Oncogene (2000) 19, 4864 ± 4875 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc C-erbB-2/ HER-2 upregulates fascin, an -bundling associated with cell motility, in human breast cell lines

Axel Grothey1, Rintaro Hashizume1, Hong Ji1, Benjamin E Tubb4, Charles W Patrick Jr2, Dihua Yu3, Erin E Mooney1 and Pierre D McCrea*,1

1Department of Biochemistry and Molecular Biology, (Program in Genes and Development), University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas, TX 77030-4095, USA; 2Department of Plastic Surgery, (Program in Genes and Development), University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas, TX 77030-4095, USA; 3Department of Surgical Oncology, University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas, TX 77030-4095, USA; 4Department of Cell Biology, Baylor College of Medicine, Houston, Texas, TX 77030, USA

The over-expression of c-erbB-2/ HER-2, a receptor Introduction tyrosine kinase, correlates with poor prognosis in patients with breast and ovarian cancer. In the human The c-erbB-2 oncogene (also known as HER-2 or neu) cell line, MDA-MB-435, c-erbB-2 over- encodes a transmembrane glycoprotein of 185 kDa expression results in increased chemoinvasion and higher (p185), which belongs to the EGF-receptor family of metastatic properties in nude mice. However, the growth factor receptor tyrosine kinases (Hung and mechanisms by which c-erbB-2 increases the malignant Lau, 1999; Klapper et al., 2000). C-erbB-2 is ampli®ed potential of cells remains unclear. We have determined or over-expressed in 20 ± 30% of all breast and ovarian that over-expression of c-erbB-2 in MDA-MB-435 cells, and is clinically associated with poor prognosis and in some additional breast cancer cell lines, is in patients with these tumors. Mutational activation of associated with graphic increases in mRNA and protein the rat neu oncogene (V?G at amino acid 659) is levels of the actin bundling protein fascin. Heightened sucient to induce a higher metastatic potential in fascin expression has been observed in other systems to mouse 3T3 ®broblasts (Yu and Hung, 1991), and c- result in greatly increased cell motility, and indeed, our erbB-2 over-expression enhances the intrinsic meta- work employing semi-automated time-lapse microscopy static properties of human lung cancer cells (Yu et al., demonstrates that MDA-MB-435 cells over-expressing 1994). In the human breast cancer cell line MDA-MB- c-erbB-2 exhibit signi®cantly heightened cellular dy- 435, stable transfection and over-expression of c-erbB-2 namics and locomotion, while visualization of bundled results in increased metastatic properties and chemoin- micro®laments within ®xed cells revealed enhanced vasion in nude mice (Tan et al., 1997), while the formation of dendritic-like processes, microspikes and transgenic expression of c-erbB-2 in mouse breast other dynamic actin based structures. To address the induces mammary tumors and metastases means by which c-erbB-2 over-expression might result in (Guy et al., 1992; Muller et al., 1988). elevated fascin levels, we identi®ed multiple perfect On the molecular level, c-erbB-2 activation results in match TCF and NF-kB consensus sites in fascin's its auto-phosphorylation at speci®c tyrosine residues, promoter and ®rst intron, which appeared consistent and thereby the association of SH2 domain containing with the greater endogenous transcriptional activities of (Reese and Slamon, 1997). For example, direct TCF and NF-kB in c-erbB-2 over-expressing MDA-MB- interaction and phosphorylation by c-erbB-2 has been 435 cells. While such transcriptional modulation may demonstrated for Shc and grb2, leading to activation of occur in the context of the intact gene/chromatin, the ras/ MAPK (Segatto et al., 1993), PLCg (Fazioli et subsequent tests using reporter constructs did not support al., 1991), and Pl3-K pathways (Peles et al., 1992). Such involvement of these signaling pathways. In conclusion, Pl3-K activation has been recently indicated to result in highly increased fascin levels were observed in MDA- increased p21-activated kinase-1 (PAK-1) function, and MB-435 over-expressing c-erbB-2, likely contributing to thereby enhanced cell motility (Adam et al., 1998). these cells' altered actin dynamics, and increased cell Conversely, independent work has demonstrated Pl3-K motility and malignancy. Studies in progress aim to dependent cell aggregation in response to heregulin b1 discern the means by which c-erbB-2 over-expression (Tan et al., 1999). While of unclear signi®cance, c-erbB- leads to transcriptional activation of the fascin gene. 2 is further associated with the multifunctional protein Oncogene (2000) 19, 4864 ± 4875. b-catenin (Ochiai et al., 1994), known to have roles in cadherin mediated cell ± and signaling Keywords: C-erbB-2; fascin; breast cancer; cell moti- within the Wnt pathway (Behrens, 1999). Thus, despite lity; actin progress upon c-erbB-2's role in , the mechanisms by which c-erbB-2 increases the malignant potential of tumor cells appear complex. In light of the central roles of cell adhesion, motility and invasion in tumor cell behavior, proteins promot- ing such functions are likely to be directly or indirectly responsive to c-erbB-2 signaling. Indeed, activation or *Correspondence: PD McCrea Received 11 January 2000; revised 17 July 2000; accepted 1 August over-expression of c-erbB-2 is correlated with increased 2000 cell motility and invasion in various systems, accom- C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4865 panied by alterations in cell and cytoskeletal morphol- cadherin, a-catenin, or b-catenin protein levels were ogies including the generation of cellular projections observed among these cell lines, although b-catenin such as microspikes, membrane ru‚es and microvilli evidenced 1.4-fold greater tyrosine phosphorylation in (Adam et al., 1998; De Corte et al., 1994; Yu et al., cells over-expressing c-erbB-2 (Figure 2). Thus, while 1994). the protein levels of cadherin-catenin adhesion compo- MDA-MB-435 breast cancer cells over-expressing c- nents remained stable, graphic elevations of the actin erbB-2, and other breast cancer cell lines having low or bundling and motility associated protein fascin were high endogenous levels of c-erbB-2, were employed to revealed in c-erbB-2 over-expressing MDA-MB-435 assess e€ects upon the cell motility associated protein cells. fascin and components of the cadherin adhesion complex. We previously reported that b-catenin, an Fascin transcription is increased in c-erbB-2 integral component of the cadherin adhesion complex over-expressing breast cancer cells (McCrea et al., 1991) and of Wnt signal transduction (Behrens, 1999), and further, a phosphorylation Fascin's heightened expression in c-erbB-2 over-expres- substrate of c-erbB-2 (Ochiai et al., 1994), co- sing breast cancer cells could result from factors immunoprecipitates with fascin to a limited extent including increased transcriptional activity of the fascin from certain cell lines and tissue extracts (Tao et al., gene and/ or decreased degradation of fascin protein. 1996). Here, we report fascin's greatly increased We evaluated fascin gene transcription via northern mRNA and protein levels in response to c-erbB-2 detection of fascin mRNA (Figure 3), and via over-expression, and correlative increases in cell luciferase reporter analysis of fascin's promoter and dynamics and motility as assessed using video and ®rst intron (Figure 4). As shown by northern analysis, semi-automated time-lapse cell tracking systems. To c-erbB-2 over-expressing 435.eb2 cells contained in- identify signaling pathways driving fascin's increased creased fascin mRNA levels relative to 435 parental expression, we examined the fascin gene promoter and cells, which displayed no detectable fascin mRNA ®rst intron, revealing a striking number of perfect signal (Figure 3). As anticipated, fascin transcript levels match TCF and NF-kB consensus binding sites. were comparably low in 435 parental and 435.neo1, Further, higher endogenous TCF/b-catenin and NF- while comparably elevated in 435.eb1 and 435.eb2 cells kB activities were assayed in c-erbB-2 over-expressing (435.eb1 data not shown). Thus, fascin's greater cells, consistent with fascin's increased expression. mRNA transcript levels in c-erbB-2 over-expressing While subsequent testing of luciferase reporter con- cells (Figure 3) is in accordance with its elevated structs did not support the role of these pathways in protein levels (Figure 1). fascin's di€erential , such regulation may occur in the context of the native gene/chromatin Fascin promoter and intron analysis reveals multiple TCF and additional unknown modulators. In summary, we and NF-kB consensus binding sites, while luciferase demonstrate the elevated expression of the actin reporter analysis does not support their role in fascin gene bundling protein fascin in a number of c-erbB-2 over- expression expressing breast cancer cell lines, one of which was examined in detail (MDA-MB-435), and found to To address possible mechanisms governing fascin's exhibit greatly enhanced cell dynamics and motility. increased transcription in c-erbB-2 over-expressing Our work raises the issue that proteins modulating MDA-MB-435 cells, we examined fascin's promoter actin structures and dynamics, such as fascin, are likely and ®rst intron (GenBank accession # U03057). Given to directly contribute to aggressive cell behaviors. our earlier ®nding that small fractions of fascin and b- catenin exist in shared protein complexes (Tao et al., 1996), it was intriguing that fascin's promoter and ®rst intron contained 24 perfect consensus TCF (LEF) sites Results (Figure 4a). The HMG-box family member TCF (or LEF) associates with DNA, and upon activation of the Upregulation of fascin protein expression in c-erbB-2 Wnt pathway, binds nuclear-translocated b-catenin to over-expressing breast cancer cells, and increased tyrosine promote the transcription of genes important to cancer phosphorylation of b-catenin progression and development (Behrens, 1999; Eastman Western blot analysis of cell extracts adjusted for and Grosschedl, 1999). protein concentration showed a dramatic increase of Indeed, consistent with the possibility that TCF/ b- fascin levels in c-erbB-2 over-expressing MDA-MB- catenin may play a role in fascin's transcriptional 435.eb1 and 435.eb2 cells when compared with parental activation, we determined that c-erbB-2 over-expres- (435) and vector-control transfected cells (neo1) sing cells had higher endogenous TCF/ b-catenin (Figure 1a). This result was con®rmed via anti-fascin transcriptional activities, as evaluated using estab- immunoprecipitation and subsequent anti-fascin Wes- lished luciferase based reporters (TOP/ FOP ¯ash) tern blotting, and by anti-fascin immunostaining, (Korinek et al., 1997) (Figure 4c). Similarly of which evidenced an almost complete absence of fascin interest, we identi®ed 22 perfect match NF-kB protein in parental and vector-control cells (data not consensus sites within fascin's promoter and ®rst shown). ECF quanti®cation of fascin protein expres- intron (Figure 4a), and elevated endogenous NF-kB sion demonstrated respective 7.4- and 11.0-fold activity within c-erbB-2 over-expressing MDA-MB- increases in MDA-MB-435.eb1 and 435.eb2 cells over 435 cells (Figure 4b). Finally, 13 hormone response levels in 435.neo1 cells (Figure 1b). In this assay, fascin consensus elements and four glucocorticoid consensus levels of the 435 parental cells were indistinguishable sites were resolved in fascin's promoter and ®rst from background. No di€erences in total cellular E- intron.

