Tuberous sclerosis complex (TSC) is a tumor suppressor gene syndrome characterized by seizures, mental retardation, autism, and tumors of the brain, kidney, heart, retina, and skin. TSC is caused by mutations in either TSC1 or TSC2, both of which are tumor suppressor genes. Hamartin, the protein product of TSC1, was found to interact with the ezrin-radixin-moesin family of cytoskeletal proteins and to activate the small GTPase Rho. To determine whether tuberin, the TSC2 product, can also activate Rho, we stably expressed full-length human tuberin in two cell types: MDCK cells and ELT3 cells. ELT3 cells lack endogenous tuberin expression. We found that expression of human tuberin in both MDCK and ELT3 cells was associated with an increase in the amount of Rho-GTP, but not in Rac1-GTP or cdc42-GTP. Tuberin expression increased cell adhesion in both cell types, and decreased chemotactic cell migration in ELT3 cells. In MDCK cells, there was a decrease in the amount of total Focal Adhesion Kinase (FAK) and an increase in the fraction of phosphorylated FAK. These findings demonstrate for the first time that tuberin activates Rho and regulates cell adhesion and migration. Pathways involving Rho activation may have relevance to the clinical manifestations of TSC, including pulmonary lymphangioleiomyomatosis.
Tuberous sclerosis complex (TSC) is a tumor suppressor gene syndrome characterized by seizures, mental retardation, autism, and tumors in the brain, kidney, heart, retina, and skin (Gomez et al., 1999). TSC is caused by germline mutations in either TSC1 (van Slegtenhorst et al., 1997) or TSC2 (European chromosome 16 Tuberous Sclerosis Consortium, 1993). TSC2 encodes tuberin, a 200 kD protein which has a domain of homology to rap1GTPase activating protein (GAP) near the carboxy terminus. Tuberin is highly conserved. The overall similarity between human, mouse, and Drosophila tuberin homologues is 32% (25% identity of amino acid residues). The GAP domain is particularly highly conserved, with 36% identity. Tuberin has been shown to possess GAP activity for rap1A (Wienecke et al., 1995) and rab5 (Xiao et al., 1997), and to function in at least three cellular pathways in mammalian cells: vesicular trafficking (Xiao et al., 1997), cell cycle regulation (Ito and Rubin, 1999; Miloloza et al., 2000; Potter et al., 2001; Soucek et al., 1997; Tapon et al., 2001), and steroid hormone function (Henry et al., 1998). The in vivo significance of these pathways and their contribution to the abnormal cell proliferation and differentiation observed in TSC are not yet completely understood. The protein product of the TSC1 gene, hamartin, was recently found to activate Rho, and to regulate focal adhesion and stress fiber formation via an interaction with the ezrin-radixin-moesin family of cytoskeletal proteins (Lamb et al., 2000). The clinical features of TSC1 and TSC2-linked disease are very similar (Dabora et al., 2001), and hamartin and tuberin are known to physically interact (Plank et al., 1998; van Slegtenhorst et al., 1998). This led us to ask whether tuberin, like hamartin, can activate Rho signaling pathways.
Development of MDCK and ELT3 cell lines with stable expression of tuberin
After retroviral transduction and neomycin selection, we successfully expressed full-length wild-type human tuberin in MDCK epithelial cells (Figure 1a), and tuberin-deficient ELT3 uterine leiomyoma cells (Figure 1b). Robust expression of tuberin was sustained over a period of at least 3 months. Vector control stable cell lines were developed in parallel. Upon expression of tuberin, we did not observe stabilization of hamartin in ELT3 cells (Figure 1c).
Tuberin expression is associated with Rho activation in MDCK and ELT3 cells
Previously, transient expression of hamartin, the product of the TSC1 gene, was shown to activate Rho in COS7 monkey kidney epithelial cells (Lamb et al., 2000). To determine whether expression of tuberin also activates Rho, we used a Rhotekin pulldown assay. Rhotekin binds the active, GTP-bound, form of Rho, but not the inactive, GDP-bound, form (Reid et al., 1996). Cells were serum starved for 16 h prior to the assay. The amount of active Rho was increased on average by threefold in MDCK cells expressing tuberin relative to the pMSCVneo vector control cell lines (Figure 2a). In ELT3 cells, the amount of active Rho was increased on average by 2.9-fold in the T3 and T9 cell lines expressing tuberin, compared to the vector control cell lines (Figure 2b). The amount of active Rho was also increased in the tuberin-expressing ELT3 cell line T4 (data not shown). Overall, the fold activation of Rho that we observed in cells expressing tuberin was similar to that observed with transient hamartin expression in COS7 cells (Lamb et al., 2000), where 2–3-fold activation was seen in a similar assay.
