TGF-b1 up-regulates paxillin protein expression in malignant ... - Nature

ã

Oncogene (2001) 20, 7976 ± 7986 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

TGF-b1 up-regulates paxillin protein expression in malignant astrocytoma cells: requirement for a ®bronectin substrate Xiaosi Han1, Jerry E Stewart Jr1, Susan L Bellis2, Etty N Benveniste3, Qiang Ding1, Kouichi Tachibana4, J Robert Grammer1 and Candece L Gladson*,1 1

The Department of Pathology, Division of Neuropathology, The University of Alabama at Birmingham, Birmingham, Alabama AL 35294, USA; 2The Department of Physiology and Biophysics, The University of Alabama at Birmingham, Birmingham, Alabama AL 35294, USA; 3The Department of Cell Biology, The University of Alabama at Birmingham, Birmingham, Alabama AL 35294, USA; 4The Department of Cancer Immunology and AIDS, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, MA 02115, USA

Cytokines can in¯uence the interactions between members of the integrin cell adhesion receptor family and the extracellular matrix thereby potentially a€ecting cell function and promoting cell adhesion, growth and migration of malignant astrocytoma tumor cells. As malignant astrocytoma cells synthesize TGF-b1 in vivo, we analysed the e€ects of TGF-b1 on signaling events associated with integrin receptor ligation, focusing on the e€ects on paxillin, a phosphorylated adaptor protein, that acts as a sca€old for signaling molecules recruited to focal adhesions. TGF-b1-stimulation of primary astrocytes and serum-starved U-251MG malignant astrocytoma cells attached to ®bronectin induced a substantial increase in the levels of paxillin protein (®vefold increase at 2.0 ng/ml) in a dose- and timedependent manner compared to the levels observed on plating onto ®bronectin in the absence of stimulation. In the astrocytoma cells, this resulted in an increase in the pool of tyrosine-phosphorylated paxillin, although it did not appear to alter the extent of phosphorylation of the paxillin molecules. In contrast, in primary astrocytes the protein levels were upregulated in the absence of a parallel increase in phosphorylation. The TGF-b1stimulated increase in paxillin levels required ligation of the ®bronectin receptor, as it was not induced when the cells were plated onto vitronectin, collagen or laminin. The increase in the pool of paxillin on TGFb1 stimulation of the ®bronectin-plated astrocytoma cells was associated with an increase in translation, but was not associated with an increase in the steady-state levels of paxillin mRNA. Stimulation with TGF-b1 on a ®bronectin substrate increased subsequent attachment and spreading of U-251MG cells onto ®bronectin and, to a lesser extent, vitronectin, but not collagen. Our results indicate that physiologic levels of TGF-b1 stimulate the expression of paxillin protein at the level of translation through a process that requires engagement of the

*Correspondence: CL Gladson, University of Alabama at Birmingham, LHRB 567, 701 S. 19th Street, Birmingham, AL 35294; E-mail: [email protected] Received 27 April 2001; revised 11 September 2001; accepted 18 September 2001

®bronectin receptor, and promotes attachment and spreading of malignant astrocytoma cells on ®bronectin.

Oncogene (2001) 20, 7976 ± 7986.

Keywords: paxillin; TGF-b1; ®bronectin; integrin a5b1; astrocytoma; astrocytes; glioma; glioblastoma

Introduction Members of the integrin cell adhesion receptor family provide a link between the extracellular environment and the cytoplasm (Giancotti and Ruoslahti, 1999), which contribute to the regulation of cell proliferation, migration and gene transcription (Giancotti and Ruoslahti, 1999; Gladson, 1999). The interactions of the integrin receptors with the extracellular matrix can be modi®ed by cytokines, including TGF-b1, as well as platelet-derived growth factor, and angiotensin II (Turner, 1998; Giancotti and Ruoslahti, 1999). Although the e€ects of cytokine regulation on integrin interactions and signaling are of particular interest in tumor cells, they have not been well characterized in astrocytoma cells. Integrin receptor attachment to extracellular matrix proteins initiates a cascade of signaling events in a dynamic process that leads to the recruitment of multiple cytoskeletal and regulatory proteins to the cytoplasmic face of cell attachment sites (focal adhesions). The formation of focal adhesions in ®broblasts requires changes in the actin-based cytoskeleton that occur through an ordered hierarchial activation of the Rho family of small GTPases, in which activation of Cdc42 progresses through Rac to Rho (Hall, 1994). The proteins that localize to the focal adhesions include the adaptor proteins, paxillin and p130CAS; the cytoskeletal proteins, vinculin and a-actinin; and the regulatory proteins, focal adhesion kinase (FAK) and Src (Giancotti and Ruoslahti, 1999). Paxillin plays a key role in this process, acting as a sca€old for signaling molecules at focal adhesions. It is a 66 kDa adaptor protein that contains multiple domains, including a proline-rich