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4866

Figure 1 Western blot (a) and enhanced chemi¯uorescence (ECF) analysis (b) of whole cell extracts of MDA-MB-435 breast cancer cells and transfected clones (435=parental cell line, 435.neo 1=vector-control transfected cells, 435.eb1 and 435.eb2= c-erbB-2 transfected/overexpressing cell lines) using polyclonal antibodies against b-catenin and fascin, and a monoclonal antibody (c-neu-Ab3) against c-erbB-2. Fascin levels were upregulated in c-erbB-2 overexpressing cell lines, whereas b-catenin remained una€ected. ECF analysis of fascin levels in 435.eb1 and 435.eb2 cell lines showed 7.4- and 11.0-fold increases, respectively, over the 435.neo1 control expression

Figure 3 Northern blot of 435 parental and 435.eb2 cells. The human fascin probe consisted of the 366 bp internal SacI fragment of the human fascin cDNA sequence. From each cell line 20, 10, or 5 mg of total RNA were resolved upon a 1% agarose/3% formaldehyde gel. Tenmg RNA of human testis Figure 2 Immunoprecipitation (IP) from HeLa, MDA-MB- served as blotting control (c), rat GAPDH cDNA was used as 435.neo1, 435.eb1 and 435.eb2 cell extract using polyclonal anti- internal loading control. Whereas fascin mRNA was readily b-catenin antibody (Ab) with subsequent Western blotting for detected in 435.eb2 cells, 435 parental cells showed no fascin- phosphotyrosine residues (P-TYR) and b-catenin. b-catenin is mRNA signal even after long ®lm exposure. The human fascin more highly tyrosine phosphorylated in 435.eb1 and 435.eb2 cells mRNA is approximately 2.7 kb, as expected (see human fascin than in 435.neo1 cells. IP of HeLa cells with rabbit IgG (IgG) cDNA Genbank #U03057). M denotes molecular weight marker served as a control (Gibco BRL)

Four luciferase based reporter constructs were c-erbB-2 over-expressing cells (data not shown). produced to test fascin's transcriptional regulation via Finally, co-transfection of established activators/sup- TCF/b-catenin, NF-kB and/or glucocorticoid (Figure pressors of Wnt (TCF/b-catenin) or NF-kB pathways 4a). Transient transfection of each of the two fascin (p50/p65), or combinations of such activators/suppres- promoter constructs indicated that they were robustly sors, did not signi®cantly alter the transcriptional transcribed relative to vector controls, but di€erences activity of fascin promoter or intron reporter con- were not apparent when comparing either construct in structs, nor did not the incubation of cells with agents c-erbB-2 over-expressing versus control MDA-MB-435 known to activate/ suppress these or the glucocorticoid cells (data not shown). Likewise, transient expression pathways (data not shown) (see Materials and of either of the two fascin intron constructs, did not methods). Thus, sequence analysis of the fascin gene reveal di€erential reporter activities coinciding with the (Figure 4a), and assay of heightened endogenous TCF/ elevated fascin mRNA and protein levels observed in b-catenin or NF-kB activities (Figure 4c), suggested a