Tuberin expression is not associated with activation of Rac1 and cdc42
To determine whether other members of the Rho family were also activated, we studied Rac1 and cdc42 using a Pak1-p21 binding domain pull-down assay. We did not observe an increase in the amount of GTP-bound Rac1 or cdc42 in ELT3 or MDCK cells expressing tuberin (Figure 3).
Tuberin expression is associated with increased cell attachment
The attachment of cells to a matrix is regulated in part by Rho (Bobak et al., 1997). To determine whether expression of tuberin affects cell attachment, cells were trypsinized, allowed to re-attach, and the number of attached cells scored. In both MDCK and ELT3 cell types, all tuberin expressing cell lines showed increased attachment at 30 min, 1 and 2 h after replating, compared to vector cell lines. The maximum differential attachment of MDCK cells was observed at 30 min after replating (Figure 4a). At the 30 min timepoint the average attachment of tuberin-expressing MDCK cell lines was increased by 53% compared to the vector control cell lines. Analysis of variance showed that this difference was statistically significant (P<0.05). For ELT3 cells at 30 min after replating (Figure 4b), the average attachment of the tuberin expressing cell lines T3, T4 and T9 was increased by 49%, compared to the vector cell lines. This difference was also statistically significant (P<0.05).
Tuberin decreases chemotactic migration of ELT3 cells
To determine whether expression of tuberin affected the chemotactic migration of ELT3 cells, 5×104 cells were placed in the upper chamber of a modified Boyden chamber, and allowed to migrate for 24 h towards a 2% Fetal Bovine Serum (FBS) gradient. The migration of ELT3 cells expressing tuberin was reduced by 42% (P<0.05, Figure 5). MDCK cells, either with or without tuberin expression, showed no migration towards 2% FBS in modified Boyden chambers, and were not used for migration assays.
To exclude the possibility that proliferation differences contributed to the migration differences observed, we measured the proliferation of ELT3 cells at 24, 48 and 72 h (data not shown). Although differences in the proliferation of tuberin expressing cell lines compared to the vector cell lines were observed at 72 h, no statistically significant differences were observed at 24 h, which is the relevant timepoint for the migration assay.
Tuberin decreases total FAK and increases the proportion of phosphorylated FAK in MDCK cells
FAK, which is a known downstream effector of Rho through the Rho-associated protein serine/threonine kinase (Flinn and Ridley, 1996; Sinnett-Smith et al., 2001), is a member of the cytoplasmic complexes associated with focal adhesions (Sastry and Burridge, 2000). Previously, hamartin was shown to regulate focal adhesion formation (Lamb et al., 2000). To determine whether the increased attachment and increased Rho activation we observed in MDCK cells expressing tuberin was associated with changes in FAK, we first assessed the amount of FAK in the whole cell lysate. The amount of FAK was decreased by fivefold in the tuberin expressing cell lines T1 and T2, and by 2.5-fold in cell line T8, relative to the vector control cell lines P2 and P3 (Figure 6). To determine whether differences in the phosphorylation status of FAK were present, FAK was immunoprecipitated and immunoblotted with an anti-phosphotyrosine antibody. Consistent with the decrease in total FAK observed in the whole cell lysate, the amount of immunoprecipitated FAK was substantially less for all three tuberin expressing MDCK cell lines (Figure 7). However, the ratio of phosphorylated FAK to total FAK was 2.3–4.6-fold higher in cells with tuberin expression, compared to the vector control cells. Differences in total FAK and phosphorylated FAK were not found between the tuberin expressing and the vector cell lines in ELT3 cells (data not shown).
Hamartin, the product of the TSC1 tumor suppressor gene, was recently identified as an interactor with the ezrin-radixin-moesin family of cytoskeletal proteins and shown to activate the GTPase Rho and regulate focal adhesion and stress fiber formation (Lamb et al., 2000). The Rho family of GTPases (cdc42, Rac1 and Rho) regulates the cytokine-induced reorganization of the actin cytoskeleton. Dysregulation of signaling by Rho family members is believed to play a critical role in cancer cell migration, invasion, and metastasis (Clark et al., 2000; Evers et al., 2000; Royal et al., 2000; Schmitz et al., 2000). Because the products of the TSC1 and TSC2 genes are known to physically interact, we hypothesized that tuberin, the product of the TSC2 gene, would also activate Rho.
