Tenascin mediates human glioma cell migration and modulates cell migration on fibronectin

The extracelluar matrix is a complex of many components interacting together and with cell receptors to provide functions essential for cell adhesion, migration and differentiation (Hynes, 1992). Development and maintenance of tissue structure and function require the continuous presence of many of the extracellular components, including fibronectin, laminin, and various collagens. However, there are specific events essential for organ development, wound healing, and tumor growth that are characterized by alterations in matrix composition and function. These events include mesenchymal-epithelial interactions during tissue remodeling and differentiation in the gut, kidney, and mammary gland as well as neural-glial interactions involved in brain development (Grumet et al., 1985; Crossin et al., 1986; Aufderheide et al., 1987, 1988; Chuong et al., 1987; Husmann et al., 1992). Wound healing is also a process involving matrix reconstruction and multicellular migration and differentiation (Luomanen and Virtanen, 1993). Finally, malignant solid tumor growth is characterized by tumor invasion and metastasis and the induction of stromal and vascular proliferation and migration, processes resulting in tumor angiogenesis. In each of these settings tenascin is expressed in the extracellular matrix. The tightly regulated expression of tenascin in normal fetal development and wound healing is characterized by transient matrix expression, while in tumors its expression is associated with the progression to a malignant phenotype, and induction of angiogenesis and tumor invasiveness (Bourdon et al., 1983; Mackie et al., 1987; Higuchi et al., 1993; Sakai et al., 1993). The pattern of tenascin expression strongly suggests that tenascin plays a significant role in these matrix settings.

The functional role of tenascin, however, has not been clear, and contradictory effects on adhesion and migration have been reported (Mackie et al., 1988; Halfter et al., 1989; Lochter et al., 1991). More recently it has emerged that tenascin supports cell adhesion and does so through integrin binding at the RGD and other sites on the tenascin molecule. The integrins shown to bind to tenascin include α₂β₁ and α_vβ₃on endothelial cells, α₂β₁ and a possible second β₁integrin on tumor cells, and α_vβ₆ and α₉β₁on epithelial cells (Bourdon and Ruoslahti, 1989; Joshi et al., 1993; Prieto et al., 1993; Sriramarao et al., 1993; Yokosaki et al., 1994). Both α_vβ₃ and α_vβ₆interact with tenascin through the SRRGDMS site of tenascin (Prieto et al., 1993; Sriramarao et al., 1993). The limited cell spreading on tenascin and its reported inhibition of fibronectin-mediated cell adhesion (Chiquet-Ehrismann et al., 1988) has suggested that tenascin like SPARC and thrombospondin is an anti-adhesive matrix protein (Sage and Bornstein, 1991) that might play a role in cell migration.

In this paper we demonstrate that tenascin is an effective mediator of tumor cell migration. Tenascin is shown here to mediate cell migration through cell interactions with the α₂β₁integrin. In contrast, cell migration on fibronectin is mediated by α_vintegrins, particularly by α_vβ₅ and α_vβ₃. The finding that tenascin can alter cell migration on fibronectin resulting in increased migration and reduced cell spreading on a substratum is significant. These effects are integrin dependent and can be selectively blocked by blocking α₂β₁ function.

U251.3 cell line is a subclone obtained from the glioma cell line U251MG by limiting dilution. The cell lines MG-63 osteoblastoma and MRC-5 diploid fibroblasts were obtained from the ATCC (Rockville, MD). Cells were grown at 37°C in a humified CO₂ incubator in Dulbecco’s MEM (DMEM, Sigma, St Louis, MO) supplemented with 10% fetal calf serum (FCS, Tissue Culture Biologicals, Tulare, CA), 2 mM L-glutamine, and antibiotics.

Human tenascin was purified from culture supernatant conditioned by U251.3 cells as previously described (Sriramarao et al., 1993). The tenascin was greater than 98% pure and free of fibronectin as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), ELISA and western blotting. Fibronectin was purified from human plasma as described (Ruoslahti et al., 1982).

