P-Rex1 Links Mammalian Target of Rapamycin Signaling to Rac Activation and Cell Migration*

Polarized cell migration results from the transduction of extra-cellular cues promoting the activation of Rho GTPases with the intervention of multidomain proteins, including guanine exchange factors. P-Rex1 and P-Rex2 are Rac GEFs connecting Gβγ and phosphatidylinositol 3-kinase signaling to Rac activation. Their complex architecture suggests their regulation by protein-protein interactions. Novel mechanisms of activation of Rho GTPases are associated with mammalian target of rapamycin (mTOR), a serine/threonine kinase known as a central regulator of cell growth and proliferation. Recently, two independent multiprotein complexes containing mTOR have been described. mTORC1 links to the classical rapamycin-sensitive pathways relevant for protein synthesis; mTORC2 links to the activation of Rho GTPases and cytoskeletal events via undefined mechanisms. Here we demonstrate that P-Rex1 and P-Rex2 establish, through their tandem DEP domains, interactions with mTOR, suggesting their potential as effectors in the signaling of mTOR to Rac activation and cell migration. This possibility was consistent with the effect of dominant-negative constructs and short hairpin RNA-mediated knockdown of P-Rex1, which decreased mTOR-dependent leucine-induced activation of Rac and cell migration. Rapamycin, a widely used inhibitor of mTOR signaling, did not inhibit Rac activity and cell migration induced by leucine, indicating that P-Rex1, which we found associated to both mTOR complexes, is only active when in the mTORC2 complex. mTORC2 has been described as the catalytic complex that phosphorylates AKT/PKB at Ser-473 and elicits activation of Rho GTPases and cytoskeletal reorganization. Thus, P-Rex1 links mTOR signaling to Rac activation and cell migration.

mTORC1 and mTORC2 are participants of type I insulin-like growth factor-stimulated motility (15).
It is not yet clear if differential signals activate mTORC1 and mTORC2 complexes. Neither is known what effector systems are directly sensitive to mTORC2 activation. Although the participation of Rho guanine exchange factors in the activation of Rho GTPases and cytoskeletal dynamics elicited by mTORC2 is expected, their identity and mechanism of action remain to be revealed. Here we identified the Rac guanine exchange factors P-Rex1 and P-Rex2 as the putative catalytic components of mTORC2 critical for the activation of Rac and cell migration elicited by leucine through the rapamycin-insensitive mTOR complex 2. These results implicate that P-Rex family members link mTORC2 signaling to Rac activation and cell migration in response to the activation of the rapamycin-insensitive mTOR complex.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screening-The DEP-DEP domains of P-Rex1 were used as a bait to screen a human brain cDNA library with the Matchmaker System III (Clontech) following both the manufacturer's instructions and the ones indicated by Vazquez-Prado et al. (29). Clones were obtained under high stringency conditions using media lacking adenine, histidine, leucine, and tryptophan. The specificity of the interaction between the DEP-DEP P-Rex1 domain with the mTOR carboxyl-terminal domain was confirmed using the control bait and prey provided by the Matchmaker System III kit.