To address this, we first tested three separate MDCK cell lines with stable expression of wild-type human tuberin. Increased Rho-GTP was present in these cells after 16 h of serum starvation, demonstrating that expression of tuberin, like hamartin, is associated with Rho activation. Activation of Rac1 or cdc42 was not observed. The expression of tuberin in MDCK cells was also associated with increased attachment to plastic, compared to vector control cells, with maximum attachment differences at 30 min following trypsinization and replating.
To determine whether similar changes occurred when tuberin was re-introduced to cells lacking endogenous tuberin, we stably expressed tuberin in ELT3 cells. The ELT3 cell line was derived from an Eker rat uterine leiomyoma (Howe et al., 1995). The Eker rat model of TSC2 (Kobayashi et al., 1995; Yeung et al., 1994) carries a germline inactivating mutation in one allele of the TSC2 gene, and develops renal carcinomas and uterine leiomyomas with an autosomal dominant pattern of inheritance. The uterine leiomyomas, as well as the renal tumors, contain somatic ‘second hit’ mutations in the remaining wild-type TSC2 allele (Yeung et al., 1995).
Similar to our findings in the MDCK cells, we observed a 2.9-fold increase in GTP-Rho in ELT3 cells expressing tuberin after 16 h of serum starvation. The Rho activation appears to be specific, since no activation of two other Rho family members, Rac1 and cdc42, was observed. The attachment of tuberin expressing ELT3 cells was increased compared to the vector control cell lines. The chemotactic migration of ELT3 cell lines with tuberin expression toward a 2% serum gradient was decreased by 24–49%, compared to the vector control cell lines. Since a strong correlation between Rho activation, migration, and attachment could not be established in all tuberin-expressing clones, this may indicate that these pathways are separable, or, alternatively, reflect saturation of key downstream effectors. Our migration data contrast with another recent study in which tuberin was found to have little or no effect on cell migration induced by platelet-derived growth factor in Eker rat-derived cells (Irani et al., 2002). Taken together with our data, this may indicate that tuberin regulates cell migration independent of the phosphatidylinositol 3-kinase pathway.
In addition to the changes in Rho-activation, attachment, and migration, we found that expression of tuberin in MDCK cells was associated with a steady-state decrease in total FAK and an increase in the proportion of phosphorylated FAK. These changes were not observed in ELT3 cells expressing tuberin, possibly reflecting their smooth muscle origin as opposed to the epithelial origin of the MDCK cells. FAK is a non-receptor tyrosine kinase that integrates growth factor and integrin signaling and promotes cell migration (Ilic et al., 1997; Sieg et al., 2000). FAK is a known downstream effector of Rho-associated kinase (Sinnett-Smith et al., 2001; Timpson et al., 2001). We hypothesize that Rho activation downstream of TSC2 increases the phosphorylated fraction of FAK, and that the decrease in total FAK that we observed is a secondary compensatory change. Interestingly, increased levels of total FAK have been found in many human epithelial cancers (Agochiya et al., 1999; Cance et al., 2000; Judson et al., 1999) and a correlation between increased FAK expression and tumor invasion and metastasis has been observed in a variety of human tumors (Cance et al., 2000; Kornberg, 1998; Weiner et al., 1993).
Taken together, these data implicate tuberin in pathways involving Rho activation, cell attachment and migration, and FAK phosphorylation. These pathways are likely to be relevant to the clinical manifestations of TSC and in particular to pulmonary lymphangioleiomyomatosis (LAM), an often-fatal disease affecting almost exclusively women. LAM is characterized by a diffuse proliferation of abnormal smooth muscle cells in the lungs (Sullivan, 1998). LAM occurs in women with TSC and also sporadically (sporadic LAM). Previously, we found mutational inactivation of both copies of the TSC2 tumor suppressor gene in LAM cells and renal angiomyolipoma cells from women with sporadic LAM (Carsillo et al., 2000; Yu et al., 2001). These mutations were not present in normal kidney, normal lung, or blood lymphocytes, suggesting the possibility of a highly unusual disease mechanism in LAM: the migration of histologically benign smooth muscle cells from the kidney to the lung. We hypothesize that abnormal Rho activation, migration, and attachment of smooth muscle cells lacking tuberin is central to the pathogenesis of LAM.