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

7977

Figure 1 TGF-b1 Stimulation increases paxillin protein expression in a dose-dependent manner in astrocytic cells. Serum-starved U-251MG cells or primary rat astrocytes were plated onto ®bronectin in serum-free media for 5 h, plus/minus TGF-b1 stimulation (0 ± 18 ng/ml), followed by 1% NP40 lysis, SDS ± PAGE of equivalent microgram of lysate and Western blot analysis with mAb anti-paxillin (a,c,h). The membranes were stripped and reprobed with mAb anti-actin (b,i). (c) U-251MG cells were transiently transfected with the Hic-5 cDNA (lane 1) or vector control (lane 2), plated onto ®bronectin for 5 h and then detergent lysed and Western blotted with mAb anti-paxillin. (d,e,f,g) U-251MG cells were plated as above, followed by lysis with RIPA bu€er with the addition of 1% SDS, SDS ± PAGE of equivalent migrogram of lysate and Western blot analysis with mAb anti-paxillin (d), or rabbit anti-a5 IgG (f) stripping and re-probing with mAb anti-actin (e,g)

consensus SH3 binding site, four cysteine/histidine rich motifs of 51 amino acids (LIM domains), ®ve leucine rich domains (LD motifs), several potential tyrosine phosphorylation sites, and other sequences that are involved in protein ± protein interactions (reviewed in Turner, 1998). Integrin receptor ligation results in integrin clustering in the cell membrane and in a temporally related manner the activation of FAK (Giancotti and Ruoslahti, 1999). Following the activation of FAK,

Src kinase associates with FAK resulting in a Src/FAK complex that phosphorylates several substrates, including paxillin. Current data suggests that integrin receptor signaling events require integrin receptor localization to focal adhesions. Therefore, cytokines that promote integrin receptor organization of the cytoskeleton could enhance signaling events occurring at focal adhesions. TGF-b1 has been reported to up-regulate the expression of Hic-5 mRNA, a paxillin family member Oncogene

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

7978

Table 1 Time course of TGF-b1 stimulation of paxillin protein expression Time (h) TGF-b1 (2.0 ng/ml) Fibronectin Substrate Vitronectin Substrate

0 7 1 1

1.5 + 1 1

7 2

+ 3.5 NP

5 7 0.8 1.9

24 + 7 1.5

7 1.6 1

+ 3.5 1

Paxillin protein level is expressed as a ratio of paxillin to actin band intensity, based on densitometric analysis. NP, denotes not performed

(Shibanuma et al., 1994). Moreover, it has been reported that TGF-b1 is synthesized by some tumors, such as malignant astrocytomas (Kjellman et al., 2000). As our laboratory has been studying the integrin-mediated signaling pathways that promote malignant astrocytoma cell invasion (Gladson, 1999; Wang et al., 2000), we investigated the regulation of paxillin by TGF-b1 in malignant astrocytoma cells. Here, we report that TGF-b1 stimulation in cooperation with the engagement of the a5b1 ®bronectin receptor integrin increases paxillin protein expression signi®cantly and enhances cell attachment and spreading.

Results TGF-b1 stimulation up-regulates paxillin protein expression in astrocytic cells adherent to fibronectin To estimate the expression of paxillin, we prepared cell extracts from U-251MG human malignant astrocytoma cells that had been allowed to attach to ®bronectin in serum-free media for 5 h in the presence of increasing concentrations of TGF-b1 (0 ± 18 ng/ml) and analysed the lysates by Western blot analysis using a monoclonal antibody to paxillin. As anticipated, the monoclonal anti-paxillin antibody revealed a doublet migrating with relative molecular masses of 64 and 66 kDa (Figure 1a). We found that the level of paxillin protein was up-regulated upon TGF-û1 stimulation with a maximal response occurring after stimulation for 5 h with 2.0 ng/ml TGF-b1 (Table 1). These conditions (2.0 ng/ml, 5 h) were used in all the subsequent studies of TGF-b1 stimulation of U-251MG cells reported here. As paxillin is a member of the family of homologous proteins that contains Hic-5, we investigated the possibility that the mAb anti-paxillin recognizes Hic-5. Hic-5 cDNA was expressed transiently in U-251MG cells and cell lysates analysed by Western blot analysis using the antipaxillin monoclonal antibody (Figure 1c). In these lysates, the monoclonal antibody did react with protein with a relative molecular mass of 50 kDa, consistent with the migration pattern of Hic-5 (Figure 1c). However, in all subsequent experiments of astrocytoma cells, the anti-paxillin monoclonal antibody revealed proteins with molecular weights Oncogene