Constructs and Antibodies-prk5myc-Rictor, prk5-myc-Raptor, and prk5-HA-G␤L were obtained from D. Sabatini (see Ref. 4). The shRNA for P-Rex1 corresponds to RNA interference codex HP_187924 cloned into the pSM2 vector, which was generated based on the sequence of the hairpin TGCTGTT-GACAGTGAGCGAGGACACAC-TGTGCTTCCAGATTAGTGAAG-FIGURE 1. mTOR interacts with P-Rex1 in yeast and in mammalian cells. A, representation of the domains of mTOR depicting the prey obtained in the yeast two-hybrid screen and the domains of P-Rex1 depicting the bait that was used. B, carboxyl terminus of mTOR, including its kinase domain, and a control, including the T antigen, were tested for interaction with the bait (DEP-DEP domains of P-Rex1) or p53 as a control. Yeast grew in media lacking leucine and tryptophan (ϪLT), which selects for the presence of the plasmids, and only the ones displaying a strong interaction grew under high stringency conditions (media lacking adenine, histidine, leucine, and tryptophan (ϪAHLT)). C, HEK 293T cells were co-transfected with 3XFLAG-tagged mTOR prey and DEP-DEP domains of P-Rex1 fused to GST. GST was affinity-purified with glutathione beads, and both total cell lysates (TCL, left) and pulldowns (AP, right) were resolved on SDS-polyacrylamide gels and analyzed by immunoblotting. WB, Western blot. A specific interaction of mTOR prey was found with the DEP-DEP domains of P-Rex1 and not with the GST protein alone used as a control. D, endogenous mTOR interacts with affinitypurified P-Rex1-DEP-DEP fused to GST and not with GST alone used as a control; right panel, affinity-purified proteins (AP); left panel, total cell lysates (TCL). Dithiobis(succinimidyl propionate) (DSP) was used to stabilize the interaction. E, HEK293T cells were co-transfected with mTOR prey fused to GST and GFP-tagged P-Rex1-DEP-DEP. Total cell lysates were analyzed (TCL, left), and affinity-purified proteins (AP, right) showed a specific interaction of the DEP-DEP domains of P-Rex1 with GST-mTOR prey and not with the GST alone used as control. F, HEK293T cells were co-transfected with full-length 3XFLAG-tagged P-Rex1 and HA-tagged mTOR prey. Immunoprecipitation was carried out. Total cell lysates (left) and HA immunoprecipitates (IP-HA, right) were analyzed by SDS-polyacrylamide gels and immunoblotting. A specific interaction of 3XFLAG-P-Rex1 full length with HA-mTOR prey is shown. G, direct interaction between P-Rex1 and mTOR was assessed by means of in vitro interaction assay between in vitro translated (IVT) mTor prey and P-Rex1 DEP-DEP expressed as GST fusion protein. P-Rex1 DEP-DEP or P-Rex1 carboxyl terminus (GST Cter), used as negative control, was affinity purified with glutathione beads and washed with 0.5 M NaCl before the interaction assay. In vitro translated mTOR prey was detected associated with P-Rex1 DEP-DEP. CCACAGATGTAATCTGGAAGCACAGTGTGTCCGTGCC-TACTGCCTCGGA. mTOR shRNA was obtained from P. Amornphimoltham. Anti-P-Rex1 rat monoclonal antibodies were obtained from M. Hoshino (see Ref. 13).
Cell Culture and Transfection-HEK 293T cells and HeLa cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Cells were transfected using Polyfect (Qiagen) according to the manufacturer's instructions and harvested after 48 h. When indicated cells were starved of leucine for 2 h in Dulbecco's modified Eagle's Specialty Media (Cell & Molecular Technologies) before stimulation.
Cross-linking Assay and Immunoprecipitations-HEK 293T cells in 10-cm dishes were placed on ice, rinsed once with phosphate-buffered saline, and lysed in 1 ml of ice-cold buffer (50 mM Tris (pH 7.5), 5 mM EDTA, 150 mM NaCl, and either 1% Triton X-100 or 0.3% CHAPS, for immunoprecipitation of mTOR complexes) containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml aprotinin). For cross-linking assays 2.5 mg/ml dithiobis(succinimidyl propionate) (Pierce) was added to the lysis buffer, and cells were rotated for 30 min on ice. Cross-linking reactions were quenched by adding 75 l of 1 M Tris-HCl (pH 7.4) followed by an additional 30 min incubation at room temperature. After clearing, 1 l of each antibody for immunoprecipitation was added to the supernatant and rotated for 1 h at 4°C. 25 l of 50% slurry of protein A-Sepharose was added and rotated again for 30 min at 4°C. Immunoprecipitates were washed three times with 1 ml of lysis buffer and resuspended in 1ϫ protein sample buffer. The interaction between endogenous P-Rex1 and mTor was detected using lysates from HeLa cells processed as described for HEK 293T cells; in this case, mTor antibody was from Santa Cruz Biotechnology and PTEN antibody, also from Santa Cruz Biotechnology, was used as a control. and both were goat polyclonal antibodies; 10 l of each antibody per 1 ml of lysate was used. Endogenous P-Rex1 was detected with Anti-P-Rex1 rat monoclonal antibodies obtained from M. Hoshino (see Ref. 13).