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4867

Figure 4 Analysis of fascin gene (a, GenBank # U03057) and endogenous NF-kB and TCF transcriptional activity determined by luciferase based reporter constructs (see Materials and methods) in transiently transfected 435.neo1 and 435.eb2 cells. Fascin's promoter and ®rst intron contain 24 consensus TCF/LEF and 22 perfect match NF-kB binding motifs (a). Endogenous NF-kB(b) and TCF (c) activity was signi®cantly higher in 435.eb2 than in 435.neo1 control cells

conceivable mechanism by which fascin expression did exist when evaluating the co-presence of c-erbB-2 becomes elevated in c-erbB-2 over-expressing cells. and b-catenin. However, our luciferase-based reporter analysis of the fascin gene promoter and intron did not support such b-catenin does not appear to associate with c-erbB-2 or a model, leaving open possibilities including the with fascin in MDA-MB-435 cells requirement for additional unknown signals, and/or that fascin's transcriptional regulation (via TCF/b- To assess if c-erbB-2 over-expression alters the limited catenin, NF-kB and/or glucocorticoid, etc.), is refrac- but reported association of fascin with b-catenin (Tao tory to assessment within the context of arti®cial et al., 1996), we examined fascin and b-catenin plasmid constructs. immunoprecipitates. In contrast to results from A431 and HeLa cells, and from brain extracts (Tao et al., 1996), we were unable to visualize the co-immunopre- Increased fascin levels coincide with c-erbB-2 cipitation of fascin with b-catenin in parental or c- over-expression in other breast cancer cell lines erbB-2 over-expressing MDA-MB-435 cells (data not Given that c-erbB-2 over-expression in MDA-MB-345 shown). Recent results additionally suggest that fascin's cells results in substantially increased fascin mRNA association with b-catenin is likely to be indirect (X and protein levels, we examined other breast cancer Fang and PD McCrea, unpublished data) lessening the cell lines for expression of c-erbB-2, fascin, E-cadherin likelihood of b-catenin's relevance to fascin's cytoplas- and catenins (Figure 5 and Table 1). Comparison of mic (versus gene) function. Likewise, since b-catenin Western blot band intensities revealed that fascin was has been reported to associate with c-erbB-2 (Ochiai et more highly expressed in cells harboring both c-erbB-2 al., 1994; Shibata et al., 1996), we assayed for the and b-catenin. The only cell line evidencing high presence of this complex in parental or c-erbB-2 over- fascin levels in the absence of c-erbB-2 over-expression expressing MDA-MB-435 cells. However, in our hands was HBL100 (Ga€ney, 1982). This cell line may be (in which positive control co-immunoprecipitations exceptional as it underwent in vitro transformation were successful), even gently washed immunoprecipi- upon integration of the MPMV retroviral and SV40 tates from parental or c-erbB-2 over-expressing MDA- genomes, with gain of tumorigenicity in nude mice MB-435 cells did not evidence an association of b- (see Discussion) (Caron de Fromentel et al., 1985; catenin with c-erbB-2 (data not shown). Thus, our Dhaliwal et al., 1990; Kzhyshkowska et al., 1996). present study does not support the involvement of Therefore, while fascin expression did not coincide either of these two previously reported complexes in with c-erbB-2 expression in all cell lines, a correlation the c-erbB-2 mediated over-expression of fascin.

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4868

Figure 5 Western blot of cell extracts of di€erent breast cancer cell lines (MDA-MB-435, 435.neo1, 435.eb2, 231, 453, 361; BT483, 474; SkBr3), the human milk-derived cell line HBL100, and the epidermoid carcinoma cell line A431 (control) for b-catenin and fascin. In all cancer cell lines, fascin was upregulated only when b-catenin expression coincided with c-erbB-2 overexpression (denoted as: 7 no or low detectable expression,+intermediate,++strong protein expression). Except for the transfected cell line 435.eb2, c-erbB-2 levels represent endogenous protein

Table 1 Western blot analysis of the expression pattern of c-erbB-2, E-cadherin, b-catenin, a-catenin, and fascin in di€erent breast cancer cell line and the human milk-derived cell line HBL100 Cell line c-erbB-2 E-cadherin b-catenin a-catenin fascin

MDA-MB-435 77 ++7 MDA-MB-231 77 ++7 BT 483 + 7777 MDA-MB-453 + 77(+) 7 SkBr3 ++ 77(+) 7 MDA-MB-361 ++ + + + + BT 474 ++ + ++ + ++ HBL100 77 ++++ eb1/eb2 ++ 7 ++++

Except in HBL100, fascin was upregulated only in cells that overexpressed c-erbB-2 and had detectable levels of b-catenin. The result for HBL100 is discussed in the text. 7, no detectable expression; (+), weak; +, intermediate; ++, strong protein expression

Increased microspikes and filopodia in c-erbB-2 over-expressing breast cancer cells Since fascin is an actin-bundling protein present in stress ®bers, microspikes, and ®lopodia, we used phalloidin-¯uorescence microscopy to visualize the distribution of ®lamentous actin (F-actin) and potential changes in the actin of MDA-MB-435 cells over-expressing c-erbB-2. Both cell line clones over-expressing c-erbB-2 (435.eb1 and 435.eb2), dis- played a graphically increased extent of microspike and Figure 6 Phalloidin-rhodamine staining of F-actin in 435.neo1 (a) and 435.eb1 (b) and c) cells. The increased expression of c- ®lopodia formation when compared with control erbB-2 and fascin in 435.eb1 cells was associated with the (vector-transfected or parental) cells (Figure 6). In formation of microspikes (arrows in c) and ®lopodia-like cellular contrast, the formation of stress ®bers was not projections resulting in a dendritic cell phenotype (b). Note the signi®cantly enhanced. It is noteworthy that cells polarized distribution of microspikes in 435.eb1 cells (506) exhibiting high fascin levels evidenced a polarized distribution of cellular projections and microspikes characteristic of motile cells. undertaken in vitro over 24 h. In contrast to parental cells, both of the c-erbB-2 over-expressing cell line clones (435.eb1 and 435.eb2) displayed highly dynamic Greatly increased cell surface activity and dendritic dendritic phenotypes characterized by the extension processes in MDA-MB-435 cells over-expressing c-erbB-2 and retraction of long branched cellular protrusions. To assess cellular dynamics in response to c-erbB-2/ Interestingly, such dendritic cell morphologies have fascin over-expression, time-lapse experiments were been associated with the presence of high fascin levels

