Expression of the β3 integrin subunit in melanoma in situ has been found to correlate with tumor thickness, the ability to invade and metastasize, and poor prognosis. Transition from the radial growth phase (RGP) to the vertical growth phase (VGP) is a critical step in melanoma progression and survival and is distinguished by the expression of β3 integrin. The molecular pathways that operate in melanoma cells associated with invasion and metastasis were examined by ectopic induction of the β3 integrin subunit in RGP SBcl2 and WM1552C melanoma cells, which converts these cells to a VGP phenotype. We used cDNA representational difference analysis subtractive hybridization between β3-positive and -negative melanoma cells to assess gene expression profile changes accompanying RGP to VGP transition. Fourteen fragments from known genes including osteonectin (also known as SPARC and BM-40) were identified after three rounds of representational difference analysis. Induction of osteonectin was confirmed by Northern and Western blot analysis and immunohistochemistry and correlated in organotypic cultures with the β3-induced progression from RGP to VGP melanoma. Expression of osteonectin was also associated with reduced adhesion to vitronectin, but not to fibronectin. Osteonectin expression was not blocked when melanoma cells were cultured with anti-αvβ3 LM609 mAb, mitogen-activated protein kinase, or protein kinase C inhibitors, indicating that other signaling pathway(s) operate through αvβ3 integrin during conversion from RGP to VGP.

Adhesion is important for cell survival and proliferation. Numerous signaling pathways are activated by the interactions of cells with matrix proteins via integrin receptors (1, 2). The expression patterns for integrins differ widely between normal and malignant tissues. A major difference between normal and malignant cells is that normal cells undergo apoptosis when prevented from attaching to substrate, whereas tumor cells can survive and continue to proliferate. Gain or loss of integrin expression in cancer cells appears to be a logical consequence of adaptation and survival in a different environment. Integrins can be expressed in cell-specific and tumor stage-specific patterns. In human melanoma, expression of the integrin αvβ3 appears to confer a tumorigenic phenotype (3).

Five distinct steps have been proposed for the progression of melanoma, based on clinical and histopathological features: common acquired and congenital nevi with structurally normal melanocytes; dysplastic nevus with structural and architectural atypia; RGP3 melanoma without and VGP melanoma with competence for metastasis; and metastatic melanoma (4). The genetic alterations responsible for the development and stepwise progression of melanoma are still unclear, but classification by gene expression profiling has been proposed to identify subsets of melanomas that correlate with phenotypic characteristics important for disease progression (5). The transition from RGP to VGP melanoma is a biologically critical step in melanoma progression. A variety of changes can be observed in sections of melanoma lesions and in cultured cells that may help explain RGP to VGP progression (6). Unlike RGP melanoma cells, VGP melanoma cells are tumorigenic and easily adapt to growth in culture. VGP melanoma cells are also less dependent on exogenous growth factors and have growth characteristics similar to those of metastatic cells, such as anchorage-independent growth in soft agar and tumorigenesis in immunodeficient mice. VGP primary melanomas display numerous cytogenetic abnormalities, suggesting considerable genomic instability. Only minor additional genetic changes may be required for further progression to metastatic dissemination because most VGP melanomas can be readily adapted to a metastatic phenotype through selection in growth factor-free media and induction of invasion through artificial basement membranes (7). This suggests that microenvironmental factors such as cell-matrix and cell-cell signaling are critical for the metastatic phenotype.

Several adhesion molecules have been studied in melanoma progression. The most notable is the β3 subunit of the αvβ3 vitronectin receptor (3). It is now recognized as a specific melanoma-associated marker that distinguishes RGP from VGP melanomas (8). β3 is a prime candidate for prognostic studies (9) because its expression correlates closely with clinical recurrence and mortality (10, 11). The contribution of β3 to metastatic behavior of melanoma was investigated by comparing cell variants with different levels of αvβ3 expression (12) and by modulating αvβ3 function with antibodies and peptides (13, 14). The strongest evidence for the role of β3 in conversion of RGP to VGP melanoma comes from expression studies of the β3 protein in RGP melanoma cell lines (15). Integrin αvβ3 was also demonstrated to promote melanoma cell survival (16).

The genetic determinants for β3 to drive metastasis have not yet been determined. Therefore, we have searched for gene expression changes using a highly efficient cDNA-RDA subtractive hybridization method (17) in β3-positive and -negative melanoma cell populations. Our results indicate that β3 overexpression up-regulates molecules associated with adhesion and that osteonectin/SPARC is critical for progression of melanoma cells from nontumorigenic RGP to tumorigenic VGP. Because osteonectin expression has previously been associated with progression in both human and mouse melanoma (18, 19) and is a marker correlated with increased incidence of distant metastases and decreased survival (20), we examined the modulation of αvβ3 integrin function on osteonectin gene induction in human melanoma cells.

Cell Culture and Adenovirus-mediated Gene Transduction.

Human melanoma cell lines (15) were maintained in 4 parts MCDB 153 and 1 part L-15 medium supplemented with 5 μg/ml insulin and 2% fetal bovine serum. Construction of the adenoviral β3/Ad5 and control LacZ/Ad5 vectors from dl-312 has been described previously (15). For viral infection, RGP melanoma cells were cultured in 150-mm dishes to 90% confluence and incubated with 20 pfu/cell in 20 ml of serum-free medium at 37°C for 4 h. Unbound virus was removed by washing cells with 10 ml of fresh medium before incubation in serum-containing medium for 48–60 h.

Flow Cytometry.

Cells transduced with viral vectors were collected with versene, washed with DMEM, and resuspended in serum-free DMEM. Cells were seeded in duplicates at 2 × 105 cells/well in a 96-well plate and incubated with 10 μg/ml anti-β3 mouse mAb SAP (11). After 45 min of incubation at 4°C with gentle shaking, cells were washed with PBS containing 0.1% BSA and 0.1% sodium azide to remove unbound antibodies before staining with 10 μg/ml FITC-conjugated rabbit antimouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA). After washing, cells were resuspended and analyzed by fluorescence-activated cell sorting using an Ortho Cytofluorograf 50 H connected to a 2150 Data Handling System (Ortho Diagnostics, Westwood, MA).

Isolation of RNA, cDNA Synthesis, and RDA.