Rac-GTP Assays-HeLa cells in 10-cm dishes were transfected with the indicated plasmids and shRNAs; 48 h later, cells were starved and stimulated as described above. When indicated, cells were incubated with rapamycin (Calbiochem) for 30 min before the assay. Cells were rinsed once with phosphatebuffered saline and lysed in 1 ml of ice-cold buffer (50 mM Tris (pH 7.5), 5 mM EDTA, 150 mM NaCl, 10 mM MgCl 2 and 1% Triton X-100) containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml aprotinin). Extracts were cleared, and aliquots were saved to check for total protein expression. 50 l of GST-PAK CRIB beads were added and rotated for 45 min at 4°C. Beads were washed three times with 1 ml of lysis buffer and resuspended in 30 l of protein sample buffer, boiled, and loaded onto a 12% gel.
Migration Assays-Serum-free media containing either L-leucine or 10% FBS, which was used as a positive control, was placed in the bottom wells of a Boyden chamber; media containing HeLa cells, starved for 2 h in serum-free media lacking leucine, were added to the top chamber. The two chambers were separated by a polycarbonate filter membrane (Neuro Probe, 8 m-pore), coated with 10 g/ml fibronectin (Calbiochem). After a 6-h incubation, membranes were stained with crystal violet, placed on a glass slide, and scanned. Densitometric quantitation was performed with ImageJ software. Where indicated cells were previously transfected with plasmids and shRNAs and starved for 2 h before this assay as described above. For the rapamycin assay, cells were starved for 2 h and preincubated with rapamycin for 30 min. Rapamycin was also added to the cell suspension placed on the top chamber wells and also FIGURE 2. Interaction of endogenous mTOR with P-Rex1. The interaction between endogenous mTor and P-REX1 was assessed by immunoprecipitation (IP) using lysates from HeLa cells. A control lysate from cells in which P-Rex1 was knocked down was included in the Western blot (WB) to further support the immunodetection of the band corresponding to endogenous P-Rex1. A goat antibody was used as a negative control. Endogenous mTOR was immunoprecipitated, and endogenous P-Rex1 was found associated with it. Total lysates (TCL, left and middle panels) and immunoprecipitation (IP, right) are shown.

FIGURE 3. mTOR interacts with all the different P-Rex family members.
HEK293T cells were co-transfected with GST-mTOR prey and GFP-tagged P-Rex1, P-Rex2a, and P-Rex2b. Pulldown assays were performed. Total cell lysates (TCL) and pulldowns were resolved on SDS-polyacrylamide gels and analyzed by immunoblotting. Interactions of mTOR prey with the three P-Rex family members were found. No interaction was seen with the GEF PDZ-Rho-GEF (PRG) used as control. WB, Western blot.
to the chemoattractant added to the bottom chamber wells in order to be present during the migration time.