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4869 in other cell systems (Edwards et al., 1995; Yamashiro lengths are similar, maximal cell lengths are much et al., 1998). Figure 7 depicts representative still images greater for 435.eb1 cells. Figure 9 further shows that of 435.eb1 versus 435.neo1 (vector transfected negative 435.eb1 and 435.neo1 cells had almost the same control) cells derived from video images acquired at 0, average and maximal lengths at the initiation of the 12, and 24 h. Dendritic cellular projections ®rst experiment immediately following cell plating. After appeared 6 h following the seeding of cells and the initial plating trauma, an almost continuous extended distances up to 6.5 times the cell body increase in maximal and average lengths can be dimension in c-erbB-2 over-expressing cells. As such observed in both cell lines. Interestingly, in 435.eb1 dynamic changes are more readily visible in recorded but not in 435.neo1 cells, cell length (`dendritic video sequences, readers may view QuickTime movies phenotype') and cell motility was positively correlated (each 13.5 MB) of high (eb1) versus low (neo1) c-erbB- (P=0.034 versus 0.642, data not shown). In summary, 2/ fascin-expressing MDA-MB-435 breast cancer cells quantitative analysis of time-lapse experiments demon- (http://research.mdacc.tmc.edu/*agrothey). strated that MDA-MB-435 cells over-expressing c- erbB-2 (and consequently the actin-bundling protein fascin) display a statistically signi®cant increased in C-erbB-2/ fascin over-expressing cells display vitro motility and formation of cell processes, cell significantly higher in vitro motility behaviors that require dynamic rearrangements of the The e€ect of c-erbB-2 over-expression in MDA-MB- actin cytoskeleton. 435 cells was evaluated via semi-automated quantita- tive analysis of time-lapse sequences. 435.eb1 cells demonstrated much higher motility than 435.neo1 cells Discussion with an average velocity of 0.34 (s.d.+0.11) mm/min compared with 0.11 (s.d.+0.03) mm/min (P50.0001). Our results reveal a striking increased expression of the Figure 8 shows that the increased velocity of 435.eb1 actin-bundling protein fascin in two independent cells was apparent during the entire observation period. MDA-MB-435 breast cancer cell lines stably trans- 435.eb2 cells were likewise signi®cantly more motile fected to over-express c-erbB-2. These lines were each than parental or 435.neo1 cells (data not shown). previously characterized as more invasive in vitro and Strikingly, even the slowest 435.eb1 cell evidenced a more metastatic in nude mice when compared to higher average velocity over 24 h than the fastest parental or vector-control transfected cells (Tan et al., 435.neo1 cell. In other words, even on the level of 1997). As assessed using standard immuno¯uorescence individual cells within the populations, the velocities of and video technology, we ®nd that c-erbB-2 over- 435.eb1 versus 435.neo1 clones did not overlap. expressing cells display graphic rearrangements in their Visually, the di€erences in cell shapes between these actin cytoskeleton, and signi®cantly increased cell two cell lines are obvious. Likewise, using cell length as dynamics and in vitro motility. Given fascin's known a marker for dendritic morphologies, a statistically role in bundling actin ®laments and in promoting cell signi®cant di€erence between the two cell types was motility (Adams et al., 1999; Yamashiro et al., 1998), observed (P=0.0396). However, this di€erence was not its dramatically increased expression in c-erbB-2 over- as apparent as expected since most cells underwent expressing MDA-MB-435 cells and in other c-erbB-2 mitosis within the observation period, such that all expressing breast cancer cell lines is consistent with the cells condensed to the same extent in 435.eb1 cells observed phenotype of increased cell velocity and versus 435.neo1 cells. Figure 9 plots the time- dendritic morphology. dependent minimum, average and maximum cell Fascin possess two actin-binding domains within a lengths observed, and indicates that while minimal cell single molecule, facilitating the tight packing of actin

Figure 7 Sequences of time-lapse video microscopy of cell lines MDA-MB-435.neo1 (a±c) and 435.eb1 (d±f) obtained at 0, 12, and 24 h. Fascin/c-erbB-2 overexpressing 435.eb1 cells displayed a higher in vitro motility, and some showed the formation of long dendritic cell processes (arrowheads) (phase contrast, 106)

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4870

Figure 8 Comparison of average velocity of 435.neo1 and 435.eb1 cells quanti®ed by semi-automated analysis of time-lapse experiments. (a) plots the time-dependent average velocity of 435.eb1 cells and 435.neo1 cells. (b) shows the overall average velocity over 24 h and the standard deviation for both cell populations. 435.eb1 cells displayed a signi®cantly increased motility at all time points in the course of the experiment

Figure 9 Comparison of changes in length of 435.neo1 and 435.eb1 cells quanti®ed by semi-automated graphic analysis of time- lapse experiments. (a) plots the time-dependent maximum, average and minimum cell length of 435 neo 1 and 435.eb1 cells. (b) shows the mean overall cell length over 24 h and the standard deviation for both cell populations. 435.eb1 cells demonstrated a much higher maximum cell length which almost gradually increased in the course of the experiment, whereas the minimum cell length was identical in both cell lines (see text). Di€erences in average cell length between the 435.neo1 and 435.eb1 cell lines were not as dramatic as those of average velocity (Figure 8), but were statistically signi®cant (b)

®laments within dynamic subcellular structures such as the gene's response to viral inducers such as EBV microspikes, lamellipodia and ®lopodia (Edwards and (Pinkus et al., 1997). Pronounced (200-fold) increases Bryan, 1995; Tilney et al., 1998). In the mouse, fascin in fascin expression is further apparent in B-lympho- mRNA and protein exist at higher levels in brain, cytes transformed with EBV, and in Burkitt tumor cell uterus, spleen and testis, and are most abundant in lines harboring MPMV, EBNAs and LMPs (Mosialos dendritic cells of the neural and immune systems. et al., 1994; Kzhyshkowska et al., 1996). Given that Human pathologies such as Hodgkin's disease display MPMV is likewise harbored within breast HBL100 highly elevated fascin expression within lymphoma cells (Caron de Fromentel et al., 1985), viral induction tissues (Reed-Stenberg cells), very likely resulting from of the fascin gene may account for fascin's observed