Transduced cells were harvested and lysed using NP40 (21). DNA was removed by pelleting the nuclei, and total cytoplasmic RNA was prepared from the cleared supernatant by proteinase K/SDS treatment. Total RNA was recovered by phenol/chloroform extraction and ethanol precipitation. Polyadenylated RNA was selected using an Oligotex mRNA Midi Kit (Qiagen, Valencia, CA), and 1 μg was used as template for double-strand cDNA synthesis using a Life Technologies, Inc. SuperScript Choice System with 1 μg of oligodeoxythymidylic acid and 100 ng of random N6 primers. RDA is a subtractive hybridization procedure coupled with PCR amplification used to rapidly identify sequence differences between complex populations of DNA molecules (22, 23). The complexity of the hybridizing material is reduced by restriction enzyme digestion, ligation of adaptors, and PCR of an amplimer product for sampling. Differential hybridization was coupled with amplification of DPs (DP1, DP2, and DP3) after each round due to removal of the adaptors from the “driver” but not the “tester” DNA representations. This approach results in the enrichment of sequences unique to the tester sample. The procedure was performed as described previously (22, 23), with the following modifications: oligonucleotides were purified by high-pressure liquid chromatography (24); mung bean nuclease digestion of amplification products was omitted (25); and excess and cleaved adaptors from the RDA amplicons made from DpnII-digested cDNA representation and tester amplification products were purified using Millipore microcon 100 filters (26). Subtractive hybridizations were performed in 4-μl volumes using 40 μg of driver with the ratio of tester/cycle set at 1:100 for DP1, 1:500 for DP2, and 1:4000 for DP3. Twenty PCR amplification cycles were used to generate each of the DP products. Supplementation of the driver with β3 coding sequences from the pcDNA1-β3 plasmid (a gift from Dr. D. Cheresh, Scripps Research Institute, La Jolla, CA) was used to suppress amplification of the β3 gene sequences derived from the adenovirus vector expression by adding 5 μg (22) of β3 plasmid cDNA generated as driver (27) to each round of hybridization.

Clone Isolation and Sequencing.

Subtracted DP2 RDA products were digested with DpnII, purified using a microcon 100 filter, and cloned into the BamHI site of pBluescript SK+ (Stratagene, La Jolla, CA). DP3 products were digested with DpnII, separated by agarose gel electrophoresis, and purified using QiaexII resin (Qiagen) before cloning. DNA samples were sequenced from mini-extracted plasmids using BigDye terminator automated sequencing. Sequences obtained were compared with DNA databases using the BLASTN program (Australian National Genomic Information Service, University of Sydney, Sydney, Australia).

Radiolabeling of Probe and Northern Blotting.

Northern blotting was performed by fractionating 10 μg of each RNA sample on formaldehyde-agarose gels (21) followed by soaking in 10× SSC, transfer to a nylon membrane, UV cross-linking, and hybridization. Efficiency of transfer and position of 18S and 28S rRNA bands were determined by UV shadowing of the nylon membrane. Loading and integrity of the RNA samples were tested using a GAPDH gene probe. Insert probes were prepared by radiolabeled PCR (28) using 0.1 ng of plasmid clone DNA as template with SK (Stratagene) and KS-20mer (5′-CCTCGAGGTCGACGGTATCG-3′) primers and fractionated on Sephadex G-50 Nick columns (Pharmacia, Piscataway, NJ). The PstI/XhoI fragment from the SPARC full-length cDNA (the pBS clone was a gift from R. Nischt, University of Cologne, Cologne, Germany) and GAPDH probes were generated by random labeling using a high prime DNA labeling kit and fractionated on Sephadex G-50 Quick Spin columns (Boehringer Mannheim; Roche, Indianapolis, IN).

Western Blot of Cell Extracts, MAPK, PKC, and Antibody Inhibition.

SBcl2 and WM1552 cells were seeded in duplicate in 6-well plates and grown to subconfluence. After adenoviral gene transduction, the cells were overlaid with growth medium for 24 h and washed with serum-free medium, and medium was replaced with serum-free medium containing 5 μg/ml insulin for an additional 48 h. MAPK inhibitors PD98059 and wortmannin (Calbiochem, San Diego, CA) were added at 10 μm and 100 nm, respectively, for the 48-h time period. Cells were harvested by scraping and vigorous pipetting on ice in 150 μl of MAPK extraction buffer consisting of 10 mm Tris-acetate (pH 8.0), 0.5% NP40 (BDH Chemicals, Darmstadt, Germany), 1 mm EDTA, 2 mm phenylmethylsulfonyl fluoride, and 1 μg/ml each of aprotinin, leupeptin, and pepstatin, and insoluble material was removed by centrifugation at 13,000 × g for 10 min at 4°C. Total protein concentrations were estimated with a BCA kit (Bio-Rad, Hercules, CA) using BSA as standard. Equal amounts of protein were separated by 10% SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes (NEN Polyscreen, Boston, MA). Membranes were blocked for 1 h at room temperature with 5% skim milk in 1× TBS/0.1% Tween 20 that was included in each antibody-binding step. Incubation with 1 μg/ml anti-osteonectin mouse mAb (Hematological Technologies Inc., Essex Junction, VT) overnight at 4°C was followed by rinsing in 1× TBS/0.1% Tween 20 and incubation with a 1:1000 dilution of peroxidase-labeled goat antimouse IgG (Kirkegaard & Perry Laboratories, Gaithersburg, MD). After extensive washing in 1× TBS/0.1% Tween 20, immunoreactive bands were identified by enhanced chemiluminescence (Amersham, Arlington Heights, IL).

A2058 cells were grown until approximately 80% confluence and washed with serum-free medium, and medium was replaced with serum-free medium containing 5 μg/ml insulin for 24 h. Cells were harvested in PBS without trypsin and transferred to the wells of a 6-well non-tissue culture plate (Falcon) in which the wells had been previously coated in duplicate for 4 h at 37°C with the following proteins: 5 μg/ml LM609 αvβ3 monoclonal blocking mAb (Chemicon Inc., Temecula, CA); 0.1 mg/ml poly-l-lysine; 2 μg/ml vitronectin; or 0.1mg/ml heat-denatured BSA. After incubation with serum-free medium containing 5 μg/ml insulin for an additional 24 h, cell extracts were prepared using MAPK extraction buffer. PKC inhibition of cells was performed in duplicate coated wells by treatment with 25 nm calphostin C (Sigma Chemical Co., St. Louis, MO). Protein concentration of the samples was normalized by incubating the membranes in a neat serum solution of intermediate filament antibody (29) for 1 h at room temperature, followed by washes in 1× PBS/0.05% Tween 20. The membrane was then treated with peroxidase-labeled sheep antimouse IgG as described above.