In Vitro Translation and in Vitro Interactions-All in vitro translation reactions were carried out using the Expressway cell-free Escherichia coli expression system (catalog number K9901-00, Invitrogen) according to the manufacturer instructions. Briefly, for each reaction we mixed in an Eppendorf tube, the E. coli SlyD-Extract, 2.5ϫ in vitro protein synthesis E. coli Reaction Buffer (Ϫamino acids), 50 mM amino acids (ϪMet), 75 mM methionine, T7 Enzyme Mix. 3XFLAG-mTor prey cloned into pEF1/His (Invitrogen) was used as DNA template containing the T7 promoter. The reactions were incubated in a shaker at 3000 rpm at 37°C for 30 min. After 30 min of incubation, 50 l of Feed Buffer (2ϫ IVPS Feed Buffer, 50 mM amino acids (ϪMet), 75 mM methionine) was added, followed by an additional incubation for 6 h. The reactions were terminated by adding 1 ml of ice-cold buffer (50 mM Tris (pH 7.5), 5 mM EDTA, 150 mM NaCl, and 1% Triton X-100 containing protease inhibitors). In vitro translated FLAG-mTor prey was incubated overnight with glutathione beads containing either P-Rex1 carboxyl-terminal domain or P-Rex1 DEP-DEP domains, both expressed as GST fusion proteins isolated from transfected HEK 293T cells. Before incubation with in vitro translated mTor prey, beads were washed three times with 0.5 M NaCl. The isolated complexes were denatured in SDS sample buffer under reducing conditions, resolved by 10% SDS-PAGE, and transferred to Immobilon membranes for Western blotting using FLAG and GST antibodies.

RESULTS
mTOR Interacts with P-Rex1 in Yeast and Mammalian Cells-In order to identify proteins that may regulate P-Rex1, we used the tandem DEP domains of P-Rex1 (Fig.  1A) as bait in a yeast two-hybrid screen using a human brain cDNA library. We found that in yeast P-Rex1 DEP-DEP interacted with the carboxyl-terminal region, including the kinase domain, of mTOR; this interaction was specific as indicated by the growth, in restrictive media lacking adenine, histidine, leucine, and tryptophan, of yeast transformed just with the combination of P-Rex DEP-DEP and mTOR prey or the positive control of the system composed of p53 and T antigen (Fig. 1, A and B). To assess in mammalian cells the interaction of P-Rex1 with mTOR, the region of P-Rex1 comprising the two tandem DEP domains and the carboxyl-terminal region of mTOR (corresponding to the prey from the yeast two hybrid) FIGURE 4. mTOR signals to Rac via P-Rex1. A, HeLa cells in serum-free media lacking leucine were stimulated with 52 g/ml of L-leucine for the indicated times, and Rac-GTP was isolated with PAK-CRIB fused to GST. Lysates were resolved on SDS-polyacrylamide gels and analyzed by immunoblotting for Rac, phosphorylated S6, and total S6. mTOR stimulation by leucine activated Rac after 1 min and remained for 10 min. Leucine also induced phosphorylation of the ribosomal protein S6 for the same period of time. Cells transfected with the Rac GEF TIAM were used as positive controls for both Rac activity and S6 phosphorylation. B, HeLa cells transfected with mTOR shRNA were stimulated with leucine; Rac activity and S6 phosphorylation were determined as indicated above. Controls included cells transfected with empty shRNA vector (control) or TIAM with and without mTOR shRNA. A representative blot detecting the indicated proteins is shown. Rac activity as well as S6 phosphorylation decreased dramatically after leucine stimulation in cells transfected with mTOR shRNA. C, HeLa cells transfected with dominant-negative P-Rex1 (EGFP ⌬DH-P-Rex1) were stimulated with leucine; Rac activity and S6 phosphorylation were determined as indicated above. Controls included cells transfected with EGFP or TIAM with and without dominant-negative P-Rex1. A representative blot detecting the indicated proteins is shown. A significant decrease in Rac activity was found in cells transfected with the dominantnegative P-Rex1. D, HeLa cells transfected with P-Rex1 shRNA were stimulated with leucine; Rac activity and S6 phosphorylation were determined as indicated above. Controls included cells transfected with empty shRNA vector (control) or TIAM with and without P-Rex1 shRNA. A representative blot detecting the indicated proteins is shown. A significant decrease in active Rac after leucine stimulation was found in cells transfected with P-Rex1 shRNA. E, HEK 293T cells were co-transfected with 3XFLAG-tagged P-Rex1, GST-mTOR prey, and either EGFP or EGFP-DEP-DEP. GST was affinity-purified with glutathione beads, and total cell lysates (TCL) and pulldowns were resolved on SDS-polyacrylamide gels and analyzed by immunoblotting. The interaction of mTOR prey with 3XFLAG-tagged P-Rex1 was weakened by the transfection of the DEP-DEP domains of P-Rex1 compared with cells transfected with EGFP. GST was used as negative control. F, HeLa cells transfected with the DEP-DEP domains of P-Rex1 were stimulated with leucine; Rac activity was determined as indicated above. Controls included cells transfected with EGFP or TIAM. A representative blot detecting the indicated proteins is shown. Rac activity decreased dramatically after leucine stimulation in cells transfected with the DEP-DEP domains of P-Rex1. WB, Western blot.