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4871 elevation in HBL100 cells in the absence of detectable small G-proteins Rac, Rho, and cdc42, and the c-erbB-2 expression. guanine nucleotide exchange factor Tiam-1 (del Peso To assess possible mechanisms by which c-erbB-2 et al., 1997; Hordijk et al., 1997; Keely et al., 1997). over-expression increases fascin expression within Increased cell motility and invasiveness require en- MDA-MB-435 cells, we tested agents that stimulate hanced plasticity of the actin cytoskeleton, which is or inhibit signaling pathways known to be activated via regulated via actin associated proteins (Mitchison and c-erbB-2, such as the ras/MAP kinase, Pl3 kinase and Cramer, 1996). In addition to our work, fascin's PLCg cascades (Fazioli et al., 1991; Peles et al., 1992; importance in generating and/or maintaining dynamic Segatto et al., 1993). Given our ®nding of 22 NF-kB, cellular processes has been indicated following fascin's 24 TCF and four glucocorticoid consensus binding stable over-expression within normal kidney epithelial sites within fascin's gene promoter and ®rst intron, we cells, causing the loss of cell ± cell junction integrity, the additionally tested agonists of the NF-kB, Wnt (TCF/ development of a ®broblastic phenotype and dramati- b-catenin) and glucocorticoid pathways. While expo- cally increased cellular dynamics and motility (Yama- sure to the NF-kB activator IL-1 increased endogenous shiro et al., 1998). Antisense studies conducted upon fascin protein levels in MDA-MB-435 cells by fetal rat hippocampal neurons correspondingly demon- approximately twofold over 48 h, additional agents strated that fascin depletion results in the loss of acting upon distinct molecular pathways (see Materials ®lopodia and well-spread growth cones (Edwards and and methods), failed to display signi®cant e€ects upon Bryan, 1995). endogenous fascin levels or reporter constructs (data In conclusion, our observation of signi®cantly not shown). Similarly, transient co-transfection of NF- elevated dynamics and cell motility in c-erbB-2 over- kB and/or Wnt (TCF/b-catenin) activators/inhibitors expressing MDA-MB-435 cells, and of increased had little impact upon constructs reporting the formation of actin-dependent, dynamic structures such transcriptional activity of fascin's promoter or intron as ®lopodia and microspikes, suggests that the (data not shown). induction of fascin in response to c-erbB-2 may play However, the NF-kB and/or Wnt (TCF/b-catenin) a role in promoting more invasive and metastatic cell pathways may none the less contribute to modulating properties. Therefore, fascin may serve as a down- fascin's in vivo expression. For example, robust stream cytoskeletal e€ector of a more aggressive increases in endogenous fascin mRNA/protein levels, cellular phenotype within c-erbB-2 over-expressing or the activities of exogenous reporters, may require cells. parallel activation of as yet uncharacterized signaling cascades, or the activation of known pathways in combinations that have not yet been discerned. As mentioned previously, EBV transformation results in Materials and methods greatly increased fascin expression (Mosialos et al., 1994). Given that EBV is known to induce NF-kB Cell lines and culture mediated signaling activity (Eliopoulos and Young, The MDA-MB-435 cell line, obtained from Dr Janet Price 1998), our observation of higher endogenous NF-kB (UT MDACC), is an estrogen receptor-negative cell line with activities in c-erbB-2 over-expressing MDA-MB-435 ®broblastic morphology isolated from the pleural e€usion of cells and the identi®cation of multiple perfect match a patient with breast cancer (Cailleau et al., 1974). Cells were NF-kB consensus sites within the fascin promoter and grown in DMEM/F12 (Life Technologies, Inc. [GIBCO BRL], Gaithersburg, MD, USA) supplemented with 10% ®rst intron suggests a role for NF-kB in the c-erbB-2 fetal bovine serum (GIBCO). During the time-lapse experi- mediated upregulation of fascin. Likewise, rat mam- ments described below, cells were maintained in Leibowitz mary epithelial cells exposed to glucocorticoids have medium (GIBCO). The generation and characterization of been reported to evidence markedly reduced fascin stably c-erbB-2 transfected clones (435.eb1 and 435.eb2), as mRNA and protein levels (Wong et al., 1999), in well as vector alone transfected control cells (435.neo1), was keeping with known glucocorticoid-induced increases previously described (Tan et al., 1997). The breast cancer cell of the NF-kB antagonist IkB (Aljada et al., 1999). lines MDA-MB-231, MDA-MB-453, MDA-MB-361, BT 483, Additional supportive evidence of a steroid hormone BT 474, and SKBR3 were obtained from the American e€ect upon fascin expression includes our recent Tissue Culture Collection (Rockville, MD, USA) and demonstration of fascin's preferential upregulation in maintained in DMEM supplemented with 10% fetal bovine serum. All cell lines were ascertained to be mycoplasma- hormone receptor negative breast cancers (Grothey et negative by 4,6-diamidino-2-phenylindole (DAPI) staining. al., 2000). Wnt (TCF/b-catenin) involvement likewise remains conceivable given our observations of the strikingly large number of TCF sites within fascin's Cell culture and transfections promoter and ®rst intron, the correlative co-presence The breast cancer cell line MDA-MB-435 and clones stably of b-catenin and c-erbB-2 in breast cancer cell lines, transfected with c-erbB-2 (MDA-MB-435-eb1 and -eb2) and c-erbB-2's reported (but not repeated in our (Tan et al., 1997), were routinely grown in RPMI 1640 hands) interaction with b-catenin (Ochiai et al., 1994; supplemented with 0.1% MEM non-essential amino acids, Shibata et al., 1996). In all cases, further analysis of the 0.1% sodium pyruvate, 0.1% vitamins and 10% fetal calf fascin gene and of impinging signaling cascades will be serum in the presence of 7% CO2. Transfections were carried out using Fugene 6 (Boehringer Mannheim), and a required to clarify c-erbB-2 driven processes contribut- total plasmid DNA load of 1 mg per well of six well cell ing to increased fascin levels, and thereby to more culture dishes. Plasmids harboring cDNAs coding for the aggressive cellular behaviors. p65 and p50 subunits of NF-kB (pRSV- p65kB and pRSV- Various proteins that modulate actin cytoskeletal p50kB) (Reddy et al., 2000), and an NF-kB luciferase dynamics have been associated with an increased reporter construct driven by four IkB consensus sites placed malignant potential in tumor cells, for example, the upstream of a minimal SV40 promoter (pGL-kB) (Reddy et

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4872 al., 2000), were gifts from WS Liao (UT MDACC). A Immunoprecipitation plasmid coding for a super-repressive form of 1kB (Van Cell line extracts were prepared in 0.75% NP-40 in calcium- Antwerp et al., 1996), was kindly provided by PE Chiao free phosphate bu€ered saline (PBS7) (136 mM NaCl, (UT MDACC). Plasmids harboring cDNAs coding for 2.6 m KCl, 8 mM Na HPO , and 1.4 mM KH PO ), human Tcf-4 (phTcf-4), dominant negative TCF (poNhTcf- M 2 4 2 4 supplemented with protease inhibitors (1 ng/ml pepstatin, 4), and a TCF/ Lef luciferase reporter construct (pTOP- 2 ng/ml TLCK, 2 ng/ml leupeptin, 4 ng/ml aprotinin, 10 ng/ FLASH/ pFOPFLASH), were gifts from Mien-Chie Hung ml antipain, 50 ng/ml benzamidine, 10 ng/ml STI, 100 ng/ml (UT MDACC) originating from H Clevers (U Hospital, iodoacetamide, 0.4 m phenylmethylsulfonyl ¯uoride Utrecht) (Korinek et al., 1997). Stabilized Xenopus b-catenin M [PMSF], 5 m EDTA). For assessing the phosphorylation cDNA with a C-terminal myc tag (pXBC69) (Yost et al., M status of b-catenin, phosphatase inhibitors were included in 1996), was kindly provided by D Kimelman (U Washington) the lysis bu€er (50 m NaFl, 10 mM sodium pyrophosphate, and inserted using standard subcloning methods into the M 1m NaMb, 2 mM vanadate). After debris removal via BamHI/ EcoRI polylinker restriction sites of pCDNA3.1 M centrifugation (14 000 r.p.m. microfuge, 30 min), the resulting (Invitrogen) (generating pXb/ pCDNA3.1). b-Engrailed (pb- supernatants were subjected to a preclearing step with normal En/ pCDNA3.1+MT), a myc-tagged chimera of stabilized rabbit serum, followed by immunoprecipitation using anity- Xenopus b-catenin and the constitutive transcriptional puri®ed anti-b-catenin, anti-fascin, or normal rabbit IgG repressor domain of Engrailed was produced as (negative control). Immunoprecipitates were washed twice in described (Montross et al., 2000). Cells were transfected at lysis bu€er and once in PBS7 before their reduction in 50 ± 60 ± 70% con¯uence (seeded 1 day prior), and harvested 2 100 mlof26SDS-sample bu€er at 958C for 5 min. days following transfection.