Cell Adhesion Assay and in Vitro Reconstruction of Human Skin.

Cell adhesion assays were performed with melanoma cells infected with either LacZ/Ad5 or β3/Ad5. Cells were harvested after 48 h, counted, and seeded in triplicate at 104 cells/well in a 96-well non-tissue culture plate that had previously been coated overnight at 4°C with the following matrix proteins: 5 μg/ml vitronectin; 10 μg/ml fibronectin; 5 μg/ml osteonectin (Calbiochem); or heat-denatured BSA. Cells were allowed to adhere for 1 h at 37°C, washed gently in PBS to remove unbound cells, fixed in 3% paraformaldehyde, and stained with crystal violet as described previously (30). After extensive washing to remove nonspecifically bound dye, cells were lysed in 1% SDS, and absorbance was read at 595 nm.

Skin reconstructs were generated as described previously (31). They consist of a “dermis” of collagen type I embedded with fibroblasts and an “epidermis” with keratinocytes and melanocytes or melanoma cells. The melanoma cells had been transduced with adenoviral vector 48 h before mixing with the keratinocytes at a 1:5 ratio. After 10–14 days of air exposure, during which the keratinocytes formed multiple layers, skin reconstructs were harvested by fixing overnight with 4% paraformaldehyde, dehydration, and embedding in paraffin. Sections were stained with H&E for histological analysis.

Ectopic Expression of β3 Integrin in RGP Primary Melanoma Cells.

The human RGP-like melanoma cell lines SBcl2 and WM1552C do not express β3 integrin (15) and were used for adenoviral vector-mediated transduction of β3. Cell surface expression of β3 was assayed 48 h after transduction with a viral load of 1, 10, and 20 pfu/cell. Fluorescence-activated cell-sorting analysis (Fig. 1) demonstrated that transduced cells expressed the β3 molecule. The highest level of expression was found at 20 pfu/cell, which was consistent with the results of previous gene transduction studies (15). For additional experiments, the cells transduced with 20 pfu/cell were chosen to look for changes associated with β3 expression.

RDA of Ectopic β3 Integrin-expressing RGP Melanoma Cells.

RGP-like SBcl2 and WM1552C melanoma cells that were either noninfected or transduced with β3 or control vectors were used to provide cDNA products for RDA analysis. Three cycles of hybridization and selection were performed using the SBcl2 β3/Ad5 PCR amplimer product as tester against the dl-312 PCR amplimer product as driver to identify those genes that were induced in the invading cells. The initial product fragments appearing in the β3/Ad5 tester DP3 cycle were blotted by Southern analysis and identified as arising from the β3 gene coding region.

To avoid amplification of the β3 gene DPs detected in our initial analysis, we doped the driver sample with β3 coding sequences (22, 23, 27). The PCR fragment profiles resulting from the β3 coding sequence-doped RDA products were separated by agarose gel electrophoresis and are shown in Fig. 2. The complex pattern of the initial driver (Lane 2) and tester (Lane 3) cDNA amplification products was sequentially reduced by subtractive hybridization coupled with PCR amplification as seen in the first (Lane 4, DP1)-, second (Lane 5, DP2)-, and third (Lane 6, DP3)-round DPs. A similar experiment was performed using the WM1552C cell line, in which only two rounds of subtractive hybridization were completed. The DP2 fragments generated from the SBcl2 and WM1552C cDNA were cloned as separate pools, whereas the SBcl2 DP3 fragments were isolated from the gel before cloning and sequencing.

Identification of Differentially Expressed Genes in RGP Melanoma Cells Adopting a VGP Phenotype.

Distinct fragments ranging from 300–600 bp in the SBcl2 DP3 samples (Fig. 2, Lane 6) were identified by comparison of their sequence with published nucleotide databases. The most intense of these fragments was a 450-bp β3/Ad5 vector SV40 sequence that was not competed out when using the β3 coding region. This served as an internal control for the fidelity of the RDA assay. Sequencing the DP3 cloned fragments then identified 14 known genes and 1 insert of unknown identity (Table 1). Seventy-two SBcl2 and 24 WM1552C DP2 isolates were sequenced to determine the range of genes whose expression levels may have been affected by β3 (data not shown). Fifty-five genes were identified, some of which overlapped with those found in the DP3 cycle. When confirming the induction of several genes by Northern blotting including KIAA0108, integrins α5 and β5, and Smad7, expression of osteonectin mRNA (32) was also observed in the β3-transduced SBcl2 cells (Fig. 3 A).

Osteonectin Protein Induction through β3 Integrin Expression.

To demonstrate that the induction of osteonectin at the mRNA level also reflected an increase in expression of osteonectin protein in β3-transduced cells, we analyzed protein by Western blot. Fig. 3 B shows the results of both SBcl2 and WM1552C cells transduced with either LacZ (Lanes 1 and 7) or β3 (Lanes 2 and 8) and sampled 72 h postinfection with anti-osteonectin antibodies. A 3–4-fold increase in protein levels was seen in both cell lines. To test for the involvement of the MAPK pathway in the β3-mediated induction of osteonectin, SBcl2 cells were treated with the MAPK inhibitors PD98059 (Lanes 3 and 4) and wortmannin (Lanes 5 and 6). Transduction with β3 continued to show increased levels of osteonectin synthesis relative to LacZ, indicating that β3-induced changes do not affect the MAPK pathway.

The A2058 melanoma cell line is known to express high levels of αvβ3 integrin and can be inhibited in attachment to substratum-bound vitronectin using the LM609 αvβ3 blocking mAb (33). The expression levels of osteonectin protein produced in these cells were assessed after plating on vitronectin, LM609 mAb, poly-l-lysine, or heat-denatured BSA-coated plastic dishes. The levels of osteonectin were reduced 10-fold on culturing these cells with LM609 blocking antibody as compared with treatment with vitronectin alone; also, the antibody was able to reduce osteonectin production when coated together with vitronectin (Fig. 3, C and D). Inhibition of αvβ3 function supports the induction of osteonectin through active β3 integrin expression. The calphostin C nonspecific PKC inhibitor was used to determine whether this pathway was involved in the expression of osteonectin through αvβ3 integrin-vitronectin binding, as has been seen for αvβ3 integrin-LM609 ligation induction of other molecules (34); however, expression levels were not affected (Fig. 3 D).