were transfected into the human epithelial cell line HEK293T as GST fusion and 3XFLAG-tagged proteins, respectively. The corresponding cell extracts were incubated with glutathione Sepharose beads. The carboxyl-terminal region of mTOR (mTOR prey) was detected associated with P-Rex1-DEP-DEP fused to GST but not with GST used as control (Fig. 1C). Endogenous mTOR also interacted with P-Rex1 DEP-DEP expressed as a GST fusion protein (Fig. 1D). To further analyze this interaction, HEK293T cells were co-transfected with the mTOR prey fused to GST and GFP-tagged P-Rex1 DEP-DEP domains (Fig. 1E) or HA-tagged mTOR prey and 3XFLAG-tagged P-Rex1 (Fig. 1F). In either case, both the bait (P-Rex1-DEP-DEP) and full-length P-Rex1 specifically interacted with mTOR prey (Fig. 1, E-F). To demonstrate a direct interaction between P-Rex1 and mTOR, an in vitro interaction assay was conducted using in vitro translated FLAG-tagged mTOR prey and recombinant P-Rex1 DEP-DEP domains or P-Rex1 carboxyl-terminal domain (used as control), both expressed as GST fusion proteins. P-Rex1 domains were affinity-purified and washed with 0.5 M NaCl before the interaction assay. mTOR prey was detected associated with P-Rex1 DEP-DEP (Fig. 1G). The interaction of both endogenous proteins was also assessed by immunoprecipitating mTOR, and detecting P-Rex1, a control lysate from cells in which P-Rex1 was knocked down, was included in the Western blot to further support the identification of the band corresponding to endogenous P-Rex1 (Fig. 2). We next wished to evaluate the possible interaction between mTOR and other members of the P-Rex family. HEK 293T cells were cotransfected with GST-tagged mTOR prey and GFP-tagged P-Rex1, P-Rex2a, or P-Rex2b (Fig. 3). The GST pulldown assays proved that mTOR not only interacts with P-Rex1 but also with P-Rex2a and P-Rex2b. No interaction was observed with the Rho GEF PDZ-RhoGEF (PRG) used as control (Fig. 3). These results, together with the two-hybrid screen, demonstrate that P-Rex family of Rac GEFs, including P-Rex1, P-Rex2a, and P-Rex2b, are able to interact directly with mTOR, opening the possibility that this group of GEFs could participate in the activation of Rac and cytoskeletal events downstream of mTORC2.
