hTCTP prolongs activation of the mTORC1 pathway. Because Glu-12 T of dTCTP has been shown to be critical for its GEF activity toward dRheb (16), a corresponding E12V mutant of hRheb was also analyzed, and we show that this mutant lacks the ability to prolong S6K phosphorylation (see details later). We further examined the effect of down-regulation of hTCTP on the phosphorylation of S6K. As shown in Fig. 1D, hTCTP siRNA significantly decreased the level of phosphorylated S6K in amino acid-depleted HEK293T cells. The discrepancies between our results and those obtained by the two other groups (11,17) may be due to the differences of the assay systems. We studied the effect of hTCTP in cells depleted of amino acids without serum starvation overnight, whereas Wang et al. carried out the experiments with cells starved with serum overnight and further depleted of amino acids for 90 min with or without insulin treatment (11) and Rehmann et al. examined the effect of hTCTP in mammalian cells with serum starvation overnight with or without supplementation of insulin (17). Based on our in vitro and in vivo results and analyses of the previously reported data, we conclude that hTCTP does harbor GEF activity on hRheb and can regulate the mTORC1 pathway. Furthermore, we examined the effect of knockdown of hRheb by RNA interference on the activation of S6K by hTCTP overexpression. As shown in Fig. 1E, at 75 min after the treatment, hRheb knockdown by either of the two Rheb siRNAs utilized in this study substantially inhibited S6K activation even when hTCTP was overexpressed (Fig. 1E), indicating that hTCTP is an upstream activator of hRheb and the activation of S6K by hTCTP is dependent on hRheb.
Homology Models of the hRheb⅐hTCTP Complexes Predict the Key Residues Involved in the Interaction-The previous results by Hsu et al. (16) and our biochemical and cell biological data have demonstrated that hTCTP has a weak GEF activity toward hRheb. To understand the molecular basis of the interaction between hTCTP and hRheb, we attempted to prepare the hRheb⅐hTCTP complex and perform structural study of the complex. However, various experiments so far have failed to obtain a stable hRheb⅐hTCTP complex for crystallization presumably due to the weak interaction between hRheb and hTCTP. Alternatively, structure-based homology modeling is a powerful method to predict transient protein-protein interaction and to provide relatively accurate interaction information (39). Recently the crystal structure of the Rab8⅐Mss4 complex has revealed the mechanism of facilitation of the nucleotide exchange of Rab8 by Mss4 (25). Small GTPases are structurally similar with moderate sequence homology across the superfamily. Comparison of the amino acid sequences of hRheb and Rab8 with the BLAST program from NCBI (blast.ncbi.nlm.nih. gov/Blast.cgi) shows that hRheb shares a sequence identity of 29% and a sequence similarity of 52% with Rab8, and the structure of hRheb resembles that of Rab8 with an r.m.s.d. of 4.3 Å for 98 C␣ atoms. Although hTCTP and Mss4 do not share evident sequence homology, the two proteins are structurally similar with an r.m.s.d. of 2.9 Å for ϳ80 C␣ atoms of the core region (23). Thus, we constructed three-dimensional homology models of the hRheb-GDP⅐hTCTP and hRheb-GTP⅐hTCTP complexes based on the crystal structure of the Rab8⅐Mss4 complex ( Fig. 2A, see "Experimental Procedures" for details of the model building). The modeled complexes resemble each other except the switch I region of hRheb (Fig. 2B). In the work by Rehmann et al., a model of the hTCTP⅐Rab8 complex was constructed by superposing hTCTP with the Mss4/Rab8 structure and using Rab8 as a model for hRheb (17). In our model, the crystal structure of hRheb is used rather than that of Rab8 in which the switch I region forms a loop instead of an ␣-helix, and thus the steric clash between the insertion of hTCTP and hRheb predicted by Rehmann et al. is not observed.