consistent with those of paxillin and proteins that migrated faster than 64 kDa, as would Hic-5 protein, was not detected. The observed increase in the levels of paxillin protein in the cell lysate could re¯ect an alteration in the detergent solubilization of a speci®c cellular compartment. To address this issue, U-251MG cells plated onto ®bronectin in serum-free media, in the presence and absence of TGF-b1 stimulation (2.0 ng/ ml) were lysed directly in RIPA bu€er with the addition of 1% SDS prior to Western blot analysis with mAb anti-paxillin (Figure 1d). A substantial increase in paxillin protein expression (&fourfold) was observed in the TGF-b1-stimulated cells (Figure 1d, lanes 1 and 2). Stripping and reprobing of the membrane with mAb anti-actin demonstrated no change in actin expression in the absence or presence of TGF-b1 stimulation (Figure 1e, lanes 1 and 2), indicating that, when normalized to actin, the increase in paxillin expression with TGF-b1 stimulation was similar to that seen when the cells were lysed in 1% NP40. These experiments demonstrate that the increase in paxillin with TGF-b1 stimulation in cells plated onto ®bronectin is not due to an alteration in the detergent solubilization of a speci®c cellular compartment. As an additional control, we Western blotted the same cell lysates with rabbit anti-a5 IgG directed toward the a5 integrin subunit that pairs with b1 to form the ®bronectin receptor. We found a band migrating at 160 kDa plus/minus TGF-b1 stimulation that was of equivalent intensity (Figure 1f, lanes 1 and 2). This indicates that the total cellular level of the a5b1 integrin does not change with TGF-b1 stimulation. To determine whether TGF-b1 stimulation increases paxillin expression in normal astrocytes, primary rat astrocytes were plated onto ®bronectin in serum-free media and stimulated with TGF-b1 for 5 h. Western blot analysis of the cell extracts indicated that 6.0 ng/ ml TGF-b1 stimulation resulted in a maximal increase in paxillin protein expression in the normal astrocytes (Figure 1h), similar to that described for the malignant astrocytic cells. Stripping and re-probing with mAb anti-actin con®rmed similar protein loading in all lanes (Figure 1i). TGF-b1 up-regulation of paxillin protein expression in astrocytic cells occurs in a matrix-dependent manner We have reported previously that U-251MG cells can attach to various extracellular matrix substrates and that di€erent integrin receptors mediate these interactions. U-251MG cells utilize integrin a5b1 in attachment to ®bronectin, integrins a2b1 and a3b1 in attachment to collagen, and integrin a6b1 in attachment to laminin (Pijuan-Thompson and Gladson, 1997). To determine if the stimulatory e€ect of TGF-b1 on paxillin protein expression is dependent on the matrix interactions, U-251MG cells were allowed to attach to vitronectin (Figure 2a,b),

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

collagen (Figure 2c,d), or laminin (Figure 2e,f) for 5 h in the presence of a range of concentrations of TGF-b1. In contrast to the results obtained with ®bronectin, the levels of paxillin protein in U-251MG cells adherent to vitronectin, collagen or laminin remained unaltered on TGF-b1 stimulation (Figure

2a,c,e, respectively). Stripping and re-probing of the membranes with mAb anti-actin con®rmed equivalent protein loading (Figure 2b,d,f). To rule out the possibility that TGF-b1 stimulation of U-251MG cells adherent to vitronectin, collagen or laminin failed to increase paxillin expression due to contam-

7979

Figure 2 TGF-b1 stimulation of paxillin protein expression in astrocytoma cells is speci®c for cells adherent to ®bronectin. Serumstarved U-251MG cells were plated onto vitronectin, collagen type IV, laminin, or ®bronectin under the conditions in Figure 1 and analysed as in Figure 1. (a,c,e,g) Western blot analysis with mAb anti-paxillin. (b,d,f,h) The membrane was stripped and reprobed with mAb anti-actin Oncogene

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

7980

ination of these puri®ed matrix proteins with TGFb1, an ELISA assay for TGF-b1 was performed. We found undetectable quantities of TGF-b1 in the vitronectin, collagen and laminin preparations, and 9.0 pg/ml of TGF-b1 in the ®bronectin preparation (data not shown). These data suggest that TGF-b1 upregulates paxillin during interaction of malignant astrocytomas with ®bronectin speci®cally. This speci®city for ®bronectin suggests that a cooperation between the integrin receptor, a5b1, and TGF-b1 is most likely necessary for TGF-b1 stimulation of paxillin protein expression in these cells. To determine whether the increase in paxillin protein expression could be due solely to engagement of the a5b1 ®bronectin receptor integrin, U-251MG cells were plated on ®bronectin, vitronectin or laminin in serum-free media for 5 h, lysed in 1% NP40 bu€er, and subjected to Western blot analysis with mAb anti-paxillin. We found that the expression of paxillin protein in cells was similar in cells adherent to ®bronectin, vitronectin, and laminin (Figure 2g). Stripping and reprobing with mAb anti-actin demonstrated similar protein loading in all lanes (Figure 2h). Thus, the increase in paxillin protein in cells adherent to ®bronectin with TGF-b1 stimulation is not due solely to engagement of the a5b1 ®bronectin receptor. To rule out the possibility that TGF-b1 stimulation of U-251MG cells adherent to ®bronectin increased cell surface expression of integrin a5b1, ¯uorescent activated cell sorter (FACS) analysis was performed. We found no increase in integrin a5b1, avb3, or avb5 expression after plating the cells on ®bronectin, plus/minus TGF-b1 stimulation, for 5 h (data not shown), indicating an altered expression of integrin a5b1 does not account for the increased paxillin expression observed with TGF-b1 stimulation. TGF-b1 stimulation of astrocytic cells attached to fibronectin does not increase tyrosine phosphorylation of paxillin The phosphorylation status of paxillin plays an important role in determining its interactions with the various proteins that interact with this adaptor protein. Therefore, we determined whether TGF-b1 stimulation increased the extent of tyrosine phosphorylation of paxillin in the astrocytic cells. U-251MG cells and primary rat astrocytes were allowed to attach to ®bronectin in serum-free media for 5 h in the presence of increasing concentrations of TGF-b1, detergent lysed, and equivalent amounts of lysate from each sample immunoprecipitated with mAb anti-paxillin, followed by SDS ± PAGE, and Western blot analysis with rabbit anti-phosphotyrosine. Subsequently, the membrane was stripped and re-probed with mAb anti-paxillin. The IgG used for immunoprecipitation was revealed by the secondary antibody used in the Western blots as a band at 55 kDa. As anticipated, expression of paxillin protein was increased in both U-251MG cells and primary astro-