mTOR Activates Rac via P-Rex1-The rapamycin-insensitive mTORcontaining complex, mTORC2, has been found to control the actin cytoskeleton through Rho GTPases (4,6). Thus, we wanted to determine whether the interaction of mTOR with P-Rex1 could be involved in the recently described ability of mTORC2 to promote the activation of Rac. Since amino acids activate mTOR, we decided to use L-leucine as a model to evaluate the role of this protein in Rac activation and cell migration. First, we evaluated whether mTOR stimulation by leucine would promote the formation of active, GTP-bound, Rac in HeLa cells (Fig. 4A). The amount of GTP-bound Rac increased with the addition of leucine after 1 min of stimulation and remained elevated for 10 min. These results correlated with the activation of S6K, a very well characterized mTOR effector, determined by an increase in the phosphorylation of its substrate, S6, in response to leucine, in samples from the same cell lysates. TIAM, a well characterized Rac GEF, used as positive control, was able to induce activation of Rac and promote the phosphorylation of the ribosomal protein S6. Next, considering that knocking down mTOR with specific short interfering RNA has been reported to decrease basal Rac-GTP in NIH 3T3 cells (6), we evaluated the participation of mTOR in leucine-induced Rac activation; mTOR shRNA was transfected, and the amount of GTP-bound Rac was measured after leucine stimulation (Fig.  4B). The ability of leucine to induce the activation of Rac was dramatically decreased in cells transfected with mTOR shRNA; as expected, the phosphorylation of S6, induced by leucine, was also prevented in cells in which mTOR was knocked down (Fig.  4B). These results correlated with the important decrease of mTOR detected by Western blot (Fig. 4B). To investigate the participation of P-Rex1 in the ability of mTORC2 to promote Rac activation in response to leucine, two independent approaches were used. First, a dominant-negative P-Rex1 (⌬DH-P-Rex1) was transfected, and the amount of GTP-bound Rac was measured after leucine stimulation (Fig. 4C). Second, P-Rex1 shRNA was used to knock down the expression of endogenous P-Rex1 (Fig. 4D). The activation of Rac induced by leucine was significantly decreased by both complementary approaches (Fig. 4, C and D). These results indicated a relevant participation of this GEF in the signaling pathway from mTOR to Rac.
It is well known that branched chain amino acids such as leucine regulate the mTORC1 complex by activation of mTOR and further phosphorylation of the S6 protein (27,32). In these experiments we wished to address the state of phosphorylation of the S6 after leucine stimulation in nontransfected cells (Fig.  4A) and in cells transfected with the mTOR shRNA (Fig. 4B), the dominant-negative P-Rex1 (Fig. 4C), and the shRNA to P-Rex1 (Fig. 4D). The results show a significant decrease in the phosphorylation of S6 only in the cells transfected with mTOR shRNA (Fig. 4B) as opposed to the cells transfected with either the dominant-negative P-Rex1 or the P-Rex1 shRNA, in which the phosphorylation of S6 increased with the addition of leucine to the media. Together, these results indicate that S6 phosphorylation is independent of P-Rex1, supporting the hypothesis that P-Rex1 is active as a component of mTORC2 and inactive when associated to mTORC1. A GST pulldown assay was performed to further analyze the role of the DEP-DEP domains of P-Rex1 with respect to mTOR interaction. As seen in Fig. 4E, the interaction of both proteins (P-Rex1 and mTOR prey) was dramatically weakened by the expression of the DEP domains of P-Rex1, which compete with the endogenous domains of this protein. In addition, P-Rex1 DEP domains inhibit the activation of Rac induced by leucine (Fig. 4F), and this result is coincident with the reduced effect of Rac activation seen with the dominant-negative P-Rex1 (Fig. 4C).
