PTEN inhibits PREX2-catalyzed activation of RAC1 to restrain tumor cell invasion

Abstract

The tumor suppressor PTEN restrains cell migration and invasion by a mechanism that is independent of inhibition of the PI3K pathway and decreased activation of the kinase AKT. PREX2, a widely distributed GEF that activates the GTPase RAC1, binds to and inhibits PTEN. We used mouse embryonic fibroblasts and breast cancer cell lines to show that PTEN suppresses cell migration and invasion by blocking PREX2 activity. In addition to metabolizing the phosphoinositide PIP₃, PTEN inhibited PREX2-induced invasion by a mechanism that required the tail domain of PTEN, but not its lipid phosphatase activity. Fluorescent nucleotide exchange assays revealed that PTEN inhibited the GEF activity of PREX2 toward RAC1. PREX2 is a frequently mutated GEF in cancer, and examination of human tumor data showed that PREX2 mutation was associated with high PTEN expression. Therefore, we tested whether cancer-derived somatic PREX2 mutants, which accelerate tumor formation of immortalized melanocytes, were inhibited by PTEN. The three stably expressed, somatic PREX2 cancer mutants that we tested were resistant to PTEN-mediated inhibition of invasion but retained the ability to inhibit the lipid phosphatase activity of PTEN. In vitro analysis showed that PTEN did not block the GEF activity of two PREX2 cancer mutants and had a reduced binding affinity for the third. Thus, PTEN antagonized migration and invasion by restraining PREX2 GEF activity, and PREX2 mutants are likely selected in cancer to escape PTEN-mediated inhibition of invasion.

Fig. 1. Pten KO–induced migration requires PREX2

(A) Immunoblot of PI3K pathway components in immortalized wild-type (WT) or KO MEFs of indicated genotypes under steady-state growth conditions. Immunoblots are representative of three experiments. pAKT 308, phospho-AKT (Thr³⁰⁸). (B) MEFs were subjected to migration assays using collagen-coated membranes and a PDGF gradient. Error bars represent the mean ± SEM of three experiments using six pairs of MEFs from three different litters. (C) MEFs from each genotype were plated in serum-free medium, and cell spreading (cell area) was measured over three experiments. Spreading was measured for at least 50 MEFs from each genotype. (D) Pten KO MEFs were transfected with siRNAs against Rac1 (KD1 and KD2) or nontargeting siRNA and subjected to migration assays. Error bars represent the mean ± SEM of three experiments. All P values were calculated using two-tailed t tests. Experiments were performed using early-passage immortalized MEFs.

Fig. 2. PTEN does not suppress invasion in the absence of PREX2

(A) PREX2 and PTEN proteins in breast cancer cell lines. (B) PTEN consists of a catalytic phosphatase domain (CAT), a C2 domain, and a tail with a PDZ binding domain. (C and D) WT PTEN, G129E, C124S, and C2-tail suppress invasion in (C) BT549 and (D) SUM149 cells. Cells were transfected and used in invasion assays with an FBS gradient. (E) PREX2 protein in ovarian cancer cell lines. (F) Knockdown (KD) of PTEN increases invasion in IGROV1 cells. (G) Knockdown of PREX2 reduces invasion in BT549 cells. (H) PTEN does not suppress invasion after PREX2 knockdown. BT549 cells stably expressing shRNA against PREX2 or nontargeting shRNA were transfected as indicated. Cells invaded toward an FBS gradient (blue) or no gradient (red). Error bars in (C), (D), and (F) to (H) represent the mean ± SEM of three experiments. P values were calculated using two-tailed t tests. Immunoblots in (A) and (C) to (H) are representative of at least two experiments.

Fig. 3. PTEN’s intact PDZ binding domain is required for antagonism of PREX2-driven invasion, and the PTEN C2-tail is sufficient to suppress invasion