In our model, at the protein-protein interface strand ␤2 of hRheb (the nomenclature of the secondary structures of hRheb is after that of Yu et al. (24)) interacts with strand ␤7 of hTCTP (the nomenclature of the secondary structures of hTCTP is shown in supplemental Fig. S1) to form an inter-molecular anti-parallel ␤-sheet as observed in the Rab8⅐Mss4 complex (25) (Fig. 2C). The main chains of residues Glu-40 R (residues of hRheb will be designated by a superscripted suffix R and residues of hTCTP by a superscripted suffix T hereafter), Asn-41 R and Thr-42 R on ␤2 of hRheb interact with those of Glu-80 T , Thr-81 T , and Ser-82 T on ␤7 of hTCTP. The main chain of Thr-44 R forms hydrogen bonds with the side chain of Gln-79 T ; the side chains of Thr-42 R and Glu-40 R form a hydrogen bond with the side chains of Glu-80 T and Thr-81 T , respectively. Moreover, the ␤1-␤2 loop (residues 9 -12) and the TCTP2 signature motif (residues 138 -140) of hTCTP also make interactions with ␤2 of hRheb (Fig. 2D). Specifically, residue Glu-138 T shares a similar physicochemical property as the equivalent residue Asp-96 of Mss4 and forms a salt bridge with Lys-45 R (equivalent to Arg-48 of Rab8), which is also conserved in the Rab8⅐Mss4 complex. Residues Glu-12 T and Met-140 T are also involved in the interactions which, however, differ from their counterparts in the Rab8⅐Mss4 complex. Glu-12 T forms a salt bridge with Lys-45 R , whereas the equivalent residue Arg-29 of Mss4 is not involved in interaction with Rab8. Met-140 T makes hydrophobic contacts with Ile-24 R , Val-32 R , and Phe-43 R , whereas the equivalent residue Glu-98 of Mss4 forms a salt bridge with Lys-46 of Rab8 (corresponding to Phe-43 R ). These distinct interactions may account for the specificities of these GEFs for their respective targets.

MD Simulations of the Modeled Complexes Suggest a Major Conformational Change of Switch I of hRheb in the GDP-GTP
Exchange-Various structural studies of small GTPases have shown that switch I, switch II, and/or the P-loop are essential for nucleotide binding and usually undergo significant conformational changes in the GDP-GTP exchange (40,41). In the Rab8⅐Mss4 complex, switch I of the unliganded Rab8 interacts directly with Mss4 and displays a conformation substantially different from that of the homolog Sec4, suggesting that Mss4 binding may induce a conformational change of switch I of Rab8 (25). Analysis of the hRheb⅐hTCTP models predicts that switch I of hRheb assumes different conformations when bound with GDP and GTP, even though it does not participate in interaction with hTCTP, whereas both switch II and the P-loop adopt similar conformations and have no interaction with hTCTP in both complexes. To identify the mobile region(s) of the hRheb⅐hTCTP complexes and sample more conformational spaces, we carried out MD simulations of the  AUGUST 28, 2009 • VOLUME 284 • NUMBER 35 modeled complexes as the MD simulation method has been increasingly successful in characterization of protein-protein interaction (42). We performed 15-ns simulations of the mod-eled complexes and the crystal structures of hRheb-GDP, hRheb-GTP, and hTCTP, respectively. The overall r.m.s.d. values of the backbone atoms of the complexes fluctuate around 2.5 Å, which is about the average r.m.s.d. between the simulated and experimental structures (43).

Structure Model of the hRheb⅐hTCTP Complex
Analysis of the MD simulation trajectory of the modeled hRheb-GDP⅐hTCTP complex shows that switch I of hRheb displays a large movement (Fig. 3A). Comparison of the initial and average structures indicates that switch I of hRheb displaces ϳ4 Å from its starting position toward hTCTP (Fig. 3B). In the average structure, switch I of hRheb interacts with hTCTP mainly through three residues: Tyr-35 R , Asp-36 R , and Ile-39 R (Fig. 3C). Tyr-35 R and Ile-39 R form a hydrophobic patch and point toward a hydrophobic cluster of hTCTP consisting of Phe-83 T , Tyr-91 T , and Pro-142 T . Additionally, Asp-36 R of hRheb forms a salt bridge with Lys-90 T of hTCTP. The movement of switch I is accompanied by conformational and positional changes of the bound nucleotide. Comparison of the initial and average structures shows that GDP exhibits a displacement of ϳ6 Å from its starting position (Fig.  3A). In the initial model, GDP is bound in the canonical mode: the nucleoside moiety of GDP is sandwiched by 3 10 helix 2 and the switch I-containing loop, and the diphosphoryl group is stabilized by the P-loop. In the average structure, displacement of switch I weakens the constraints on the nucleoside, leading to the movement of the nucleoside with the tip of the base pointing away from the binding pocket (Fig. 3B). The diphosphoryl group maintains a similar position which appears to be constrained by the stable P-loop.