Oncogene

cytes upon TGF-b1 stimulation (Figure 3b,d, respectively), consistent with the previous Western blot analyses of paxillin protein in U-251MG cells (Figure 1a) and in primary astrocytes (Figure 1h). In U-251MG cells, a TGF-b1-stimulated increase in tyrosine phosphorylation was observed (Figure 3a), which re¯ected the increase in paxillin protein (Figure 3b). In the primary rat astrocytes, TGF-b1 stimulation increased paxillin protein expression (Figure 3d), but in these cells an equivalent increase in the level of tyrosine phosphorylation of paxillin was not observed (Figure 3c). These data indicate that TGFb1 stimulation increases the pool of tyrosine phosphorylated paxillin in the malignant astrocytic cells attached to ®bronectin, but does not increase the extent of tyrosine phosphorylation of the paxillin in either the malignant astrocytic cells or normal astrocytes attached to ®bronectin.

TGF-b1 stimulation increases paxillin protein translation, but fails to alter the steady-state levels of paxillin mRNA We then investigated the mechanism(s) by which TGF-b1 stimulation increases the expression of paxillin protein. Northern blot analysis indicated that the level of paxillin mRNA increased by 30% on stimulation with TGF-b1 (2.0 ng/ml) (Figure 4a, lanes 1 and 2). Repeats of this experiment showed a 15% variability in the increase in paxillin message after TGF-b1 stimulation, as assessed by Northern blot analysis. This suggests that TGF-b1 does not signi®cantly alter paxillin transcription under these conditions. Ethidium-bromide staining of the gel indicated an equivalent loading of ribosomal RNA in both lanes (Figure 4b). To determine whether TGF-b1 stimulation altered the half-life of paxillin protein, U-251MG cells plated onto ®bronectin, in the presence or absence of TGFb1 stimulation, were pulsed with [35S]Methionine/ cysteine, and then chased with cold media. Equivalent TCA-precipitable counts were immunoprecipitated with mAb anti-paxillin, and subjected to SDS ± PAGE, followed by autoradiography (Figure 4c). At time 0, the intensity of the 35S-labeled paxillin band was fourfold greater in cells stimulated for 5 h with TGF-b1, than in cells grown in the absence of TGFb1 stimulation (Figure 4c), which is consistent with TGF-b1 stimulation of paxillin translation. At subsequent time points between 4 and 24 h, the intensity of the 35S-labeled paxillin band in the TGFb1-treated cells was increased slightly as compared to that of the non-treated cells. The 35S-labeled paxillin band decreased signi®cantly in intensity between 8 and 24 h, in both treated and untreated cells, indicating that the half-life of paxillin protein is between 8 and 24 h in both cases. Taken together, our data suggest that TGF-b1 up-regulation of paxillin protein in U-251MG cells is associated with an increase in paxillin translation, but is not associated with altered paxillin transcription.

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

7981

Figure 3 TGF-b1 stimulation fails to increase paxillin phosphorylation on tyrosine residues. Serum-starved U-251MG cells (a,b) and primary rat astrocytes (c,d) were plated onto ®bronectin, plus/minus TGF-b1 stimulation (0 ± 18 ng/ml), in serum-free media for 5 h, followed by 1% NP40 lysis, immunoprecipitation of equivalent micrograms of lysate with mAb anti-paxillin, SDS ± PAGE of the immunoprecipitates, and transfer to Immobilon. Subsequently, the membranes were probed with rabbit anti-phosphotyrosine (a,c), stripped and reprobed with mAb anti-paxillin (b,d). The IgG used for immunoprecipitation was revealed as a band at 55 kDa by the secondary antibody used in the Western blots

TGF-b1 stimulation promotes malignant astrocytoma cell attachment and spreading

TGF-b1 stimulation promotes extension of filopodia and lamellipodia in cells plated onto fibronectin

To determine whether the increase in paxillin protein with TGF-b1 stimulation altered cell adhesion properties, cell attachment assays were performed. TGF-b1 stimulation of U-251MG cells adherent to ®bronectin, followed by replating of the cells onto ®bronectin or vitronectin, increased cell attachment by 40 and 20%, respectively, but failed to alter cell attachment to collagen (Figure 5a). Furthermore, the U-251MG cells replated onto ®bronectin or vitronectin for 30 min after TGF-b1 treatment showed a twofold and a 1.2fold increase in cell spreading, respectively, as compared to untreated cells (Figure 5b). At a later time point, 1 h after replating onto ®bronectin or vitronectin, no di€erence in cell spreading was detected between cells previously stimulated with TGF-b1 or not. These data suggest that paxillin promotes adhesion and more rapid spreading of malignant astrocytoma cells to some ligands.