Tenascin cDNA have been subcloned in the yeast shuttle/expression vector pPIC9 and transfected into P. pastoris yeast (Invitrogen, San Diego, CA) by RT-PCR of U251.3 mRNA. Expression in P. pastoris provides glycosylation similar to that in mammalian cells and disulfide bond formation insuring proper protein folding. Three subcloned regions of tenascin used in these studies are designated rTNfnIII1-5 (fibronectin type III repeats 1-5), rTNfnIII4-9, and rTNfnIII8-13 and were obtained by RT-PCR using the primer pairs: GAAGCTTACGTATGCATCTGCAACGAGGG, AAGACCTAG-GTTTTCCAGCTCAGGGGCTT; AAGCTTACGTAAAAGAGAC-CTTCACAACAGG, AAGACCTAGGCTATTTGGCTGTGGAGG-CCTCA; GAAGCTTACGTAGAGGAGGTTCCAGATATGGG, AAGACCTAGGCTAGACTTCCTTTGGGGAGCCC.

PCR products were cloned into the TA cloning vector PCRII (Clonetech, Palo Alto, CA). Tenascin subcloned cDNA were restricted with SnaBI and AvrII and directionally cloned into pPIC9. The plasmid was propagated in TOP10F bacteria and plasmid DNA was electroporated into P. pastoris strain GS115. Transfected GS115 clones were selected on histidine minus plates. Expression was induced from the AOX-1 promoter by placing transformed yeast into methanol medium, according to the manufacturer’s instructions (Invitrogen). Recombinant proteins were expressed at 50-70 mg/liter culture. Following induction in methanol minimum medium, the culture supernatants were collected, assayed for recombinant protein by SDS-PAGE, ELISA and western blotting, and recombinant proteins were isolated by either immunoaffinity or Mono-Q chromatography.

Wells of high-binding 96-well plates (Costar) were coated with 100 µl of tenascin (2.5 µg/ml), fibronectin (1.5 µg/ml) or BSA in PBS overnight at 4°C. Unbound sites were blocked with 1% BSA (Sigma) in DMEM for 1 hour at 37°C. A total of 50,000 cells were added per well in 100200 µl DMEM containing 1% BSA. For blocking experiments, ascites were used at a final dilution of 1:500 (anti-β₁ and anti-α_vβ₅) or 1:100 (all other ascites), and purified mAbs were used at a final concentration of 50 µg/ml. Cells were allowed to attach for 45 minutes at 37°C in a CO₂ incubator. Thereafter, plates were gently washed three times with Dulbecco’s PBS (DPBS, BioWittaker) to remove unattached cells. The adherent cells were fixed and stained with 0.2% Crystal Violet (Sigma) in 10% ethanol. Thereafter plates were washed three times with PBS, and the dye was extracted with 0.05 M sodium phosphate (pH 4.5) in 50% ethanol (Kilian et al., 1991). The absorbance was measured at 540 nm using a microplate reader (Bio-Tek Instruments, Winooski, VT).

Boyden chamber type Transwells (Costar, Cambridge, MA) were used to examine the ability of cells to invade the 8 µm pore size membrane and migrate onto matrix-coated surfaces. The undersurfaces of the 6.5 mm Transwell membranes were coated with tenascin, fibronectin or BSA at 20 µg/ml in PBS overnight at 4°C. After blocking with BSA, 3×10⁵ cells were plated in serum-free AIM-V medium (Gibco, Grand Island, NY) into the inner chamber and allowed to migrate onto the coated undersurface at 37°C in a CO₂ incubator. One, 2 or 4 hours later, cells from the inner surface of the insert were wiped from the surface and the inserts were fixed and stained with Crystal Violet solution. Following PBS washing of the membranes, the dye was extracted as described above for the cell attachment assay.