In Vitro Effect of Osteonectin on Melanoma Cell Adhesion to ECM Proteins.

To test the effects of osteonectin protein on cell function, we examined the adhesion of SBcl2 cells after β3 overexpression. Vitronectin, fibronectin, osteonectin, or heat-denatured BSA was used to coat plastic dishes. Attachment of LacZ- or β3-transduced SBcl2 cells was tested. Cells expressing β3 integrin showed a 2-fold increase in adhesion to both vitronectin- and fibronectin-coated wells as compared with LacZ-expressing cells; however, osteonectin alone did not significantly affect adhesion (Fig. 4). When vitronectin and osteonectin were mixed before coating, cell binding was reduced by >30%. Inhibition of adhesion of the vitronectin/osteonectin mixture was not seen in LacZ-transduced SBcl2 cells nor when osteonectin was mixed with fibronectin.

Induction of Dermal Invasion of SBcl2 Cells Overexpressing β3 Integrin.

Osteonectin expression by melanoma cells after β3 integrin transduction was determined in an orthotopic melanoma invasion model. SBcl2 cells transduced with the control vector LacZ were not invasive (15). The cells did not express β3 and showed only a weak presence of osteonectin (Fig. 5, A and C). In contrast, SBc12 cells transduced with β3 were deeply invasive in the dermis (Fig. 5, B and D). The cells expressed β3 and osteonectin, confirming the up-regulation of osteonectin by β3 seen in monolayer cultures.

A multitude of cellular changes occur during transition from RGP to VGP melanoma. Dissection of gene activation and suppression steps is required to fully understand the metastatic process whereby melanoma cells first gain the ability to invade through the dermis, intravasate, and survive beyond the primary site of the skin. Currently, β3 integrin expression status in melanoma is a leading indicator for this transition (3). Although it is known that αvβ3 dimer formation promotes melanoma tumorigenicity through escape from apoptosis (15, 16), other pathways remain to be investigated. This study has found a range of genes whose expression was up-regulated on ectopic expression of β3 integrin. SBcl2 and WM1552C RGP melanoma cells were used as a test system to profile gene expression responsible for aggressive behavior. Both DP3 and DP2 cycle gene fragment products amplified from the cDNA-RDA procedure were analyzed to increase the range of genes whose expression may be affected by induction of the β3 integrin subunit (25, 35). The DP3 cycle favors the identification of more abundant gene transcripts with most dramatic changes in gene expression.

Fourteen genes and one unknown transcript were selectively amplified as products of the DP3 cDNA-RDA procedure in the SBcl2 cells transduced with β3. Of these genes, osteonectin is expressed by melanoma cells (32) but not by normal melanocytes. This pattern has been associated with progression in both human and mouse melanoma (18, 19). Expression of osteonectin has been correlated with increased incidence of distant metastases and decreased survival (20); however a relationship with primary tumor growth has not yet been determined. Studies of invasive meningiomas and prostate carcinomas have also demonstrated a correlation between osteonectin production and tumor progression (36, 37). Addition of osteonectin protein to culture medium of carcinoma cells stimulated cell migration and invasion (38, 39). Antisense RNA down-regulation of osteonectin transcripts in melanoma cells reduced the invasive and adhesive capacities in vitro and inhibited tumor formation in vivo(40). The loss of aggressive growth properties was associated with loss of matrix metalloproteinase-2 expression. Osteonectin can activate matrix metalloproteinase-2 in invasive breast cancer cells (41). Thus, osteonectin may be involved in cell-matrix interactions during tissue remodeling, morphogenesis, migration, and proliferation by acting as an antiadhesive molecule stimulating invasion; it may also modulate angiogenesis (42).

VGP melanoma cells have acquired the ability to invade the dermis. This process requires separation of the melanoma cells from neighboring keratinocytes, attachment to the basement membrane, proteolytic degradation of the basal lamina, and proliferation in the dermis. The induction of osteonectin during this phase is expected to provide the cells with the appropriate matrix to migrate on.

Our finding that osteonectin inhibits vitronectin-mediated binding of melanoma cells suggests that the melanoma cells can readily detach from substrate despite expression of the vitronectin receptor αvβ3. Because melanoma cells also secrete vitronectin (43), it appears that the secretion of these two molecules establishes a balance of adhesion and counter-adhesion. Similarly, binding of osteonectin to vitronectin can modulate attachment of endothelial cells (44). In human glioma cells, expression of osteonectin correlates with the angiogenic potential of the cells (45).

It is evident that integrins are involved in adhesion and signaling. Stimulation of integrins triggers intracellular signaling events that can be integrated with those originating from growth factor receptors to organize the cytoskeleton, stimulate MAPK cascades, and regulate gene expression (2, 33, 34, 46). Ligand binding to integrins can activate protein kinases, in particular, activation and autophosphorylation of the cytoplasmic focal adhesion kinase. β3 integrin appears to participate in specific signaling events that are poorly delineated. Potentially, there are cell type-specific functions of signaling pathways (47). An increase in osteonectin protein production that accompanies β3 integrin expression in RGP melanoma cells is independent of MAPK and PKC, suggesting that other, as yet undefined pathways are important. It is notable that the small GTPase RhoC and osteonectin are up-regulated in highly metastatic melanoma cells, suggesting that enhanced expression of genes involved in ECM assembly (48) are critical. Additional studies will need to establish a link between RhoC signaling and osteonectin expression in melanoma.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

      
1

During this work, R. A. S. was supported by an SSP Award from the University of Queensland and a Queensland Cancer Fund travel grant; B. B. G. is a Queensland Cancer Fund Ph.D. scholar. The Center for Functional and Applied Genomics is a Special Research Center of the Australian Research Council. This work was also Supported by NIH Grants CA-47159 and CA-10815 and the Australian NHMRC-102565.

            
3

The abbreviations used are: RGP, radial growth phase; DP, difference product; mAb, monoclonal antibody; MAPK, mitogen-activated protein kinase; pfu, plaque-forming unit; RDA, representational difference analysis; TBS, Tris-buffered saline; VGP, vertical growth phase; PKC, protein kinase C; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ECM, extracellular matrix.

Fig. 1.