Northern blot Antibodies Con¯uent monolayers of 435 and 435.eb2 cells were lysed Anity-puri®ed anti-fascin rabbit polyclonal antibodies were and their RNA extracted using the Ultraspec RNA isolation raised against mouse fascin protein, expressed and puri®ed as reagent from Biotecx (Houston, TX, USA). For each cell a thioredoxin fusion protein. The thioredoxin domain was line, 20, 10, and 5 mg of total RNA were resolved upon a 1% cleaved from fascin via enterokinase treatment (Invitrogen, agarose gel with 3% formaldehyde. RNA of human testis, San Diego, CA, USA) (Dickason et al., 1995). Hybridoma which is known to have high fascin levels, served as control. supernatant of the mouse monoclonal anti-fascin antibody Transfer occurred in 106SSC using a Stratagene pressure clone 55 k-2 (Mosialos et al., 1994) was a generous gift of blotter at 80 mmHg for about 90 min onto a GeneScreen Dako Corp. (Carpinteria, CA, USA). Anti-b-catenin rabbit Plus nylon membrane (NEN Life Science Products, Boston, polyclonal antibodies were raised and anity puri®ed as MA, USA). After transfer, the transferred RNA was cross- previously indicated (McCrea et al., 1993). The c-neu-Ab3 linked to the membrane using an UV cross-linker (Strata- mouse monoclonal antibody against c-erbB-2 was obtained gene, La Jolla, CA, USA). UV visualization of the gel from Oncogene Science, Inc. (Uniondale, NY, USA); mouse revealed that almost all of the RNA had been evenly monoclonal antibodies against E-cadherin, a-catenin, and transferred to the membrane. Only a reduced amount of phosphotyrosine were purchased from Transduction Labora- RNA larger than 7 ± 10 kb remained in the gel. The human tories (Lexington, KY, USA). fascin probe consisted of the 366 bp internal (coding region) SacI fragment of the human fascin cDNA sequence in Bluescript (SK-hfas, kindly provided by G Mosialos) (see Western blot GenBank accession # U03057; probe comprised of positions Cell line extraction for Western blotting analysis occurred in 865 ± 1231 of this sequence). The probe was labeled with 32P- RIPA bu€er (1% NP-40; 0.1% sodium dodecyl sulfate dATP (Amersham) and the Strip-EZ random-priming cDNA [SDS]; 0.5% desoxycholate [DOC]; 150 mM NaCl; 50 mM labeling kit (Ambion Corp., Austin, TX, USA). Both pre- Tris, pH 7.5) or in boiling 26SDS-sample bu€er (4% SDS; hybridization and hybridization were performed in ZIP-HYB 10% glycerol; 2% b-mercaptoethanol; 250 mM Tris-HCl, solution (Ambion) with addition of 0.1 mg/ml ®nal concen- pH 6.8; 0.006% bromophenol blue). Protein concentrations tration boiled, sheared salmon sperm DNA for blocking. Pre- of lysates were measured using the Bio-Rad DC Protein hybridization was approximately at 658C within a rotating Assay kit (Bio-Rad Laboratories, Hercules, CA, USA). hybridization oven. Then 50 ml of labeled probe was mixed Either 10 or 25 mg of total protein of each lysate were with 150 ml of 1 mg/ml sheared salmon sperm DNA and resolved upon 7% SDS ± PAGE gels and electrotransferred 50 ml206SSC, boiled 5 min, heated at 688C for 30 min and to nitrocellulose. Monoclonal or anity-puri®ed polyclonal ®nally added to the hybridization tube. Hybridization primary antibodies were incubated with blots at 1 : 500 to continued at 658C overnight. For the control blot, the 1 : 1000 dilution in minimal volumes of TBS-T bu€er membrane was stripped at 658C using Strip-EZ dCTP- (20 mM Tris-HCl, pH 7.6; 137 mM NaCl; 0.1% Tween 20) degrading stripping solutions (Ambion), rinsed twice in and 5% dry milk for 1 ± 2 h at room temperature or for 0.1% SDS, and put back into a pre-hybridization solution 16 h at 48C. Anti-mouse or anti-rabbit goat-HRP-conjugated as described above. A standard 32P-dATP labeled GADPH secondary antibodies (Bio-Rad) were incubated at 1 : 3000 probe fragment (approximately 1 kB) was used as control. dilution for 1 h at room temperature. Proteins of interest were detected by enhanced chemiluminescence (ECL) Fascin promoter/intron reporter constructs (Amersham Corp., Arlington Heights, IL, USA). Protein expression was quanti®ed by enhanced chemi¯uorescence Two overlapping reporter constructs of the mouse fascin (ECF) analysis, in which proteins were separated by SDS ± promoter (GenBank # U90355) were PCR ampli®ed using PAGE and electrotransferred onto PVDF membranes. primer sequences harboring KpN1 restriction sites: up-stream Immunodetection was performed using the Vistra Fluores- primers mFgp1/F-KpnI, -3448 NT, 5'-GGGGTACCCAGCT- cence Western blotting kit (Amersham), with signal GGAGAGTGCTTGT-3' or mFgp2/F-KpnI, -1168, 5'-GGG- ampli®cation by detection of the secondary antibody with GTACCATGGGAAAGCCATGCCATC-3', opposing down- an anti-¯uorescein alkaline phosphatase conjugate followed stream primer mFgp/B-KpnI, 18 NT, 5'-GGGGTACCTTA- by the ¯uorescent substrate AttoPhos (Cano et al., 1992). GCCCGCTGCAGTTCT-3'. Ampli®ed fragments were in- Image analysis of ECF scans (Storm 840, Molecular serted into the KpN1 polylinker site of pGL-3 Basic Dynamics, Inc., Sunnyvale, CA, USA) was performed with (Promega), which lacks an endogenous promoter. Two non- ImageQuaNT (Molecular Dynamics). overlapping reporter constructs spanning fascin's ®rst intron

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4873 were produced via PCR ampli®cation employing primers Electronic Products Ltd., Hawthorne, NY, USA); IPLab harboring BamHI restriction sites: upstream primer mFgI1/F- Spectrum software (Scanalytics Inc., Fairfax, VA, USA) BamHI, 893 NT, 5'-CGGGATCCAACGAGGGGAACGT- orchestrated the entire acquisition process, providing GTCC-3' opposing down-stream primer mFgI1/B-BamHI, complete automation. The microscope stage was enclosed 3632 NT, 5'-CGGGATCCTCACTCATTCTTAGCCAGC-3'; within a temperature-controlled incubator (Olympus, Mel- and upstream primer mFgI2/F-XhoI, 3633 NT, 5'-CCGG- ville, NY, USA) which maintained the ambient tempera- CTCGAGCAGGTGAGGCTGGGCTT-3', opposing down- ture at 37.0+0.18C. stream primer mFgI2/B-XhoI, 6451 NT, 5'-CCGGCTCGAG- GAGAAGGTCTCACAGTGTA-3'. Intron fragments were inserted into the luciferase reporter vector pGL3-Promoter Digital time-lapse recording (Promega), which contains the SV40 basal promoter. Correct In conventional time-lapse recording of extended experi- orientation of the fascin promoter and ®rst intron fragments ments, a ®xed ®eld of view is continuously monitored and were veri®ed by cDNA sequencing of inserts from 5' and 3' recorded on video tapes (i.e. analog system). The video tapes ends. are then post-processed to obtain a sequence of digital images. This approach, however, has two limitations: video Gene transcription reporter assays recording results in loss of image quality (as much as 60%), and one is limited to monitoring a single ®eld, resulting in the Cells cultured within six well dishes were grown to 60 ± 70% need to conduct many experiments to ascertain the behavior con¯uence and transiently co-transfected (FuGENE 6, of a cell population. To overcome these limitations, we used Boehringer Mannheim, Indianapolis, IN, USA) with 0.2 mg digital time-lapse recording to continuously monitor tumor of the selected luciferase based fascin reporter construct and cell cultures over extended periods. At ®xed and user- 0.2 mg of pRL-TK (Promega Madison, WI, USA), which speci®ed time intervals, the computer moved the motorized harbors cDNA encoding Renilla luciferase and is employed stage according to preprogrammed locations of monitored to control for transfection eciencies. As above indicated culture wells and triggered the acquisition of a series of and referenced, further co-transfections utilized plasmids digital images. Experiments were typically conducted for 24 h (0.2 mg) harboring cDNAs encoding either NF-kB (p65 with 30 min acquisition intervals. Each gray scale 6406480 and/or p50), IkB, hTcf-4, dnTCF, b-catenin or b-Engrailed pixel image required 314 Kb of disk space so that the (data not shown). Total transfected cDNA loads were resulting 48 frame sequence took 15 Mb of disk space. uniformly made to 1 mg per cell culture well using plasmid Images were stored on 1 Gb Jaz drive cartridges (Iomega, lacking insert (pGL3, Promega). Negative control transfec- Roy, UT, USA) for post-processing. Sequences of digitized tions likewise employed plasmid lacking insert (pGL Basic). images were converted to QuickTime movies within IPLab Transcriptional activities were measured using the dual- Spectrum software for visualization and analysis of cell luciferase reporter assay system according to the manufac- migration. turer's instructions (Promega).