In contrast, simulation of the modeled hRheb-GTP⅐hTCTP complex shows that both switch I and the bound GTP are stable, and Tyr-35 R covers the top of the nucleotide and forms a hydrogen bond with the ␥-phosphate, which is also seen in the crystal structure (24). Similarly, simulations of the GDP-and GTP-bound hRheb structures show that, in the GDP-bound hRheb, switch I exhibits some conformational change at the beginning of the simulation but converges to a stable conformation along the simulation and GDP exhibits a displacement of ϳ5 Å; whereas in the GTP-bound hRheb, both switch I and GTP maintain stable positions (Fig. 3A). Analyses of the GTP-and GDP-bound hRheb structures provide a possible explanation for these results. In the hRheb-GDP structure, switch I of hRheb has little contacts with the nucleotide, and thus both switch I and GDP have relatively high flexibility; whereas in the GTP-hRheb structure, GTP binds tightly to switch I via four hydrogen bonds, which restrict the flexibility of switch I and GTP (supplemental Fig. S2) (24). During all of the simulations, switch II and the P-loop undergo little conformational changes (data not shown), which are consistent with the observation that these regions do not show obvious conformational differences in the GDP-and GTP-hRheb structures (24). The simulation results, together with the previous crystal structure results, suggest that switch I of hRheb has a great flexibility and undergoes a significant conformational change upon GDP-GTP exchange. In the presence of hTCTP, switch I of hRheb moves toward hTCTP and decreased contacts between switch I and GDP result in increased dynamics of switch I and GDP, suggesting that hTCTP binds to the switch I region of hRheb and opens the nucleotide-binding site to facilitate GDP dissociation. This is in consistent with the notion that GEFs function by disturbing the nucleotide-binding site of small GTPases through induction of conformational changes of the switch regions and/or the P-loop (44).
Validation of the Key Residues in the hRheb-hTCTP Interaction-Analyses of the hRheb⅐hTCTP models and the MD simulations have allowed us to predict the key residues participating in the hRheb-hTCTP interaction and the region undergoing significant conformational change during the GDP-GTP FIGURE 4. Validation of the key residues in the hRheb-hTCTP interaction. Binding assay of wild-type and mutant hRheb with the GST-fused wild-type hTCTP (A) and that of wild-type hRheb with the GST-fused wild-type and mutant hTCTP (B). The GST pulldown results are shown in the upper panel, and equal loading of hRheb and GST-fused hTCTP proteins are in the middle and lower panels, respectively. The apparent molecular weight of the K45D mutant hRheb is slightly larger than that of the other hRheb proteins for an unknown reason. GST served as a negative control. C, in vitro GDP release assay of hRheb accelerated by wild-type and mutant hTCTP. D, in vivo functional analysis of the capability of wild-type and mutant hTCTP to activate the mTORC1 pathway. The phosphorylation level of S6K was examined at 75after removal of the amino acids. Actin served as a loading control. exchange of hRheb. Alignments of the available Rheb sequences from 24 species and the corresponding