Ectopic expression of β3 integrin in RGP-like SBcl2 melanoma cells. Cell surface expression of the β3 integrin subunit was assayed by flow cytometry after sequential incubation with anti-β3 mAb SAP and FITC-conjugated goat antimouse IgG. The plot shows the relative cell number of SBcl2 cells (Y axis) after dl-312 (shaded area, β3−) or β3/Ad5 (dotted curve, β3+) adenoviral vector infection against the log fluorescence (X axis). Almost all β3-transduced cells expressed the β3 integrin subunit. A nonbinding control antibody showed the same results as SAP on β3-negative cells.

Fig. 1.

Ectopic expression of β3 integrin in RGP-like SBcl2 melanoma cells. Cell surface expression of the β3 integrin subunit was assayed by flow cytometry after sequential incubation with anti-β3 mAb SAP and FITC-conjugated goat antimouse IgG. The plot shows the relative cell number of SBcl2 cells (Y axis) after dl-312 (shaded area, β3−) or β3/Ad5 (dotted curve, β3+) adenoviral vector infection against the log fluorescence (X axis). Almost all β3-transduced cells expressed the β3 integrin subunit. A nonbinding control antibody showed the same results as SAP on β3-negative cells.

Close modal
Fig. 2.

Resolution of differentially expressed gene fragments detected by cDNA-RDA analysis of SBcl2 melanoma cells overexpressing the β3 integrin subunit. The 1.5% agarose gel was stained with ethidium bromide after electrophoretic separation. Lane 1, øX174 DNA/HaeIII markers; Lanes 2 and 3, initial cDNA amplicon representations generated from dl-312- and β3-SBcl2-transduced cells used as driver and tester populations, respectively. Lanes 4–6, DP arising after amplification of each subtractive hybridization step at the following driver:tester ratios: DP1, 1:100 (Lane 4); DP2, 1:500 (Lane 5); and DP3, 1:4000 (Lane 6). The driver was supplemented with 5 μg of β3 coding sequence cDNA amplicons to suppress β3 gene fragment amplification.

Fig. 2.

Resolution of differentially expressed gene fragments detected by cDNA-RDA analysis of SBcl2 melanoma cells overexpressing the β3 integrin subunit. The 1.5% agarose gel was stained with ethidium bromide after electrophoretic separation. Lane 1, øX174 DNA/HaeIII markers; Lanes 2 and 3, initial cDNA amplicon representations generated from dl-312- and β3-SBcl2-transduced cells used as driver and tester populations, respectively. Lanes 4–6, DP arising after amplification of each subtractive hybridization step at the following driver:tester ratios: DP1, 1:100 (Lane 4); DP2, 1:500 (Lane 5); and DP3, 1:4000 (Lane 6). The driver was supplemented with 5 μg of β3 coding sequence cDNA amplicons to suppress β3 gene fragment amplification.

Close modal
Fig. 3.

Identification of osteonectin gene induction in melanoma cells expressing β3 integrin. A, induction of the osteonectin mRNA transcript was confirmed by Northern blotting in which SBcl2 cells were noninfected (−), transduced with LacZ (Z) or β3 (β3), and then harvested 72 h posttransduction. The relative osteonectin band intensity ratio is indicated below each lane, normalized against GAPDH using the Bio-Rad Molecular Analysist v2.1 program. B, induction of osteonectin protein by ectopic expression of β3 integrin in SBcl2 and WM1552C melanoma cells. Cells were transduced with LacZ (Z) or β3 (β3) and harvested 72 h later. Samples equivalent to approximately 104 cells were separated by 10% SDS-PAGE and then electroblotted onto polyvinylidene difluoride membranes. Detection of the 43-kDa osteonectin protein using mAb AON-5031 was performed by enhanced chemiluminescence. Increased osteonectin protein production expressed as a ratio between each lane is seen after β3 transduction (Lanes 1 and 2; Lanes 7 and 8) in both SBcl2 and WM1552C cells. This induction was not significantly quenched by incubating transduced SBcl2 cells with MAPK inhibitor PD98059 (PD, Lanes 3 and 4) or MAPK inhibitor wortmannin (W, Lanes 5 and 6). C, osteonectin protein levels in A2058 melanoma cells plated on dishes coated with 5 μg/ml LM609 αvβ3 blocking mAb, 0.1 mg/ml poly-l-lysine, 2 μg/ml vitronectin (V), 0.1 mg/ml heat-denatured BSA, and LM609 mAb plus vitronectin were harvested 24 h later. The relative osteonectin band intensity ratio is indicated below each lane, normalized against intermediate filament antigen mAb quantitating the decreased expression seen plating on LM609 mAb. D, osteonectin protein levels in A2058 melanoma cells plated on dishes coated with LM609 mAb or vitronectin (V) in the presence (+) or absence (−) of calphostin C.

Fig. 3.

Identification of osteonectin gene induction in melanoma cells expressing β3 integrin. A, induction of the osteonectin mRNA transcript was confirmed by Northern blotting in which SBcl2 cells were noninfected (−), transduced with LacZ (Z) or β3 (β3), and then harvested 72 h posttransduction. The relative osteonectin band intensity ratio is indicated below each lane, normalized against GAPDH using the Bio-Rad Molecular Analysist v2.1 program. B, induction of osteonectin protein by ectopic expression of β3 integrin in SBcl2 and WM1552C melanoma cells. Cells were transduced with LacZ (Z) or β3 (β3) and harvested 72 h later. Samples equivalent to approximately 104 cells were separated by 10% SDS-PAGE and then electroblotted onto polyvinylidene difluoride membranes. Detection of the 43-kDa osteonectin protein using mAb AON-5031 was performed by enhanced chemiluminescence. Increased osteonectin protein production expressed as a ratio between each lane is seen after β3 transduction (Lanes 1 and 2; Lanes 7 and 8) in both SBcl2 and WM1552C cells. This induction was not significantly quenched by incubating transduced SBcl2 cells with MAPK inhibitor PD98059 (PD, Lanes 3 and 4) or MAPK inhibitor wortmannin (W, Lanes 5 and 6). C, osteonectin protein levels in A2058 melanoma cells plated on dishes coated with 5 μg/ml LM609 αvβ3 blocking mAb, 0.1 mg/ml poly-l-lysine, 2 μg/ml vitronectin (V), 0.1 mg/ml heat-denatured BSA, and LM609 mAb plus vitronectin were harvested 24 h later. The relative osteonectin band intensity ratio is indicated below each lane, normalized against intermediate filament antigen mAb quantitating the decreased expression seen plating on LM609 mAb. D, osteonectin protein levels in A2058 melanoma cells plated on dishes coated with LM609 mAb or vitronectin (V) in the presence (+) or absence (−) of calphostin C.