A two-dimensional spheroid outgrowth assay was used to assess the migration of tumor cells on different adhesion proteins. Spheroids were prepared from single cell suspension essentially as described (Sano et al., 1983). Briefly, 4×10⁶ to 10×10⁶ cells were placed in 10 ml DMEM/FCS in 50-ml conical siliconized glass flasks. Flasks were rocked on a gyratory shaker in a CO₂ incubator at 37°C for 2-4 days for formation of spheroids. Spheroids with similar diameters were selected and placed into 96-well Costar plates. Plates were previously coated with adhesion proteins overnight and blocked with BSA (10 mg/ml). To determine the coating concentration optimal for cell migration, extracellular matrix proteins were titrated and thereafter the protein concentration supporting maximal cellular outgrowth from spheroids was used in all experiments (10 µg/ml). Spheroids were washed and added to wells in 50 µl serum-free AIM-V medium and then tested antibodies or recombinant proteins were added in a volume of 100 µl/well. In the cell migration assays all blocking antibody ascites were used at a final dilution of 1:100 and purified mAbs or recombinant proteins at a final concentration of 100 µg/ml. Plates were then incubated in the humidified CO₂ incubator at 37°C. The diameter of the spheroids at the time of plating and diameter of the area covered with cells migrated from the spheroids following 16-48 hours migration was measured under an inverted microscope using a calibrated reticle. The radial distance of migration was calculated after subtraction of the mean initial spheroid diameter at time zero.

The anti-human tenascin monoclonal antibody (mAb) 9F8 was obtained after immunization of RBF mouse with purified tenascin. This mAb specifically binds to tenascin and to RGD-containing rTNfnIII1-5 (but not to rTNfnIII4-9 or rTNfnIII8-13) in ELISA. In cell adhesion assays, 9F8 mAb partially blocks cell attachment of the MG-63 cell line to tenascin and blocks the attachment of MG-63 and D54 cell lines to rTNfnIII1-5 up to 98% and of human umbilical vein endothelial cells (HUVEC) up to 74% (data not shown). The blocking by mAb 9F8 on tenascin is partial, since two independent receptor sites on the molecule mediate cell adhesion (Sriramarao and Bourdon, 1995). The mAb 9F8 does not cross-react with other matrix proteins in solid phase ELISA. Blocking integrin-specific mAbs were obtained from GibcoBRL (anti-β₁, P4C10, and anti-α_vβ₅, P1F6) and Chemicon Int. (Temecula, CA) (anti-α₂, P1E6; anti-α₃, P1B5; anti-α₅, P1D6; anti-α_vβ₃, LM609; anti-α_vβ₅, P1F6). Anti-α_v blocking mAbs, L230 (ATCC) and L1A3 (this murine mAb was obtained at LJIEM and will be described elsewhere), were purified from serum-free supernatants. Control mAb produced by myeloma 45.6 (ATCC) was purified from ascitic fluid.

The experiments were reproduced at least twice (some as many as 20 times). For quantitative analysis the arithmetical mean value and standard error (s.e.m.) were calculated. The significance of the data was evaluated by paired and independent t-tests using the SigmaStat software program (Jandel Corporation, Corte Madera, CA). Data were interpolated with the Sigmaplot software program (Jandel Corporation).

We first examined cell migration on tenascin and fibronectin in the Transwell migration/invasion assay. In Transwell assays U251.3 cells were placed in serum-free medium onto the 8 µm pore size membranes, which were coated on the opposite membrane surface with tenascin, fibronectin or BSA. After 1, 2 and 4 hours of incubation the relative numbers of cells trans-migrating to the undersurface of the membrane were determined. The results showed that U251.3 cells were capable of rapid motility onto both tenascin and fibronectin but not BSA-coated surfaces (Fig. 1). For comparison, MRC-5 normal fibroblasts were also examined in the Transwell assay. Migration of MRC-5 cells onto tenascin was at least 5-fold less than for U251.3 cells. While both fibronectin and tenascin supported cell migration, there were significant differences in glioma cell migration on the two matrices.