The individual roles of hamartin and tuberin in activating Rho are not yet understood. Hamartin stabilizes tuberin by protecting tuberin from ubiquitin-mediated degradation, and tuberin can also stabilize hamartin, but to a substantially lesser extent (Benvenuto et al., 2000). No increase in hamartin expression was observed in our ELT3 cells expressing tuberin. We speculate that tuberin could be the key effector in Rho activation, and the previously observed Rho activation by hamartin could have resulted from stabilization of tuberin by hamartin. Consistent with this model, it has been reported that hamartin activates Rho via amino acids 145–510 (Lamb et al., 2000), a region that overlaps with hamartin's tuberin-interaction domain (amino acids 302–430) (Hodges et al., 2001; Nellist et al., 2001). Further studies will be required to address whether tuberin can activate Rho independently of hamartin.
In summary, we have shown for the first time that expression of tuberin, the product of the TSC2 tumor suppressor gene, is associated with activation of Rho, both in MDCK cells with endogenous tuberin expression and in ELT3 cells that lack endogenous tuberin. MDCK cells expressing tuberin had increased cell adhesion and a relative increase in the amount of phosphorylated FAK, and ELT3 cells expressing tuberin had increased cell attachment and decreased cell migration. Increased cell attachment and decreased cell migration are consistent with activation of pathways downstream of Rho.
Materials and methods
Madin-Darby canine kidney epithelial cells MDCK-1 (American Type Culture Collection, Manassas, VA, USA) were cultured in low-glucose DMEM medium, supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM L-glutamine and 10% FBS. GP-293 packaging cells (Clontech, Palo Alto, CA, USA) were cultured in regular DMEM medium supplemented as above. Tuberin-deficient ELT3 cells, derived from a uterine leiomyoma of Eker rats (Howe et al., 1995), were cultured in IIA complete medium (50% DMEM, 50% F-12, 1.2 g/ml NaHCO3, 1.6 μM FeSO4, 50 nM sodium selenite, 25 μg/ml insulin, 200 nM hydrocortisone, 10 μg/ml transferrin, 1 nM triiodothyronine, 10 μU/ml vasopressin, 10 nM cholesterol, 10 ng/ml epidermal growth factor) containing 15% FBS.
Generation of tuberin-expressing cell lines
GP-293 cells were co-transfected with 2 μg of the retroviral vector pMSCVneo (Clontech), 1 μg pVSV-G (Clontech) encoding for the viral glycoprotein, and 6 μl FuGene6 (Roche, Indianapolis, IN, USA). Replication-deficient retroviruses were collected from the culture after 72 h and applied to subconfluent target cells (MDCK and ELT3) in the presence of 8 μg/ml polybrene (Sigma, St. Louis, MO, USA). Cells were transduced with empty pMSCVneo vector as a control, or a pMSCVneo construct containing the coding region of the human TSC2 gene, which was sequence confirmed. Stable clones were selected for 2 weeks in the presence of 200 μg/ml or 300 μg/ml G418 (Life Technologies, Rockville, MD, USA) for MDCK and ELT3 cells, respectively.
Western blot analysis
Cells were lysed in RIPA buffer [1×PBS, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM sodium orthovanadate, 10 μg/ml PMSF, 3 : 100 aprotinin (Sigma)] unless otherwise specified. Protein concentration was determined by the Bradford assay (Biorad, Hercules, CA, USA) at 595 nm. Proteins were denatured by boiling in Laemmli sample buffer (Biorad) and resolved in Tris-glycine polyacrylamide gels (Biorad). The proteins were transferred on Immobilon P membrane (Millipore, Bedford, MA, USA) in 25 mM Tris, 192 mM glycine, 10% methanol, at 20 V for 16 h. Five per cent BSA in PBST (1×PBS, 0.05% Tween 20) was used as a blocking agent for anti-phosphotyrosine and anti-Rho antibodies. Five per cent non-fat dry milk in PBST was used for blocking for anti-tuberin, anti-beta-tubulin and anti-FAK antibodies. Five per cent non-fat dry milk/1% BSA in PBST was used as blocking agent for anti-hamartin.
Rabbit polyclonal anti-tuberin C20 and anti-cdc42 (Santa Cruz, Santa Cruz, CA, USA), and mouse monoclonal anti-beta-actin (Sigma), anti-Rho (Santa Cruz), anti-Rac1 (Upstate Biotechnology, Charlottesville, VA, USA), anti-FAK (BD Biosciences, San Jose, CA, USA) and anti-phosphotyrosine clone 4G10 (Upstate Biotechnology) were used for Western blot analysis. Rabbit polyclonal anti-hamartin antibody was previously described (Plank et al., 1998). HRP-conjugated anti-rabbit IgG and anti-mouse IgG, and Enhanced Chemiluminescence reagents were purchased from Amersham (Piscataway, NJ, USA).