As paxillin recruits a complex of proteins, including small GTPases, to the plasma membrane that mediate reorganization of the actin-based cytoskeleton, we investigated the e€ect of TGF-b1 on cytoskeletal organization. No change was detected by immuno¯uorescent analysis in the morphology of the U251MG cells, stress ®ber formation, or the localization of a5b1 and paxillin to focal adhesions after 5 h of attachment to ®bronectin independent of TGF-b1 stimulation (data not shown). However, in cells plated onto ®bronectin plus TGF-b1 stimulation for 5 h, followed by cell harvesting with bu€ered EDTA and replating onto ®bronectin for 30 min, paxillin localized to the plasma membrane in ®lopodia and lamellipodia (Figure 6b), phalloidin staining demonstrated polymerized actin in the cytoplasmic extensions (Figure 6d), and integrin a5b1 and Crk colocalized to the plasma membrane in ®lopodia and Oncogene

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

7982

Figure 4 TGF-b1 stimulation fails to alter the steady-state level of paxillin mRNA, but increases protein translation. (a,b) Serumstarved U-251MG cells were plated onto ®bronectin in serum-free media for 5 h, plus/minus 2.0 ng/ml TGF-b1 stimulation, followed by extraction of total RNA, and Northern blot analysis with [32P]Paxillin cDNA (a). The ethidium bromide stained gel is shown in b. (c) Paxillin half-life in the U-251MG cells adherent to ®bronectin plus/minus TGF-b1 stimulation was estimated by 35Smethione-cysteine labeling, followed by a cold chase, and immunoprecipitation with mAb anti-paxillin, as described in the Materials and methods

lamellipodia (Figure 6f). In contrast, in unstimulated U-251MG cells, paxillin remained localized to the cytoplasm, along with integrin a5b1 and Crk (Figure 6a,e, respectively), and cortical actin was not detected (Figure 6c). Discussion In this report, we characterized the ability of TGF-b1 to regulate paxillin protein expression in malignant astrocytoma cells. We found that physiologic levels of TGF-b1 up-regulate the expression of paxillin protein in malignant astrocytoma cells and in primary astrocytes, but did not increase paxillin phosphorylation on tyrosine residues. Furthermore, we demonstrate that TGF-b1 cooperates with the a5b1 ®bronectin receptor integrin in up-regulating paxillin protein, this regulation occurs largely at the level of translation, and results in increased cell attachment and spreading on some ligands. We show that TGF-b1 up-regulates paxillin protein in malignant astrocytoma cells in a dose- and timedependent manner. This ®nding is consistent with the previous report that TGF-b1 up-regulates the expression of Hic-5 mRNA, a paxillin family member expressed in ®broblasts (Shibanuma et al., 1994). Hic5 has signi®cant homology with paxillin and the monoclonal antibody used in these studies reacts with this molecule, but Hic-5 was not detected in the primary astrocytes and malignant astrocytoma cells used in this study under any of the experimental conditions. This is Oncogene

Figure 5 TGF-b1 stimulation promotes U-251MG cell attachment and spreading on ®bronectin and vitronectin. (a) 35Smethionine/cysteine labeled U-251MG cells that had been serumstarved were plated onto ®bronectin in serum-free media for 5 h, plus/minus 2.0 ng/ml TGF-b1 stimulation, harvested with bu€ered EDTA, replated onto wells coated with ®bronectin, vitronectin, collagen or ovalbumin, and allowed to attach for 30 min (378C, 5% CO2). Subsequently, the wells were washed 36, cells were harvested with trypsin and counted in a scintillation counter. Attachment to ovalbumin was subtracted. Cell conditions were plated in replicas of six, and the data is presented as the mean+s.e.m. (b) U-251MG cells treated as in A were harvested, replated onto wells coated with ®bronectin, vitronectin, or collagen and allowed to attach for 30 min (378C, 5% CO2). Cells with ®lopodia and /or lamellipodia were counted in eight ®elds at 206magni®cation in each condition. The data is presented as the mean percentage of cells+s.e.m. with ®lopodia and/or lamellipodia in each condition

consistent with the reported di€erence in the tissue distribution of Hic-5 and paxillin (Shibanuma et al., 1994) and the apparent di€erences in function, with Hic-5 protein being up-regulated in senescent ®broblasts and undetectable in many tumor cell lines (Fujita et al., 1998). The mechanism by which TGF-b1 up-regulates paxillin protein expression in the ®bronectin-adherent astrocytomas was associated largely with translation whereas the mechanism underlying TGF-b1 up-regulation of Hic-5 has been reported to be associated with transcription. Given the dissimilarity in distribution and function of these two proteins, the utilization of di€erent regulatory mechanisms is not surprising.