The U251.3 cell migration in tenascin-coated Transwells resulted in 52.2±9.6% more cells transmigrating onto tenascin than on fibronectin-coated membranes (Fig. 2). Transmigration for both tenascin and fibronectin required a haptotactic gradient as shown by coating both sides of the Transwell membrane. Fibronectin migration was reduced 90% when both membrane surfaces were coated while tenascin migration was reduced 80% (data not shown). The glioma cells also differed in their morphology on tenascin and fibronectin. Cells that had transmigrated onto fibronectin were highly spread forming a confluent monolayer on the membrane. In contrast, cells that transmigrated onto tenascin were less spread and did not form a monolayer but had a more motile appearance as indicated by long cell processes and a tendency for cells to overlay one another (Fig. 3).

These results show that tenascin promotes the rapid migration or invasion of U251.3 cells across a membrane. The relative rate of migration on tenascin is significantly greater than on fibronectin and correlates with differences in the morphology of transmigrated cells. The more rapid migration of U251.3 cells on tenascin as compared to fibronectin correlates with reduced cell spreading and a motile phenotype.

While tumor cell invasion as modeled in Transwell assays is an important parameter of tumor malignancy, tumor cell migration, and in particular glioma cell migration, often takes place over extended distances. Therefore U251.3 cell migration was next examined in a two-dimensional spheroid outgrowth assay. The tumor spheroid migration assay allowed the initiation of migration from a defined number of cells from a single focus and the subsequent observation of cellular morphology and cell migration in two dimensions over time. Spheroids of tumor cells containing 1,000-3,000 viable cells depending on spheroid size were allowed to form over 2-4 days in suspension culture and then were plated into wells coated with tenascin, fibronectin, or BSA. The tumor spheroids at the time of plating had formed a tightly bound microtissue with a cross-sectional diameter of 150-350 µm. Spheroids were size selected, ±10 µm for each experiment, and their initial diameters were recordered. Following incubation of spheroids in coated plates, the morphology of cells migrating out onto the matrix was observed and the average radial distance of cell outgrowth was determined.

The results showed that tenascin directly supports cell migration, confirming the results from the Transwell assays (Table 1). There was no cell migration on BSA-coated wells in serum-free medium. As shown, glioma cell migration on tenascin was significantly greater than on fibronectin as measured by radial distance migrated and area covered (Table 1). Over a 48 hour period the radial distance migrated on tenascin averaged 1.54 times greater than that measured on fibronectin and as a consequence the outgrowth area on tenascin was two times greater than that measured on fibronectin (Table 1). Migration on tenascin occurred at coating concentrations as low as 0.5 µg/ml but were maximal at coating concentrations of 2.5 to 20 µg/ml (data not shown). Average radial migration distance on tenascin was typically 425 µm or approximately 20 cell diameters in a 16-24 hour experiment. This would mean that cells migrating from the spheroid traveled radially about 18-26 µm each hour. However, this is likely a significant underestimate of motility rates, since cellular movement does not appear to be radial alone and outgrowth on substrata appears to lag in the first 23 hours following initiation of the assay.

The morphology of cells migrating on tenascin is also different than that seen on fibronectin (Fig. 4). These differences in the cell morphology are identical to those seen in Transwell assays although much more readily observed. Cells migrating out on tenascin have highly extended cell bodies with ruffled edges and trailing processes, and limited intercellular contacts. In contrast, cell migration on fibronectin is characterized by highly spread cells in close contact with the adjacent cells. The leading edges of the front of cells migrating on fibronectin is very distinct as the leading cells maintain cell-substratum and cell-cell contacts, while the leading edge of migration on tenascin is less distinct, since cells have few intercellular contacts and appear oriented in a variety of directions (Fig. 4). The visual impression, confirmed by migration measurements, is that U251.3 cells are highly motile on tenascin.

Overall the results show that tenascin promotes rapid cell migration and increased motility compared to fibronectin. We next examined whether the mechanism of tenascin-mediated migration was integrin dependent as has been shown for fibronectin.