TCTP sequences show that the involved residues Tyr-35 R , Glu-12 T , and Glu-138 T are strictly conserved and residues Phe-43 R , Lys-45 R , and Met-140 T are highly conserved (supplemental Fig. S1). Besides, residues Asp-36 R and Lys-90 T are conserved in mammals and birds. To further investigate the functional roles of the residues at the interaction interface, we performed site-directed mutagenesis studies and in vitro GST pull-down assays. We first examined Lys-45 R of switch I of hRheb and Glu-12 T and Glu-138 T of hTCTP2, which are predicted to form two salt-bridging interactions (Fig. 2D). The GST pulldown results show that mutation of Lys-45 R to Asp abrogates the ability of hRheb to bind hTCTP (Fig. 4A), and similarly mutation of either Glu-12 T to Val or Glu-138 T to Ala also abolishes the binding of hTCTP with hRheb (Fig. 4B). The MD simulation results of the hRheb-GDP⅐hTCTP complex model predict that the switch I region of hRheb interacts with hTCTP via extensive hydrophobic interactions. In particular, Tyr-35 R and Ile-39 R of hRheb make hydrophobic contacts with several residues of hTCTP, including Phe-83 T , Tyr-91 T , and Pro-142 T . Mutation of Tyr-35 R to Ala impairs these hydrophobic contacts and therefore diminishes the binding of hRheb with hTCTP ( Fig.  4A). In the simulated hRheb-GDP⅐hTCTP complex, Asp-36 R of hRheb forms a salt bridge with Lys-90 T of hTCTP. Consistently, the K90E mutant hTCTP exhibits impaired binding with hRheb due to loss of the salt bridge (Fig. 4B).
We further examined the abilities of the hTCTP mutants to accelerate the GDP release of hRheb in vitro and to regulate the mTORC1 pathway in vivo. Our results show that the activities of the hTCTP mutants are correlated well with their binding abilities with hRheb in general. The E12V, K90E, and E138A mutants of hTCTP exhibit significantly diminished capabilities to stimulate GDP dissociation of hRheb (Fig. 4C). The R5A mutant seems to have a slightly enhanced GEF activity compared with the wild-type protein, but the difference appears to be statistically not significant. Because the Y35A and K45D mutants of hRheb showed impaired binding with hTCTP, their abilities to respond to hTCTP in the GDP release assay were also examined. The velocities of GDP release by the mutants alone were similar to that of the wild-type hRheb (supplemental Fig. S3). However, neither mutant showed enhanced GDP release in the presence of hTCTP (Fig. 4C). The in vivo activities of those mutants were further analyzed. Consistently, the mutations (E12V, K90E, and E138A) that impair the GEF activity of hTCTP in vitro also impede its ability to activate the mTORC1 pathway in vivo (Fig. 4D). Quantitative analysis of results in Fig. 4D shows that the level of the phosphorylated S6K in cells transfected with the E138A mutant is ϳ50% of that in cells with the wild-type hTCTP. In particular, phosphorylation of S6K was not observed in the cells overexpressing the E12V and K90E mutants of hTCTP. These results indicate that the key residues of hTCTP involved in the interaction with hRheb are important for its GEF activity toward hRheb and its regulatory function in the mTORC1 pathway, supporting the notion that hTCTP regulates the mTORC1 pathway via its interaction with hRheb.