Close modal
Fig. 4.

Differential modulation of cell adhesion to ECM proteins by ectopic expression of β3 integrin in SBcl2 melanoma cells. Dishes were coated with 5 μg/ml vitronectin, 10 μg/ml fibronectin, 5 μg/ml osteonectin, or heat-denatured BSA. Cell attachment was measured after 1 h at 37°C, when cells were fixed and stained with crystal violet. Absorbance at 595 nm was measured.

Fig. 4.

Differential modulation of cell adhesion to ECM proteins by ectopic expression of β3 integrin in SBcl2 melanoma cells. Dishes were coated with 5 μg/ml vitronectin, 10 μg/ml fibronectin, 5 μg/ml osteonectin, or heat-denatured BSA. Cell attachment was measured after 1 h at 37°C, when cells were fixed and stained with crystal violet. Absorbance at 595 nm was measured.

Close modal
Fig. 5.

Immunohistochemical detection of osteonectin in invading SBcl2 melanoma cells in human skin reconstructs. Melanoma cells transduced with lacZ or β3 were mixed with keratinocytes and seeded onto a collagen matrix with embedded fibroblasts. When the three-dimensional structures were exposed to air, keratinocytes differentiated to form multiple layers. After 2 weeks, skin reconstructs were fixed and stained with anti-osteonectin and anti-β3 mAb. A, SBcl2 cells transduced with lacZ and stained for β3 to show nonspecific background intensity. B, SBcl2 cells transduced with β3 and stained for β3, with the diffuse signal surrounding the cells evident at the dermal epidermal junction and invading the artificial dermis. C, SBcl2 cells transduced with lacZ and stained for osteonectin. D, SBcl2 cells transduced with β3 and stained for osteonectin, with the diffuse signal surrounding the cells evident at the dermal epidermal junction and invading the artificial dermis.

Fig. 5.

Immunohistochemical detection of osteonectin in invading SBcl2 melanoma cells in human skin reconstructs. Melanoma cells transduced with lacZ or β3 were mixed with keratinocytes and seeded onto a collagen matrix with embedded fibroblasts. When the three-dimensional structures were exposed to air, keratinocytes differentiated to form multiple layers. After 2 weeks, skin reconstructs were fixed and stained with anti-osteonectin and anti-β3 mAb. A, SBcl2 cells transduced with lacZ and stained for β3 to show nonspecific background intensity. B, SBcl2 cells transduced with β3 and stained for β3, with the diffuse signal surrounding the cells evident at the dermal epidermal junction and invading the artificial dermis. C, SBcl2 cells transduced with lacZ and stained for osteonectin. D, SBcl2 cells transduced with β3 and stained for osteonectin, with the diffuse signal surrounding the cells evident at the dermal epidermal junction and invading the artificial dermis.

Close modal
Table 1

Gene fragments induced in SBcl2 RGP melanoma cells by ectopic expression of β3 integrin

List of genes identified by RDA in β3 integrin-expressing SBcl2 RGP-like melanoma cells compared with non-β3 integrin-expressing cells. The genes were identified by cloning and sequencing of fragments isolated by RDA. More detailed information can be obtained from http://www.wistar.upenn.edu/herlyn

CloneAccession no.IdentityInsert size (bp)Homology region
NM002205 Integrin α5 384 1943–2325 
NM005904 Smad7 329 1509–1815 
10, 21, 22 D14696 Human KIAA0108 343 233–573 
and 31 NM008640 Homologous to mouse Golgi 4-transmembrane spanning transporter MTP  831–1171 
25 and 26 NM003118 Secreted protein, acidic, cysteine-rich (osteonectin; SPARC) 296 963–1258 
29 AL117442 Clone DKFZp434P106 296 995–1290 
32 NM003299 Tumor rejection antigen (gp96) 1 (TRA1) 310 2239–2528 
34 NM001096 ATP citrate lyase (ACLY) 305 34–306 
35  Unknown 310  
39 NM003333 Ubiquitin A-52 residue ribosomal protein fusion product 1 (UBA52) 279 43–322 
45 AF092131 51-kDa subunit of NADH dehydrogenase 272 33–284 
46 NM002213 Integrin β5 (ITGB5) 291 2024–2315 
47 NM001642 Amyloid β (A4) precursor-like protein 2 (APLP2) 282 1579–1861 
48a AF151832 CGI-74 protein 272 1446–1698 
 AF151817 CGI-54 protein  1420–1673 
52 NM003001 Succinate dehydrogenase complex, subunit C, integral membrane protein, 15-kDa (SDHC) 504 662–1152 
53 and 59 L80004 u1A-IC/SNRPN transcript 515 360–868 
CloneAccession no.IdentityInsert size (bp)Homology region
NM002205 Integrin α5 384 1943–2325 
NM005904 Smad7 329 1509–1815 
10, 21, 22 D14696 Human KIAA0108 343 233–573 
and 31 NM008640 Homologous to mouse Golgi 4-transmembrane spanning transporter MTP  831–1171 
25 and 26 NM003118 Secreted protein, acidic, cysteine-rich (osteonectin; SPARC) 296 963–1258 
29 AL117442 Clone DKFZp434P106 296 995–1290 
32 NM003299 Tumor rejection antigen (gp96) 1 (TRA1) 310 2239–2528 
34 NM001096 ATP citrate lyase (ACLY) 305 34–306 
35  Unknown 310  
39 NM003333 Ubiquitin A-52 residue ribosomal protein fusion product 1 (UBA52) 279 43–322 
45 AF092131 51-kDa subunit of NADH dehydrogenase 272 33–284 
46 NM002213 Integrin β5 (ITGB5) 291 2024–2315 
47 NM001642 Amyloid β (A4) precursor-like protein 2 (APLP2) 282 1579–1861 
48a AF151832 CGI-74 protein 272 1446–1698 
 AF151817 CGI-54 protein  1420–1673 
52 NM003001 Succinate dehydrogenase complex, subunit C, integral membrane protein, 15-kDa (SDHC) 504 662–1152 
53 and 59 L80004 u1A-IC/SNRPN transcript 515 360–868 
a

Double insert.