P-Rex1 Is Part of Both mTOR Complexes, mTORC1 and mTORC2-Since mTOR has been found in two different multiprotein complexes (4, 6, 26 -28), one of which is able to activate Rac and induce cytoskeletal changes by a rapamycin-resistant mechanism; we predicted that P-Rex1 would be a component of the mTORC2, able to induce this effect. In order to assess this possibility, we explored the presence of P-Rex1 in isolated mTOR complex 1 (mTORC1) or complex 2 (mTORC2). HEK 293T cells were transfected with 3XFLAG-tagged P-Rex1, HA-tagged G␤L/mLST8, and either Myc-tagged Raptor or Myc-tagged Rictor; mTORC1 and mTORC2 were isolated by immunoprecipitation of Raptor and Rictor, respectively (Fig. 5A). Unexpectedly, P-Rex1 was found associated with both mTORC1 and mTORC2 (Fig. 5A), and it did not interact with an unrelated Myctagged protein (associate molecule FIGURE 6. mTOR promotes cell migration via P-Rex1. A, HeLa cells transfected with EGFP or dominantnegative P-Rex1 (EGFP ⌬DH-P-Rex1) were subjected to chemotaxis assays 48 h post-transfection. Cells starved for 2 h in serum-free media lacking leucine were stimulated with L-leucine (520 g/ml) for 6 h at 37°C in a Boyden chamber. 10% FBS was used as positive control. Migrating cells were stained, scanned, and analyzed by densitometry. A remarkable increase in the number of control migrating cells after a leucine or FBS stimulation was found. Cells transfected with dominant-negative P-Rex1 showed a dramatic decrease in their migration when stimulated by leucine. Relative cell migration (% of FBS) was determined by comparing control cells with cells transfected with the dominant-negative. Three independent experiments were averaged and plotted. Error bars represent S.E. *, p Ͻ 0.05 difference with control group. B, HeLa cells transfected with empty shRNA vector (control), P-Rex1 shRNA, or mTOR shRNA were subjected to chemotaxis assays, stained, and analyzed by densitometry as described in A. An increase in the number of control migrating cells was found after leucine or FBS stimulation. Cells transfected with P-Rex1 shRNA and mTOR shRNA showed a significant decrease in their migration when stimulated by leucine. Relative cell migration (% of FBS) was determined by comparing control cells with cells transfected with the indicated shRNAs. Three independent experiments were averaged and plotted. Error bars represent S.E. *, p Ͻ 0.05 difference with control group. C, HeLa cells were starved for 2 h, preincubated with either vehicle (DMSO) or rapamycin (20 and 100 ng/ml, respectively) for 30 min, and placed in a Boyden chamber. Rapamycin or vehicle was also added to both the cell suspension and the chemoattractants (leucine and 10% FBS) to the same concentrations indicated above. Cells were thus left to migrate in the presence of rapamycin for 6 h at 37°C. Migrating cells were stained, scanned, and analyzed by densitometry. Only a slight decrease in the number of migrating cells toward serum was found when rapamycin was present at the higher concentration. Relative cell migration (% of FBS) was determined by comparing vehicle-treated cells with rapamycin-treated cells. Four independent experiments were averaged and plotted. Error bars represent S.E. *, p Ͻ 0.05 difference with vehicle group. D, HeLa cells were starved for 2 h, incubated with either vehicle (DMSO) or rapamycin (20 ng/ml), and stimulated with L-leucine (52 g/ml) for the indicated times. Cells were lysed, and a Rac-GTP pulldown assay was carried out. Lysates were resolved on SDS-polyacrylamide gels and analyzed by immunoblotting for GTP-bound Rac, total Rac, phosphorylated S6, and total S6. A representative blot is shown. No significant decrease in active Rac after leucine stimulation was found in cells treated with rapamycin compared with vehicle-treated cells. E, model depicting P-Rex1 as part of both mTOR complexes and only active in mTORC2, in which this GEF is involved in Rac signaling and cell migration.
with the Src homology 3 domain of STAM) that was used as a negative control (Fig. 5B).