Rhotekin Rho binding domain and Pak1-p21 binding domain pull-down assays
Cells were plated in the presence of serum for 24 h, rinsed with PBS, and refed with serum-free medium. After 16 h of serum starvation (70% confluency), the cells were lysed.
For Rhotekin pull-down, cells were lysed in 50 mM Tris pH 7.2, 500 mM NaCl, 10 mM MgCl2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, and 1 : 100 Protease Inhibitor Cocktail II (Sigma). Lysates were passed through a 21G needle and clarified by centrifugation at 14 000 r.p.m. for 20 min. Fifty μl GST-fused Rhotekin Rho Binding Domain bound to glutathione-agarose beads (Upstate Biotechnology) was incubated with 750 μg of protein lysate in a final volume of 500 μl. The beads were washed three times in 50 mM Tris pH 7.2, 1% Triton X-100, 150 mM NaCl, 10 mM MgCl2, 0.1 mM PMSF.
For Pak1-p21 binding domain pull-down, cells were lysed in 25 mM HEPES pH=7.5, 150 mM NaCl, 1% Igepal CA-630, 0.25% sodium deoxycholate, 10% glycerol, 25 mM NaF, 10 mM MgCl2, 1 mM EDTA, 1 mM sodium orthovanadate, 1 : 100 Protease Inhibitor Cocktail II (Sigma). Lysates were passed through a 21G needle and clarified by centrifugation at 14 000 r.p.m. for 20 min. Ten μg recombinant Pak1-p21 binding domain-GST agarose beads (Upstate Biotechnology) was incubated with 750 μg of protein lysate in a final volume of 750 μl. The beads were washed three times in lysis buffer.
Bound proteins were eluted in Laemmli sample buffer, and analysed by SDS–PAGE and Western blotting.
Cell attachment assays were performed using standard techniques (Bonifacino et al., 2001). Briefly, 2.5×104 cells were plated in triplicate wells of 96-well tissue culture plates and allowed to attach for 30 min, 1, 2 and 6 h, in a 37°C humidified CO2 incubator. Non-attached cells were removed by gentle rinsing with phosphate-buffered saline (PBS), and the remaining attached cells were fixed in 5% glutaraldehyde, washed three times with distilled water, and stained with crystal violet. Cells were washed in distilled water, and crystal violet was released from the cells in 10% acetic acid. Absorbance was measured at 570 nm. Attachment at the given time points was reported relative to 100% attachment of cells at 6 h.
5×104 ELT3 cells were resuspended in serum-free IIA complete medium and added to the upper compartment of a 6.5 mm diameter, 12 μm pore size Transwell insert (Costar, Corning, NY, USA). Two per cent FBS in IIA complete medium was added to the lower chamber of the Transwell insert, and the cells were allowed to migrate for 24 h in a 37°C humidified CO2 incubator. The cells that remained on the upper surface of the Transwell were removed with a cotton swab. The cells that migrated to the lower surface of the Transwell were stained with DiffQuik (Dade, Deerfield, IL, USA). The number of migrating cells was determined by direct counting under the microscope. Each experiment was performed in triplicate.
FAK phosphorylation assay
Cells at 50% confluency were lysed in PTY buffer [50 mM HEPES, 50 mM NaCl, 5 mM EDTA, 1% Triton X-100, 50 mM NaF, 10 mM Na4P2O7, 1 mM Sodium Orthovanadate, 10 μg/ml PMSF, 3 : 100 aprotinin (Sigma), 1 : 100 phosphatase inhibitor cocktail II (Sigma)]. Two μg mouse monoclonal anti-FAK antibody (BD Biosciences) was added to 500 μg of protein lysate and rotated at 4°C for 4 h. One hundred μl of protein A-agarose beads (Invitrogen, Carlsbad, CA, USA) washed in PTY buffer was added and incubated at 4°C for 16 h under constant rotation. The beads were washed in PTY buffer and an equal volume of Laemmli sample buffer (Biorad) was added. FAK was analyzed by Western blotting.
Computational and statistical analysis
The two-tailed t-test for paired samples and analysis of variance was used to calculate P values between controls and subjects for attachment and migration assays. The NIH Image or Scion Image (Scion Corp, Frederick, MD, USA) software was used for densitometry of scanned immunoblots.
We are grateful to Drs Erica Golemis and Joseph Testa for critical review of this manuscript. This work was supported by NIH grants HL 60746 and DK 51052, and the LAM Foundation (Cincinnati, OH, USA).