1
Giancotti F. G. Integrin signaling: specificity and control of cell survival and cell progression.
Curr. Opin. Cell Biol.
,
9
:
691
-700,  
1997
.
2
Kumar C. Signaling by integrin receptors.
Oncogene
,
17
:
1365
-1373,  
1998
.
3
Seftor R. E. B. Role of the β3 integrin subunit in human primary melanoma progression.
Am. J. Pathol.
,
153
:
1347
-1351,  
1998
.
4
Meier F., Satyamoorthy K., Nesbit M., Hsu M. Y., Schittek B., Garbe C., Herlyn M. Molecular events in melanoma development and progression.
Front. Biosci.
,
3
:
D1005
-D1010,  
1998
.
5
Bittner M., Meltzer P., Chen Y., Jiang Y., Seftor E., Hendrix M., Radmacher M., Simon R., Yakhini Z., Ben-Dor A., Sampas N., Dougherty E., Wang E., Marincola F., Gooden C., Lueders J., Glatfelter A., Pollock P., Carpten J., Gillanders E., Leja D., Dietrich K., Beaudry C., Berens M., Alberts D., Sondak V., Hayward N., Trent J. Molecular classification of cutaneous malignant melanoma by gene expression profiling.
Nature (Lond.)
,
406
:
536
-540,  
2000
.
6
Satyamoorthy K., DeJesus E., Linnenbach A. J., Kraj B., Kornreich D. L., Rendle S., Elder D. E., Herlyn M. Melanoma cell lines from different stages of progression and their biological and molecular analyses.
Melanoma Res.
,
7 (Suppl. 2)
:
S35
-S42,  
1997
.
7
Kath R., Jambrosic J. A., Holland L., Rodeck U., Herlyn M. Development of invasive and growth factor-independent cell variants from primary human melanomas.
Cancer Res.
,
51
:
2205
-2211,  
1991
.
8
Albelda S. M., Mette S. A., Elder D. E., Stewart R., Damjanovich L., Herlyn M., Buck C. A. Integrin distribution in malignant melanoma: association of the β3 subunit with tumor progression.
Cancer Res.
,
50
:
6757
-6764,  
1990
.
9
Hieken T. J., Farolan M., Ronan S. G., Shilkaitis A., Wild L., Das G. T. β3 integrin expression in melanoma predicts subsequent metastasis.
J. Surg. Res.
,
63
:
169
-173,  
1996
.
10
Natali P. G., Hamby C. V., Felding H. B., Liang B., Nicotra M. R., Di Filippo F., Giannarelli D., Temponi M., Ferrone S. Clinical significance of αvβ3 integrin and intercellular adhesion molecule-1 expression in cutaneous malignant melanoma lesions.
Cancer Res.
,
57
:
1554
-1560,  
1997
.
11
Van Belle P. A., Elenitsas R., Satyamoorthy K., Wolfe J. T., Guerry D. t., Schuchter L., Van Belle T. J., Albelda S., Tahin P., Herlyn M., Elder D. E. Progression-related expression of β3 integrin in melanomas and nevi.
Hum. Pathol.
,
30
:
562
-567,  
1999
.
12
Gouon V., Tucker G. C., Kraus-Berthier L., Atassi G., Kieffer N. Up-regulated expression of the β3 integrin and the 92-kDa gelatinase in human HT-144 melanoma cell tumors grown in nude mice.
Int. J. Cancer
,
68
:
650
-662,  
1996
.
13
Seftor R. E. B., Seftor E. A., Gehlsen K. R., Stetler-Stevenson W. G., Brown P. D., Ruoslahti E., Hendrix M. J. C. Role of the αVβ3 integrin in human melanoma cell invasion.
Proc. Natl. Acad. Sci. USA
,
89
:
1557
-1561,  
1992
.
14
Mitjans F., Sander D., Adan J., Sutter A., Martinez J. M., Jaggle C. S., Moyano J. M., Kreysch H. G., Piulats J., Goodman S. L. An anti-αv-integrin antibody that blocks integrin function inhibits the development of a human melanoma in nude mice.
J. Cell Sci.
,
108
:
2825
-2838,  
1995
.
15
Hsu M. Y., Shih D. T., Meier F. E., Van Belle P., Hsu J. Y., Elder D. E., Buck C. A., Herlyn M. Adenoviral gene transfer of β3 integrin subunit induces conversion from radial to vertical growth phase in primary human melanoma.
Am. J. Pathol.
,
153
:
1435
-1442,  
1998
.
16
Petitclerc E., Stromblad S., von Schalscha T. L., Mitjans F., Piulats J., Montgomery A. M., Cheresh D. A., Brooks P. C. Integrin αvβ3 promotes M21 melanoma growth in human skin by regulating tumor cell survival.
Cancer Res.
,
59
:
2724
-2730,  
1999
.
17
Lisitsyn N. A. Representational difference analysis: finding the differences between genomes.
Trends Genet.
,
11
:
303
-307,  
1995
.
18
Ledda F., Bravo A. I., Adris S., Bover L., Mordoh J., Podhajcer O. L. The expression of the secreted protein acid and rich in cysteine (SPARC) is associated with the neoplastic progression of human melanoma.
J. Investig. Dermatol.
,
108
:
210
-214,  
1997
.
19
Kato Y., Frankenne F., Noel A., Sakai N., Nagashima Y., Koshika S., Miyazaki K., Foidart J. M. High production of SPARC/osteonectin/BM-40 in mouse metastatic B16 melanoma cell lines.
Pathol. Oncol. Res.
,
6
:
24
-26,  
2000
.
20
Massi D., Franchi A., Borgognoni L., Reali U. M., Santucci M. Osteonectin expression correlates with clinical outcome in thin cutaneous malignant melanomas.
Hum. Pathol.
,
30
:
339
-344,  
1999
.
21
Ausubel F. M., Brent R., Kingston R. E., Moore D. D., Seidman J. G., Smith J. A., Struhl K. .
Current Protocols in Molecular Biology
, Green Publishing Associates and Wiley Interscience New York  
1989
.
22
Hubank M., Schatz D. G. Identifying differences in mRNA expression by representational difference analysis of cDNA.
Nucleic Acids Res.
,
22
:
5640
-5648,  
1994
.
23
Braun B. S., Frieden R., Lessnick S. L., May W. A., Denny C. T. Identification of target genes for the Ewing’s sarcoma EWS/FLI fusion protein by representational difference analysis.
Mol. Cell. Biol.
,
15
:
4623
-4630,  
1995
.
24
O’Neill M. J., Sinclair A. H. Isolation of rare transcripts by representational difference analysis.
Nucleic Acids Res.
,
25
:
2681