mTOR Induces Cell Migration via P-Rex1-Regulation of the actin cytoskeleton by mTORC2 has been described (4,6). Moreover, recent findings in Dictyostelium show the involvement of the TORC2 complex in chemotaxis and cell polarity (33). So far, we have demonstrated that leucine activates Rac via mTOR and P-Rex1-dependent pathways; therefore, we wished to determine whether leucine could also induce cell migration dependent on P-Rex1. The ability of HeLa cells to migrate in the presence of leucine was demonstrated (supplemental Fig. 1). Then a migration assay was performed using HeLa cells transfected with the dominant-negative P-Rex1 or the shRNA to P-Rex1, and cells transfected with empty shRNA vector (pENTR) were also used as a control. Cells were either nonstimulated or stimulated with leucine or fetal bovine serum as a positive control and left to migrate for 6 h at 37°C. The results show a remarkable increase in the number of migrating cells when stimulated with leucine or serum in control cells. On the contrary, cells transfected with either dominant-negative P-Rex1 or P-Rex1 shRNA showed a dramatic decrease in the number of migrating cells only when stimulated with leucine ( Fig. 6, A and B). As expected, cells in which mTOR was knocked down with specific shRNA were unable to migrate in response to a leucine gradient (Fig. 6B). A transfection efficiency of at least 50% (monitored with GFP) was accomplished for these experiments. To discern if mTORC1 is relevant for the chemokinetic activity induced by leucine, we evaluated the migration of HeLa cells incubated with rapamycin, an inhibitor of mTORC1; the results, shown in Fig. 6C, indicated that only serum-induced migration of HeLa cells was only slightly decreased in presence of a high dose of rapamycin, whereas the one induced by leucine was not affected by it. A Rac activation assay was also carried out; the results ( Fig. 6D) showed no decrease in Rac activity after leucine stimulation in cells treated with rapamycin compared with control cells. Overall, these results demonstrate that leucine induces cell migration and Rac activation in a P-Rex1-dependent manner involving the participation of the mTORC2 complex.

DISCUSSION
Our findings provide a novel mechanism by which mTORC2 can control the actin cytoskeleton through the activation of Rho GTPases (4,5). Based on our results, we can postulate that by virtue of its ability to form molecular complexes with mTOR, P-Rex1 can link mTORC2 activation to the stimulation of Rac signaling, leading to the regulation of the actin cytoskeleton and cell migration. This molecular pathway involves proteins that have previously been identified as part of the mTORC2 complex, mLST8 and Rictor (4, 5) (Fig. 6E). Our data show that both P-Rex1 and mTOR are needed for Rac activation and cell migration induced by leucine, and whether they are part of the same signaling route or acting through independent pathways remains to be elucidated. However, as P-Rex1 is also a constituent of mTORC1, it may also participate in mTORC1-initiated Rac activation, as recently observed in cell migration in response to insulin-like growth factor (15); or alternatively, P-Rex1 may be a component of oligomeric complexes contain-ing both mTORC1 and mTORC2 (16 -18). Both possibilities are under current evaluation. The fact that TIAM was able to activate both Rac and the ribosomal protein S6 drew our attention. These results are coincident with demonstrated ability of Rac to promote the activation of p70 S6K (30) and the ability of this Rac GEF, in a complex with spinophilin, to contribute to the activation of p70 S6K (31). The dominant-negative P-Rex1 and the shRNA to P-Rex1 had no effect on pS6 levels; this could be due to P-Rex1 participating only in Rac activation processes (15) even when it is also associated to complex 1 of mTOR.
Eukaryotic cells rely on the polarization of the actin cytoskeleton and chemotaxis to find nutrients in order to survive starvation conditions; the participation of TOR signaling in these events has been recognized in yeast and Dictyostelium discoideum (18,19). In yeast, ROM2 is the RhoGEF acting downstream of TOR2 in the pathway leading to the reorganization of the actin cytoskeleton (20). Interestingly, ROM2 contains a DEP domain as part of its structure; thus the finding that P-Rex1 uses its DEP domains to interact with mTOR suggests that P-Rex1 and its homologs might represent the mammalian orthologs of ROM2. Indeed, our findings showing that HeLa cells can migrate in the presence of leucine, and the central role of P-Rex1 in this process, may reflect the existence of a mammalian counterpart of this ancestral survival mechanism. Furthermore, the identification of P-Rex1 as a participant in the signaling route by which mTOR controls Rac activation and cell migration may also provide a novel therapeutic target for pharmacological intervention in many pathological conditions that are characterized by the overactivity of mTOR, including human cancer (21).