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

TGF-b1 stimulation failed to increase the level of tyrosine phosphorylation of paxillin in primary astrocytes resulting in an increase in the pool of unphosphorylated paxillin. In the U-251MG malignant astrocytoma cells, the level of tyrosine phosphorylation

paralleled the increase in the level of paxillin protein, resulting in an increase in the pool of tyrosine phosphorylated paxillin. Other investigators have shown previously that TGF-b1 stimulation increased phosphorylation of paxillin on tyrosine residues in rat

7983

Figure 6 TGF-b1 stimulation promotes extension of ®lopodia and lamellipodia in U-251MG cells plated onto ®bronectin. Serumstarved U-251MG cells were plated onto ®bronectin, plus/minus TGF-b1 stimulation (2.0 ng/ml), for 5 h in serum-free media, and then harvested with bu€ered EDTA and replated onto ®bronectin-coated coverslips in serum-free media without TGF-b1 stimulation for 30 min (378C, 5% CO2), followed by immuno¯uorescent analysis. (b,d,f) Cells stimulated previously with TGF-b1. (a,c,e) Cells not stimulated previously with TGF-b1. (a,b) Mab anti-paxillin staining. (c,d) Phalloidin staining. (e,f) Sequential staining with rabbit anti-integrin a5 and mAb anti-Crk; shown as merged images. 1006magni®cation Oncogene

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

7984

Oncogene

aortic smooth muscle cells by approximately twofold over 1 h of stimulation (Riedy et al., 1999). In our study, we focused on later time points, as we had found only a small increase in paxillin protein expression at 1.5 h. Other investigators have reported previously that transfection of the a5 subunit of the ®bronectin receptor integrin into myoblasts promoted cell proliferation and increased paxillin protein expression (Sastry et al., 1999). In our experiments, we found similar levels of paxillin in the malignant astrocytoma cells plated onto ®bronectin, laminin, or vitronectin for 5 h. Thus, although our experiments are not directly comparable to those reported in myoblasts (Sastry et al., 1999), our data demonstrate that engagement of the a5b1 integrin by plating on a ®bronectin substrate does not in itself result in increased paxillin protein expression in U-251MG cells, as compared to engagement of the a6b1 laminin receptor integrin and the avb3/avb5 vitronectin receptor integrins. However, our data do suggest that there is a cooperation of TGF-b1 with the a5b1 ®bronectin receptor integrin in the malignant astrocytoma cells, as TGF-b1 failed to increase the expression of paxillin protein when the cells were plated onto other substrates (laminin, collagen, or vitronectin). Plating of the cells on mAb anti-a5 that had been immobilized on plastic followed by TGF-b1 stimulation failed to increase paxillin protein, perhaps due to the incomplete cell spreading that occurs when the U-251MG are plated onto mAb anti-a5 (Gladson, unpublished data). The increase in paxillin protein expression that we detected upon TGF-b1 stimulation was associated with a signi®cant increase in cell attachment to ®bronectin and vitronectin at 30 min. TGF-b1 stimulation also resulted in increased cell spreading on ®bronectin and less so on vitronectin at 30 min, but not on collagen, that correlated with paxillin, integrin a5b1, and Crk localization to the plasma membrane in ®lopodia and lamellipodia. No change in cell spreading was detected at 1 h in cells prestimulated with TGF-b1 as compared to those not pre-stimulated, suggesting that elevated levels of paxillin protein promote a more rapid spreading of the cell. Thus, the mechanism by which increased paxillin protein expression promotes cell attachment over a 30 min assay is likely due to increased and/or a more rapid paxillin recruitment to the plasma membrane of a complex of proteins that help to mediate reorganization of the actin-based cytoskeleton. Other investigators have shown previously that paxillin helps to recruit a Crk-DOCK 180 complex to the plasma membrane that can activate Rac1, and a complex containing PAK1, Nck, members of the Cdc42/Rac1 guanine nucleotide exchange factor family, and the ARF-GAP protein p95PKL to the cell membrane (Kiyokawa et al., 1998a,b; Norman et al., 1998; Turner et al., 1999). In addition, TGF-b1 has been reported to induce expression of E-cadherin, as well as its undercoat-associated proteins a- and bcatenins, in a mechanism that required induction of