-2682,  
1997
.
25
Welford S. M., Gregg J., Chen E., Garrison D., Sorensen P. H., Denny C. T., Nelson S. F. Detection of differentially expressed genes in primary tumor tissues using representational differences analysis coupled to microarray hybridization.
Nucleic Acids Res.
,
26
:
3059
-3065,  
1998
.
26
Vician L., Basconcillo R., Herschman H. R. Identification of genes preferentially induced by nerve growth factor versus epidermal growth factor in PC12 pheochromocytoma cells by means of representational difference analysis.
J. Neurosci. Res.
,
50
:
32
-43,  
1997
.
27
Feldman J. D., Vician L., Crispino M., Tocco G., Marcheselli V. L., Bazan N. G., Baudry M., Herschman H. R. KID-1, a protein kinase induced by depolarization in brain.
J. Biol. Chem.
,
273
:
16535
-16543,  
1998
.
28
Schowalter D. B., Sommer S. S. The generation of radiolabeled DNA and RNA probes with polymerase chain reaction.
Anal. Biochem.
,
177
:
90
-94,  
1989
.
29
Pruss R. M., Mirsky R., Raff M. C., Thorpe R., Dowding A. J., Anderton B. H. All classes of intermediate filaments share a common antigenic determinant defined by a monoclonal antibody.
Cell
,
27
:
419
-428,  
1981
.
30
Domanico S. Z., Pelletier A. J., Havran W. L., Quaranta V. Integrin α6Aβ1 induces CD81-dependent cell motility without engaging the extracellular matrix migration substrate.
Mol. Biol. Cell
,
8
:
2253
-2265,  
1997
.
31
Meier F., Nesbit M., Hsu M. Y., Martin B., Van Belle P., Elder D. E., Schaumburg-Lever G., Garbe C., Walz T. M., Donatien P., Crombleholme T. M., Herlyn M. Human melanoma progression in skin reconstructs: biological significance of bFGF.
Am. J. Pathol.
,
156
:
193
-200,  
2000
.
32
Howe C. C., Kath R., Mancianti M. L., Herlyn M., Mueller S., Cristofalo V. Expression and structure of human SPARC transcripts: SPARC mRNA is expressed by human cells involved in extracellular matrix production and some of these cells show an unusual expression pattern.
Exp. Cell Res.
,
188
:
185
-191,  
1990
.
33
Aznavoorian S., Stracke M. L., Parsons J., McClanahan J., Liotta L. A. Integrin αvβ3 mediates chemotactic and haptotactic motility in human melanoma cells through different signaling pathways.
J. Biol. Chem.
,
271
:
3247
-3254,  
1996
.
34
Khatib A. M., Nip J., Fallavollita L., Lehmann M., Jensen G., Brodt P. Regulation of urokinase plasminogen activator/plasmin-mediated invasion of melanoma cells by the integrin vitronectin receptor αVβ3.
Int. J. Cancer
,
91
:
300
-308,  
2001
.
35
Chang D. D., Park N-H., Denny C. T., Nelson S. F., Pe M. Characterization of transformation related genes in oral cancer cells.
Oncogene
,
16
:
1921
-1930,  
1998
.
36
Rempel S. A., Ge S., Gutierrez J. A. SPARC: a potential diagnostic marker of invasive meningiomas.
Clin. Cancer Res.
,
5
:
237
-241,  
1999
.
37
Thomas R., True L. D., Bassuk J. A., Lange P. H., Vessella R. L. Differential expression of osteonectin/SPARC during human prostate cancer progression.
Clin. Cancer Res.
,
6
:
1140
-1149,  
2000
.
38
Golembieski W. A., Ge S., Nelson K., Mikkelsen T., Rempel S. A. Increased SPARC expression promotes U87 glioblastoma invasion in vitro.
Int. J. Dev. Neurosci.
,
17
:
463
-472,  
1999
.
39
Jacob K., Webber M., Benayahu D., Kleinman H. K. Osteonectin promotes prostate cancer cell migration and invasion: a possible mechanism for metastasis to bone.
Cancer Res.
,
59
:
4453
-4457,  
1999
.
40
Ledda M. F., Adris S., Bravo A. I., Kairiyama C., Bover L., Chernajovsky Y., Mordoh J., Podhajcer O. L. Suppression of SPARC expression by antisense RNA abrogates the tumorigenicity of human melanoma cells.
Nat. Med.
,
3
:
171
-176,  
1997
.
41
Gilles C., Bassuk J. A., Pulyaeva H., Sage E. H., Foidart J. M., Thompson E. W. SPARC/osteonectin induces matrix metalloproteinase 2 activation in human breast cancer cell lines.
Cancer Res.
,
58
:
5529
-5536,  
1998
.
42
Yan Q., Sage E. H. SPARC, a matricellular glycoprotein with important biological functions.
J. Histochem. Cytochem.
,
47
:
1495
-1506,  
1999
.
43
Sanders L. C., Felding-Habermann B., Mueller B. M., Cheresh D. A. Role of αV integrins and vitronectin in human melanoma cell growth.
Cold Spring Harbor Symp. Quant. Biol.
,
57
:
233
-240,  
1992
.
44
Rosenblatt S., Bassuk J. A., Alpers C. E., Sage E. H., Timpl R., Preissner K. T. Differential modulation of cell adhesion by interaction between adhesive and counter-adhesive proteins: characterization of the binding of vitronectin to osteonectin (BM40, SPARC).
Biochem. J.
,
324
:
311
-319,  
1997
.
45
Vajkoczy P., Menger M. D., Goldbrunner R., Ge S., Fong T. A., Vollmar B., Schilling L., Ullrich A., Hirth K. P., Tonn J. C., Schmiedek P., Rempel S. A. Targeting angiogenesis inhibits tumor infiltration and expression of the pro-invasive protein SPARC.
Int. J. Cancer
,
87
:
261
-268,  
2000
.
46
Clark E. A., King W. G., Brugge J. S., Symons M., Hynes R. O. Integrin-mediated signals regulated by members of the Rho family of GTPases.
J. Cell Biol.
,
142
:
573
-586,  
1998
.
47
Keely P., Parise L., Juliano R. Integrins, and GTPases in tumour cell growth, motility and invasion.
Trends Cell Biol.
,
8
:
101
-106,  
1998
.
48
Clark E. A., Golub T. R., Lander E. S., Hynes R. O. Genomic analysis of metastasis reveals an essential role for RhoC.
Nature (Lond.)
,
406
:
532
-535,  
2000
.