Tenascinand fibronectin-mediated cell adhesion and migration were examined in the presence of specific adhesionblocking anti-integrin antibodies.

Cell adhesion assays were carried out on tenascin and fibronectin with known anti-integrin blocking mAbs to determine which integrins mediate U251.3 cell adhesion. As expected, U251.3 cells adhered to fibronectin via α₅β₁ and α_vβ₃(Fig. 5). Cell adhesion to tenascin was mediated by α₂β₁ and α_vβ₃. Blocking was complete when both β₁ and α_v or α₂ and α_v were blocked in combination. When α₃, α₄, α₅ and α₆ were blocked, only anti-α₃ inhibited adhesion. The combination of anti-α₃ and anti-α_v had an additive inhibition, indicating that α₃ integrins also mediated adhesion. The finding that α₂β₁ and α_vβ₃ are tenascin receptors is supported by previous results with endothelial cells (Sriramarao et al., 1993; Sriramarao and Bourdon, 1995). Other tumors such as MG-63 adhere to tenascin via β₁ integrins but not through either α₂β₁ or α₃β₁(Sriramarao and Bourdon, 1993).

Since integrins mediate the attachment of U251.3 cells to tenascin we examined the role of these integrins in mediating cell migration (Fig. 6). In these experiments the migration of U251.3 cells on tenascin was inhibited 87.3±8.8% by β₁ blocking monoclonal antibodies and up to 34.1±4.9% by α₂ blocking antibodies at 24 hours and remained blocked at the 48 hour time point following spheroid plating. The inhibition was evident in terms of both radial distance of migration (Fig. 6) and cell outgrowth morphology. No significant inhibition of migration on tenascincoated wells could be observed with blocking α₃ and α₅ antibodies. The partial blocking of cell migration on tenascin by anti-α₂ antibody may indicate incomplete receptor blocking as is also observed on collagen type I. Alternatively, another unidentified β₁ integrin could also be involved. A critical role for α₂β₁ integrin is indicated by results shown below in which α₂ antibody completely blocks the stimulatory effect of soluble tenascin on fibronectin-mediated cell migration.

The effects of blocking integrins α_vβ₃ and α_vβ₅ on cell migration on tenascin were also examined (Fig. 6). Blocking α_v integrins with either L230 or L1A3 mAb did not inhibit cell migration on tenascin. Cell migration was also not inhibited by blocking α_vβ₃ and α_vβ₅ with LM609 and P1F6 mAb, respectively. Tenascin-mediated cell migration therefore does not appear to be critically mediated by α_v integrins. As shown below, α_v integrins on U251.3 cells are active in mediating fibronectin cell migration and are effectively blocked by α_v blocking antibodies.

Fibronectin-mediated migration, in contrast to cell migration on tenascin, is α_vβ₅/α_vβ₃ dependent (Fig. 6). As shown, α_v blocking strongly inhibits fibronectin-mediated migration up to 92% by L1A3 mAb and blocking α_vβ₅ and α_vβ₃ partially inhibits migration by 29.5±5.5% and 17.6±6.2%, respectively, in 24 hour migration assays (Fig. 6). There was no blocking of migration in the presence of other integrin blocking antibodies, including those to β₁ or α₅. In adhesion assays, blocking with α₅ or α_vβ₃ antibodies significantly inhibits cell attachment to fibronectin (Fig. 5). In this context α_vβ₅ did not mediate cell adhesion, indicating a clear distinction between integrin function in adhesion of cells reattaching to fibronectin and the motility of cells already attached on fibronectin.

We examined whether the RGD-adhesion site on tenascin was involved in U251.3 cell migration using recombinant tenascin fragments and a monoclonal antibody that blocks the RGD site of tenascin.