Cell tracking Immunofluorescence microscopy Time-lapse experiments recorded frame sequences captured For staining of the actin-cytoskeleton, the MDA-MB- over 24 h. To determine the spatial movement of individual 435.neo1, 435.eb1, and 435.eb2 cells were grown for 18 h cells, phase-contrast optical cell centroids were ®rst marked on glass coverslips to 50 ± 80% con¯uence, ®xed for 15 min at by crosshairs and recorded as x-y-coordinates relative to the room temperature using 3.75% formaldehyde-PBS, and upper left edge of the frame displayed in the software permeabilized with 0.1% Triton X-100 in PBS. Cells were package IP Lab (Scanalytics). A standardized calibration of blocked with 0.2% BSA and 5% normal donkey serum the optical system for recording and image analysis allowed (Jackson ImmunoResearch Labs, Inc., West Grove, PA, measurement determination in metric values (mm). To avoid USA), and thereafter incubated with rhodamine (TRITC)- a selection bias, all vital cells visible in the initial frame were conjugated phalloidin (1 : 200) (Molecular Probes, Inc., included in the assessment and tracked over 24 h. If a Eugene, OR, USA) for 1 h at 378C. For visualization of mitotic event occurred, one of the daughter cells was fascin, cells were ®xed and permeabilized in methanol:acetone randomly selected for further tracking. In total, the analysis 1 : 1 for 2 min at room temperature, rinsed with 16PBS and included 24 435.eb1 and 21 435.neo1 cells in sequences of 48 blocked with 3% BSA in PBS. Hybridoma supernatant of frames. Thus, 1152 and 1008 x-y-coordinates were respec- mouse anti-fascin antibody was added in 1 : 5 ± 1 : 10 dilutions tively recorded for 435.eb1 and 435.neo1 cells. The data was for 4 h at room temperature followed by three washes with transferred to an Excel spreadsheet (Microsoft Corp., PBS/0.75% NP-40. The secondary rhodamine-conjugated Redmond, WA, USA) and StatView 5.0 (Abacus Software, goat anti-mouse antibody (1 : 20 ± 1 : 50) was incubated for Berkeley, CA, USA) for further calculations and statistical 1 h at room temperature. analysis. For each cell and frame, the distance D d traveled between time points t0 and t1 was calculated as Dd=((x 7x )2+(y 7y )2)2. The velocity v of a cell at Video microscopy and digital image acquisition 11 10 11 10 1 a given time point was expressed as v1=D d/30 [mm/min]. In The tumor cell cultures were visualized using an inverted view of the apparent dendritic morphology of c-erbB-2 over- microscope (Olympus, Melville, NY, Model IX 70) with a expressing cells, we sought to obtain quantitative informa- 46objective (0.13 NA), phase optics, and a black-and- tion on di€erences in cell shape between 435.eb1 and white charge-coupled device (CCD) camera (Panasonic, 435.neo1 cells. To this end, the shape of each cell was Secaucus, NJ, Model BP500). The composite video signal outlined graphically in all frames by drawing a straight line from the camera was fed to a 24-bit frame grabber (Scion from one cell edge to the centroid and then another linked Corp., Frederick, MD, USA) housed in a PowerTower- line from the centroid to the furthest opposite cell edge. The Pro225 computer (Power Computing, Round Rock, TX, summed length of the two lines was employed as an USA) for digitization at maximum data integrity. The indication of overall `cell length', i.e. a marker for the frame grabber had a resolution of 6406480 pixels, spatial extension of cellular projections. As the same cells corresponding to a 3.262.4 mm ®eld of view for the 4x were analysed for cell length and cell tracking, changes in objective. A high-resolution monitor allowed real-time cell length and motility could be correlated. The paired visualization of the digitized images. The computer also Student t and Spearman rang correlation tests were used for controlled the microscope XYZ motorized stage (Ludl statistical analysis.

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4874 Assay of signaling pathways modulating endogenous fascin gene 6, 8, 16, 24 and 36 h prior to cell lysis and Western blotting; and protein expression and (6) glucocorticoid agonist dexamethasone (Wong et al., 1999), cells cultured at 1 mM for 48 h prior to cell lysis and To identify signaling cascades that might modulate endogen- Western blotting. Some of these antagonists/agonists were ous fascin gene transcription or protein expression, we further tested with respect to their ability to inhibit/stimulate exposed cells to selected pathway antagonists and/or agonists transcription of our exogenously introduced (transiently (data not shown). Exogenous agents included: (1) Ras transfected) luciferase based fascin reporter constructs (data pathway antagonist HFP (a-hydroxyfarnesyl-phosphoric not shown). acid) (Gibbs et al., 1993), cells cultured at 1 mM for 0, 6, 12, 24, 36 and 48 h prior to cell lysis and Western blotting; (2) P13 kinase antagonist Ly-294002 (Vlahos et al., 1994), cells cultured at 50 mM for 0, 6, 12, 24, 36 and 48 h prior to Acknowledgments cell lysis and Western blotting; (3) Small G-protein and p21 Axel Grothey was supported by a grant of the German ras antagonist perillic acid (Crowell et al., 1991), cells Cancer Foundation (Mildred-Scheel-Stiftung). We also cultured at 1 mM for 0, 6, 12, 24, 36 and 48 h prior to cell gratefully acknowledge the current and past sources of lysis and Western blotting; (4) Src family PTK antagonist support that have contributed to the execution of this Pp1/2 (Calbiochem, La Jolla, CA, USA), cells cultured at work: NIH RO1 Grant GM 52112; Texas ARP Grant 15 ± 500 nM for 2, 6, 12 and 24 h prior to cell lysis and Western 135; March of Dimes Basil O'Connor Grant 5-0926; blotting; (5) NF-kB agonists IL-1 and TNF-a (Roche, Kleberg Foundation Award; and CCSG Developmental Indianapolis, IN, USA), cells cultured at 10 ng/ml for 0, 2, Funds CCSG-CA 16672.