®bronectin and its integrin receptor and resulted in increased cell-cell adhesion (Wang and Chakrabarty, 2001). Whether the latter downstream event of TGFb1 stimulation is occurring in the U-251MG cells and modulates cell-matrix adhesion remains to be investigated. These tumors synthesize ®bronectin in vitro and probably in vivo, as well as synthesizing vitronectin in vivo (reviewed in Gladson, 1999). Transient increases in TGF-b1 in these tumors due to an autocrine or paracrine mechanism could thus result in increased, or a more rapid, cell attachment that promotes subsequent tumor cell migration and invasion. The increased cell attachment is not due to upregulation of the respective integrin receptors, as we found within the time frame of these assays that TGF-b1 stimulation did not increase cell surface expression of the ®bronectin or vitronectin receptor integrins. TGF-b1 stimulation can modulate cell adhesion in some cell types in speci®c conditions through the upregulation of ®bronectin and its integrin receptor (Zimmerman and Padgett, 2000). TGF-b1 is known to have pleotropic e€ects on cells in vitro, including regulation of gene transcription, activation of JNK kinase, inhibition of cell growth, and modulation of the immune response (Zimmerman and Padgett, 2000). In contrast to its e€ects on most cells propagated in vitro, TGF-b1 has a growthpromoting e€ect on some malignant astrocytoma cells (reviewed in Kjellman et al., 2000). In the most malignant astrocytic tumor, known as glioblastoma, there is elevated expression of TGF-b1 and b2 mRNAs, as well as the TGFb receptor mRNAs, TbR-I and TbR-II (Kjellman et al., 2000). However, there is markedly lower expression of Smad2, Smad3 and Smad4 mRNA, which are downstream e€ectors of TGF-b, in these tumors (Kjellman et al., 2000). This downregulation of Smad2, Smad3 and Smad4 may account for the lack of growth inhibition observed when malignant astrocytoma cells are treated with TGF-b1 in vitro. We investigated the ability of TGF-b1 to modulate cell proliferation of the U251MG cells plated onto ®bronectin in serum-free conditions and stimulated with 2.0 ng/ml TGFG-b1 for 5 h, followed by propagation in serum free media with no TGF-b1. We found no di€erence in cell proliferation at 24 and 48 h, suggesting that TGF-b1 does not modulate cell proliferation of U-251MG malignant astrocytoma cells under these conditions (Gladson, unpublished data). In summary, our data demonstrate for the ®rst time that TGF-b1 up-regulates paxillin protein expression in astrocytic cells and acts largely at the level of translation, but fails to increase paxillin phosphorylation on tyrosine residues. Furthermore, TGF-b1 upregulation of paxillin protein requires a cooperative interaction with the a5b1 ®bronectin receptor integrin. As TGF-b1 is synthesized by malignant astrocytoma cells in vivo (Kjellman et al., 2000), our data suggest an autocrine or paracrine mechanism whereby TGF-b1 up-regulates paxillin protein expression in vivo and thereby promotes tumor cell adhesion to ®bronectin.

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

Materials and methods Cells and reagents U-251MG human malignant astrocytoma cells were propagated in complete media as described (Pijuan-Thompson and Gladson, 1997). U-251MG cells were maintained at 430% con¯uency for these studies. Primary neonatal rat astrocytes were isolated and maintained as described (Benveniste et al., 1991). All experiments were performed in serum-free medium, DMEM with 1% BSA, fraction V (Fisher Scienti®c, Atlanta, GA, USA). Prior to paxillin analysis, U-251MG cells were plated in complete media overnight, and the following morning the cells were washed with PBS and then cultured in media containing 1% fetal bovine serum for one week (serum starvation). Fibronectin and vitronectin were purchased (Boehringer Manheim Corp., Indianapolis, IN, USA) and were 495% pure on reduced SDS ± PAGE. Laminin and collagen were purchased (ICN Biomedical Research Products, Costa Mesa, CA, USA). Human platelet-derived TGFb1 was purchased (R&D Systems, Minneapolis, MN, USA). MAb anti-paxillin (Transduction Labs, Lexington, KY, USA), mAb anti-Crk (Transduction Labs), mAb antiphosphotyrosine (mAb PY20) (Upstate Biotechnology, Lake Placid, NY, USA), and rabbit anti-integrin a5 (Chemicon International, Inc., Temecula, CA, USA) were purchased. Alexa 488-conjugated anti-mouse and Alexa 594-conjugated anti-rabbit IgGs, as well as Oregon-Green conjugated phalloidin, were purchased (Molecular Probes, Eugene, OR, USA). The full-length cDNA for Hic-5 (Fujita et al., 1998), was transiently transfected into U-251MG cells using DMRIE-C reagent (Life Technologies, Inc.) to determine the cross-reactivity of mAb anti-paxillin with Hic-5. Western blot analysis Western blot analysis was performed as described previously (Wang et al., 2000). Brie¯y, cells were lysed in 1% NP-40 lysis bu€er with protease inhibitors on ice for 30 min in the ¯ask, scraped with a rubber policeman and the entire lysate incubated on ice for an additional 30 min with occasional vortexing (Wang et al., 2000). Alternatively, cells were lysed in RIPA bu€er (Pijuan-Thompson et al., 1999) with the addition of 1% SDS and containing protease inhibitors on ice for 10 min. The intensity of the band(s) on autorad was estimated using densitometric analysis.