(A) PREX2 drives invasion. BT549 cells were transfected as indicated and subjected to invasion assays using an FBS gradient or no gradient. (B) PTEN constructs used in invasion assays. (C) PTEN C2-tail, G129E, and C124S block PREX2-driven invasion. BT549 cells were transfected as indicated, and invasion experiments were performed using an FBS gradient. (D) Immunoblots showing PREX2 and PTEN from cells used in invasion assays. Lysates from breast cancer cell lines with endogenous PTEN are shown. (E) C2-tail antagonizesPREX2-driven invasion in SUM149 cells. SUM149 cells were transfectedas indicated, and invasion experiments were performed. (F) Coimmunoprecipitation of PTEN constructs and PREX2. BT549 cells were transfected with V5-PREX2 along with the indicated FLAG-PTEN constructs. (G) Schematic showing the binding of the PTEN PDZ binding domain to the IP4P domain of PREX2 and the binding of the C2 domain of PTEN to the Dbl homology–pleckstrin homology (DHPH) domain of PREX2. Error bars in (A), (C), and (E) represent the mean ± SEM of three experiments. P values were calculated using two-tailed t tests. Immunoblots in (D) to (F) are representative of two experiments.

Fig. 4. PTEN inhibits PREX2 GEF activity

(A to C) PTEN suppresses RAC activation. HEK-293 cells were cotransfected with RAC and PREX2 plus either (A) WT PTEN, (B) G129E, or (C) C2-tail. Cells were starved overnight, and RAC activation was measured by pull-down of RAC-GTP. Immunoblots in (A) to (C) are representative of at least two experiments. (D) Coomassie stains of purified PREX2, PTEN, and RAC1 proteins. (E) Effect of PREX2 and PTEN on the kinetics of mant-GDP dissociation from RAC as measured by a fluorimeter. RAC, PTEN (G129E, C124S, and C2-tail), and PREX2 (WT and GEF-dead double mutant E30A N212A) were present in exchange reactions at concentrations of about 20, 10, and 5 nM, respectively. One thousand–fold molar excess of GTP was added to the reaction to initiate dissociation of mant-GDP from RAC. Lines represent the average of at least three independent runs, and the slopes of the linear phases of nucleotide exchange were calculated. Two-tailed Mann-Whitney nonparametric tests were used to compare slopes of nucleotide exchange experiments. au, arbitrary units.

Fig. 5. Cancer-derived PREX2 mutants evade inhibition by PTEN

(A) PREX2 mutation is associated with high PTEN expression in melanoma. mRNA z-scores for PTEN and PREX2 expression in melanomas with WT PREX2 (n =160) or with PREX2 mutations (n = 49). Whiskers extend from the 10th to 90th percentiles, and black dots denote values outside this range. Boxes delineate the first to third quartiles, and the central bar represents the median. P values were calculated using the Mann-Whitney test. Data were obtained from cBioPortal (13). Eighteen samples had PTEN mutations and were excluded. (B) PREX2 consists of DH, PH, DEP, PDZ, and IP4P domains. *, locations of cancer-derived PREX2 mutations used in this study. (C) The PTEN C2-tail is unable to inhibit invasion driven by PREX2 cancer mutants. BT549 cells were transfected as indicated, and invasion assays were performed. C2-tail and WT PREX2 are shown for comparison. Error bars represent the mean ± SEM of at least three experiments. (D) Coimmunoprecipitation of PREX2 mutants along with full-length PTEN or C2-tail. Immunoblots are representative of three experiments.

Fig. 6. PREX2 cancer mutants are resistant to PTEN C2-tail–mediated inhibition of RAC-GEF activity

(A to C) Effect of PTEN C2-tail on PREX2 mutant-induced mant-GDP dissociation from RAC. Nucleotide exchange catalyzed by a GEF-dead PREX2 double mutant (E30A N212A) is also shown. Each line represents the average of at least three experiments. Slopes of nucleotide exchange were quantified (fig. S6B). (D) Cancer-derived PREX2 mutants retain the ability to inhibit PTEN signaling activity. U87-MG cells were transfected as indicated, and lysates were collected from cells growing under steady-state conditions. Immunoblots were probed as indicated. Immunoblots are representative of two experiments.

Fig. 7. Model of PTEN-mediated inhibition of PREX2 GEF activity

(A) PTEN binds PREX2 and inhibits its GEF activity, blocking RAC1 activation and suppressing invasion. WT PREX2 inhibits PTEN activity and amplifies PI3K signaling. (B) Cancer-derived PREX2 mutants are resistant to inhibition by PTEN. PREX2 cancer mutants stimulate invasion and maintain the ability to antagonize PTEN activity and amplify PI3K signaling. Arrows and blunted lines denote activation and inhibition, respectively. Black and gray lines denote full and reduced activity, respectively. PREX2 is activated downstream of receptor tyrosine kinases (RTK) or G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptors (GPCRs).