References

Adam L, Vadlamudi R, Kondapaka SB, Cherno€ J, Hordijk PL, ten Klooster JP, van der Kammen RA, Michiels Mendelsohn J and Kumar R. (1998). J. Biol. Chem., 273, F, Oomen LCJM and Collard JG. (1997). Science, 278, 28238 ± 28246. 1464 ± 1466. Adams JC, Clelland JD, Collett GD, Matsumura F, Hung MC and Lau YK. (1999). Semin. Oncol., 26, 51 ± 59. Yamashiro S and Zhang L. (1999). Mol. Biol. Cell., 10, Keely PJ, Westwick JK, Whitehead IP, Der CJ and Parise 4177 ± 4190. LV. (1997). Nature, 390, 632 ± 636. Aljada A, Ghanim H, Assian E, Mohanty P, Hamouda W, Klapper LN, Kirschbaum MH, Sela M and Yarden Y. Garg R and Dandona P. (1999). J. Clin. Endocrinol (2000). Adv. Cancer Res., 77, 25 ± 79. Metab., 84, 3386 ± 3389. Korinek V, Barker N, Morin PJ, van Wichen D, de Weger R, Behrens J. (1999). Cancer Metastasis Rev., 18, 15 ± 30. Kinzler KW, Vogelstein B and Clevers H. (1997). Science, Cailleau R, Young R, Olive M and Reeves W. (1974). J. Natl. 275, 1784 ± 1787. Cancer Inst., 53, 661 ± 674. Kzhyshkowska JG, Kiselev AV, Gordina GA, Kurmashow Cano RJ, Torres MJ, Klem RE and Palomares JC. (1992). VI, Portjanko NM, Ostashkin AS and Ilyin KV. (1996). Biotechniques, 12, 264 ± 269. Immunol. Lett., 53, 101 ± 104. Caron de Fromentel C, Nardeux PC, Soussi T, Lavialle C, McCrea PD, Brieher WM and Gumbiner BM. (1993). J. Cell. Estrade S, Carloni G, Chandrasekaran K and Cassingena Biol., 123, 477 ± 484. R. (1985). Exp. Cell. Res., 160, 83 ± 94. McCrea PD, Turck CW and Gumbiner B. (1991). Science, Crowell PL, Chang RR, Ren ZB, Elson CE and Gould MN. 254, 1359 ± 1361. (1991). J. Biol. Chem., 266, 17679 ± 17685. Mitchison TJ and Cramer LP. (1996). Cell., 84, 371 ± 379. DeCorteV,DePotterC,VandenbergheD,VanLaerebeke Montross WT, Ji H and McCrea PD. (2000). J. Cell. Sci., N, Azam M, Roels H, Mareel M and Vandekerckhove J. 113, 1759 ± 1770. (1994). J. Cell. Sci., 107, 405 ± 416. Mosialos G, Yamashiro S, Baughman RW, Matsudaira P, delPesoL,Hernandez-AlcocebaR,EmbadeN,CarneroA, Vara L, Matsumura F, Kie€ E and Birkenbach M. (1994). Esteve P, Paje C and Lacal JC. (1997). Oncogene, 15, J. Virol., 68, 7320 ± 7328. 3047 ± 3057. Muller WJ, Sinn E, Pattengale PK, Wallace R and Leder P. Dhaliwal MK, Giovanella BC and Pathak S. (1990). (1988). Cell., 54, 105 ± 115. Anticancer Res., 10, 113 ± 117. Ochiai A, Akimoto S, Kanai Y, Shibata T, Oyama T and Dickason RR, Edwards RA, Bryan J and Huston DP. (1995). Hirohashi S. (1994). Biochem. Biophys. Res. Commun., J. Immunol. Methods, 185, 237 ± 244. 205, 73 ± 78. Eastman Q and Grosschedl R. (1999). Curr. Opin. Cell. Biol., Peles E, Lamprecht R, Ben-Levy R, Tzahar E and Yarden Y. 11, 233 ± 240. (1992). J. Biol. Chem., 267, 12266 ± 12274. Edwards RA and Bryan J. (1995). Cell. Motil. Cytoskeleton, Pinkus GS, Pinkus JL, Langho€ E, Matsumura F, Yama- 32, 1±9. shiro S, Mosialos G and Said JW. (1997). Am. J. Pathol., Edwards RA, Herrera-Sosa H, Otto J and Bryan J. (1995). J. 150, 543 ± 562. Biol. Chem., 270, 10764 ± 10770. Reddy SA, Huang JH and Liao WS. (2000). J. Immunol., 164, Eliopoulos AG and Young LS. (1998). Oncogene, 16, 1731 ± 1355 ± 1363. 1742. Reese DM and Slamon DJ. (1997). Stem. Cells., 15, 1±8. Fazioli F, Kim UH, Rhee SG, Molloy CJ, Segatto O and Di Segatto O, Pelicci G, Giuli S, Digiesi G, Di Fiore PP, Fiore PP. (1991). Mol. Cell. Biol., 11, 2040 ± 2048. McGlade J, Pawson T and Pelicci PG. (1993). Oncogene, 8, Ga€ney EV. (1982). Cell. Tissue Res., 227, 563 ± 568. 2105 ± 2112. Gibbs JB, Pompliano DL, Mosser SD, Rands E, Lingham ShibataT,OchiaiA,KanaiY,AkimotoS,GotohM,Yasui RB, Singh SB, Scolnick EM, Kohl NE and Oli€ A. (1993). N, Machinami R and Hirohashi S. (1996). Oncogene, 13, J. Biol. Chem., 268, 7617 ± 7620. 883 ± 889. Grothey A, Hashizume R, Sahin AA and McCrea PD. Tan M, Grijalva R and Yu D. (1999). Cancer Res., 59, 1620 ± (2000). Br.J.Cancer,88. 1625. Guy CT, Webster MA, Schaller M, Parsons TJ, Cardi€ RD Tan M, Yao J and Yu D. (1997). Cancer Res., 57, 1199 ± and Muller WJ. (1992). Proc. Natl. Acad. Sci. USA, 89, 1205. 10578 ± 10582.

Oncogene C-erbB-2 upregulates fascin in breast cancer cell lines A Grothey et al 4875 Tao YS, Edwards RA, Tubb B, Wang S, Bryan J and Yamashiro S, Yamakita Y, Ono S and Matsumura F. (1998). McCrea PD. (1996). J. Cell. Biol., 134, 1271 ± 1281. 9, 993 ± 1006. Tilney LG, Connelly PS, Vranich KA, Shaw MK and Guild Yost C, Torres M, Miller JR, Huang E, Kimelman D and GM. (1998). J. Cell. Biol., 143, 121 ± 133. Moon RT. (1996). Genes. Dev., 10, 1443 ± 1454. Van Antwerp DJ, Martin SJ, Kafri T, Green DR and Verma Yu D, Wang SS, Dulski KM, Tsai CM, Nicolson GL and IM. (1996). Science, 274, 787 ± 789. Hung MC. (1994). Cancer Res., 54, 3260 ± 3266. Vlahos CJ, Matter WF, Hui KY and Brown RF. (1994). J. Yu DH and Hung MC. (1991). Oncogene, 6, 1991 ± 1996. Biol. Chem., 269, 5241 ± 5248. Wong V, Ching D, McCrea PD and Firestone GL. (1999). J. Biol. Chem., 274, 5443 ± 5453.

Oncogene