Glu-12 of hTCTP Is Essential for Its Biological Function-The
previous biochemical data have shown that mutation of Glu-12 T of dTCTP to Val abolishes its GEF activity toward dRheb (16). Our in vitro biochemical and in vivo cell biological assays also demonstrate that the E12V mutant of hTCTP has undetectable binding to hRheb and an abolished GEF activity to hRheb (Fig. 4, A and C) and can no longer activate the mTORC1 pathway (Fig. 4D). Sequence analysis shows that Glu-12 of TCTP is strictly conserved in all species (supplemental Fig. S1). These results indicate that Glu-12 T of hTCTP is a key residue involved in the interaction with hRheb. To understand the molecular basis of the effect caused by the E12V mutation of hTCTP, we determined the crystal structure of the E12V mutant of hTCTP at 2.6-Å resolution ( Table 1). The space group of the E12V mutant belongs to P2 1 2 1 2 1 , which is different from that of the wild-type hTCTP (P2 1 ) (PDB code 1YZ1). The overall structure of the mutant is very similar to that of the wild-type protein with an overall r.m.s.d. of 1.2 Å for all atoms, indicating that the E12V mutation does not cause obvious conformational change of the protein (Fig. 5A). The previous structural analysis of the wildtype hTCTP by Thaw et al. suggests that Glu-12 T , Leu-78 T , and Glu-138 T of hTCTP may form a potential small GTPase-binding groove (23). In the structure models of the hRheb⅐hTCTP complexes, both Glu-12 T and Glu-138 T of hTCTP form saltbridging interactions with Lys-45 R of hRheb. The importance of these two residues is further supported by both in vitro and in vivo assay results. However, Leu-78 T of hTCTP is not involved in direct interaction with hRheb. Detailed structural comparison of the wild-type and the E12V mutant hTCTP in the putative GTPase-binding groove region shows that both Leu-78 T and Glu-138 T maintain similar side-chain conformations, whereas Arg-5 T adopts different side-chain conformations. In the wild-type hTCTP structure, the side chain of Arg-5 T points toward and may form a potential electrostatic interaction with the side chain of Glu-12 T (ϳ4 Å); while in the E12V mutant hTCTP structure, the side chain of Arg-5 T points away from the side chain of Val-12 T apparently due to the change of the side chain (Fig. 5B). As the potential electrostatic interaction between Arg-5 T and Glu-12 T could weaken the interaction between Glu-12 T and Lys-45 R by neutralizing the negative charge of Glu-12 T , we predict that mutation of Arg-5 T to Ala would enhance the salt-bridging interaction between Glu-12 T and Lys-45 R and thus the hRheb-hTCTP interaction. As expected, the R5A mutant hTCTP shows an increased binding ability with hRheb (Fig. 4B) and displays a GEF activity comparable to if not stronger than that of wild-type hTCTP (Fig. 4C), further supporting the importance of the interaction between Glu-12 T of hTCTP and Lys-45 R of hRheb. The R5A mutant hTCTP did not exhibit an increased activating ability for the mTORC1 pathway probably due to the very low expression level of the mutant in the cells (Fig. 4D). These results not only validate our homology models but also support the notion that Glu-12 T is essential for hTCTP function due to its critical role in the interaction with hRheb. In addition, these results suggest that the observed effect of the E12V mutation is not through causing significant conformational change of hTCTP, instead, it is through subtle changes of the interactions of Glu-12 T with the surrounding residues of hTCTP and hRheb, particularly Lys-45 R of hRheb. It is noteworthy that there are four hTCTP molecules in the asymmetric unit that form two homodimers, and there is an intermolecular disulfide bond between Cys-172 T of adjacent monomers (Fig. 5A). Considering that murine TCTPs are prone to interaction with each other through a C-terminal region of residues 126 -172 (45), the disulfide bond seen in the mutant hTCTP structure may account for the tendency of TCTP to dimerize or oligomerize. Further investigation is needed to find out whether dimerization of TCTP and formation of the disulfide bond have any roles in the modulation of TCTP function.
In summary, our biochemical, cell biological, modeling, and structural data together demonstrate that hTCTP positively regulates the mTORC1 pathway via acceleration of the GDP-GTP exchange of hRheb. Our results not only resolve the argument about the involvement of hTCTP in the mTORC1 pathway but also provide insights into the molecular mechanism of the biological function of hTCTP in the mTORC1 pathway.
Protein Data Bank Accession Code-The structure of the E12V mutant hTCTP has been deposited with the RCSB Protein Data Bank under accession code 3EBM. FIGURE 5. Crystal structure of the E12V mutant hTCTP. A, overall structure of the E12V mutant hTCTP. The E12V mutant forms a homodimer (one subunit in cyan and the other in magenta) with an inter-subunit disulfide bond formed between Cys-172 (shown with side chains) of the two monomers. The two Val-12 residues are shown with side chains and indicated by circles. B, structural comparison of the wild-type (cyan) and E12V mutant (magenta) hTCTP in the putative GTPase-binding groove region.