7985

plated onto 25 cm2 ¯asks coated with 10 mg/ml ®bronectin. TGF-b1 was added at initial plating onto ®bronectin as indicated. At 3 h, [35S]Methionine/cysteine was added at 0.1 mCi/ml of media. At 5 h the supernatant was removed and the label was chased with cold serum-free media for 0, 4, 8 and 24 h. Cells were then lysed in RIPA lysis bu€er (Pijuan-Thompson et al., 1999). Incorporation of the 35Slabel was determined by TCA precipitation and scintillation counting (Pijuan-Thompson and Gladson, 1997). Equivalent TCA-precipitable counts from each sample were used to isolate paxillin by immunoprecipitation using mAb antipaxillin, the immunoprecipitates separated on a 7.5% SDS ± PAGE, transferred to Immobilon membrane, and visualized by autoradiography. The intensity of the paxillin band was assessed by densitometric analysis. Northern blot analysis Northern blot analysis was performed as described previously (Gladson et al., 1995). The cDNA probe for Northern blot was synthesized by random priming using a 1.1 kb paxillin cDNA fragment (corresponding to nucleotides 1 to 1146 of the paxillin cDNA sequence) as the template. The intensity of the 1.7 kb paxillin message was assessed by phosphorimage analysis. Cell attachment assays Cell attachment assays were performed as described previously (Pijuan-Thompson et al., 1999). TGF-b1 ELISA The TGF-b1 ELISA was performed as per the instructions with the Quantikine Human TGF-b1 Immunoassay Kit (R&D Systems, Minneapolis, MN, USA). Fibronectin, vitronectin, collagen, and laminin were tested at 1.0, 10.0 and 100.0 mg/ml in the TGF-b1 ELISA. The readings from the TGF-b1 standard dilutions were plotted and found to be linear.

Abbreviations TGF-b1, Transforming growth factor-b1; TCA, Trichloroacetic acid

Immunofluorescent analysis Immuno¯uorescence was performed as described previously (Pijuan-Thompson et al., 1999). Estimation of paxillin protein half-life Serum-starved U-251MG cells (1.256106) were harvested with bu€ered EDTA, resuspended in serum-free media, and

Acknowledgments This work was supported by grants #CA75682 and CA59958 from the National Institutes of Health, National Cancer Institute to CL Gladson and #NS34856, Project #2, from the National Institutes of Health, National Institutes for Neurologic Diseases and Stroke to EN Benveniste and CL Gladson.

References Benveniste EN, Vidovic M, Panek RB, Norris JG, Reddy AT and Benos DJ. (1991). J. Biol. Chem., 266, 18119 ± 18125. Fujita H, Kamiguchi K, Cho D, Shimanuma M, Morimoto C and Tachibana K. (1998). J. Biol. Chem., 273, 26516 ± 26521. Giancotti FG and Ruoslahti E. (1999). Science, 285, 1028 ± 1032.

Gladson CL. (1999). J. Neuropathol. Exp. Neurol., 58, 1029 ± 1040. Gladson CL, Wilcox JN, Sanders L, Gillespie GY and Cheresh DA. (1995). J. Cell Sci., 108, 947 ± 956. Hall A. (1994). Ann. Rev. Cell Biol., 10, 31 ± 54.

Oncogene

TGF-b1 regulation of paxillin in astrocytoma cells X Han et al

7986

Oncogene

Kiyokawa E, Hashimoto Y, Kobayashi S, Sugimara H, Kurata T and Matsuda M. (1998a). Genes Dev., 12, 3331 ± 3336. Kiyokawa E, Hashimoto Y, Kurata T, Sugimura H and Matsuda M. (1998b). J. Biol. Chem., 273, 24479 ± 24484. Kjellman C, Olofsson SP, Hansson O, Von Schantz T, Lindvall M, Nilsson I, Salford LG, SjoÈgren H-O and Widegren B. (2000). Int. J. Cancer (Pred. Oncol.), 89, 251 ± 258. Norman JC, Jones D, Barry ST, Holt MR, Cockcrofts S and Critchley DR. (1998). J. Cell Biol., 143, 1981 ± 1995. Pijuan-Thompson V and Gladson CL. (1997). J. Biol. Chem., 272, 2736 ± 2743. Pijuan-Thompson V, Grammar R, Stewart J, Tuszynski G, Silverstein RL, Pierce SF, Murphy-Ullrich J and Gladson CL. (1999). Exp. Cell Res., 249, 86 ± 101. Riedy MC, Brown MC, Molloy CJ and Turner CE. (1999). Exp. Cell. Res., 251, 194 ± 202.

Sastry SK, Lakonishok M, Wu S, Truong TQ, Huttenlocker A, Turner CE and Horwitz AF. (1999). J. Cell Biol., 144, 1295 ± 1309. Shibanuma M, Mashimo JI, Kuroki T and Nose K. (1994). J. Biol. Chem., 269, 26767 ± 26774. Turner CE. (1998). Int. J. Biochem. Cell Biol., 30, 955 ± 959. Turner CE, Brown MC, Perrotta JA, Riedy MC, Nikolopoulos SN, McDonald AR, Bagrodia S, Thomas S and Leventhal PS. (1999). J. Cell Biol., 145, 851 ± 863. Wang D, Grammer JR, Cobbs CS, Stewart Jr JE, Liu Z, Rhoden R, Hecker T, Ding Q and Gladson CL. (2000). J. Cell Sci., 113, 4221 ± 4230. Wang H and Chakrabarty S. (2001). J. Cell. Physiol., 187, 188 ± 195. Zimmerman CM and Padgett RW. (2000). Gene, 249, 17 ± 30.