Several cell lines including MG-63 as well as normal fibro-blasts and endothelial cells adhere to tenascin via the RGD site and a second non-RGD site (Sriramarao et al., 1993; Prieto et al., 1993). These cells also adhere to the RGD site within the fnIII3 repeat of rTNfnIII1-5. Recombinant tenascin proteins covering 13 fibronectin type III repeats including fnIII3 within rTNfnIII1-5 did not support cell attachment or migration of U251.3 cells even at 50-100 µg/ml coating concentrations (Table 2). In addition, these soluble or substratum attached recombinant tenascin fragments did not inhibit or stimulate migration on tenascin or fibronectin (Table 2). The mAb 9F8, which blocks adhesion of MG-63 cells and HUVEC to the SRRGDMS site of tenascin and rTNfnIII1-5, had no effect on tenascin-mediated cell migration of U251.3 cells, indicating that the SRRGDMS site on tenascin does not contribute to supporting cell migration (Table 2). In addition, as shown above, the RGD-dependent integrins α₅β₁, α_vβ₃, and α_vβ₅ do not mediate U251.3 migration on tenascin. These data indicate that cell migration of U251.3 on tenascin is independent of the RGD site of human tenascin.

The ability of tenascin to inhibit cell adhesion to fibronectin (Chiquet-Ehrismann et al., 1988) and its co-expression with fibronectin in tissues (Bourdon et al., 1983) suggested that tenascin could modulate migration on fibronectin substrata. Migration on fibronectin was therefore assayed in the presence of soluble tenascin.

The morphology of cells migrating on fibronectin in the presence of tenascin was significantly altered. Cells appeared distinctly less spread than in control fibronectin-coated cultures and fewer intercellular contacts were present within 24 hours of treatment. Following 48 hours in culture with tenascin, the migrating cells had acquired the appearance of cells migrating on tenascin alone (Fig. 7). Titration of the tenascin indicated that concentrations as low as 11 µg/ml could significantly enhance cell migration on fibronectin (Fig. 8A). After 24 hours, cell migration on fibronectin in the presence of soluble tenascin (100 µg/ml) was enhanced on average 25.8±2.9% (P<0.002) over that seen on fibronectin alone. The migration rate of cells continued to be greater in tenascin-treated cultures for at least 48 hours, reaching on average 29.7±.5.6% (P<0.005) greater migration than in untreated fibronectin-coated wells (Fig. 8B). The effect of tenascin on cells migrating on fibronectin was to increase cell migration to a level intermediate between the rate of migration on fibronectin and the higher rate of migration on tenascin. Another important aspect of the interaction of tenascin and fibronectin is the dominant effect of tenascin in terms of alterations in cell morphology and rate of migration. This is further supported by the findings that while addition of tenascin significantly increases cell migration on fibronectin, adding soluble fibronectin up to 100 µg/ml to either tenascinor fibronectin-coated wells does not affect cell migration positively or negatively (data not shown).

Since cell migration on tenascin is mediated by α₂β₁ integrins, the effects of tenascin on fibronectin cell migration were examined in the presence of β₁ and α₂ blocking antibodies. The results show that after 24 hours the effect of tenascin on fibronectin-mediated cell migration was completely blocked if either β₁ or α₂ blocking antibodies were added (Fig. 9) and continued to be blocked for at least 48 hours. The overall effect of α₂β₁ blocking was to block the migration effects of tenascin while leaving fibronectin adhesion and migration unaltered. When in both control and tenascin-treated cultures α_v blocking occurred, cell migration on fibronectin was substantially but not completely blocked, as the cells appeared to adapt to migration that was α_v independent as shown in Fig. 9. The results demonstrate that the effects of tenascin on cell migration on fibronectin are receptor-mediated and distinct from any interaction of the cells or tenascin with fibronectin, since the cell migration activities of fibronectin and tenascin are independently mediated by different receptors.

The results presented here demonstrate a direct functional role for tenascin in supporting and promoting tumor cell migration. These studies show that tenascin directly supports U251.3 cell migration but also stimulates more rapid cell migration on the major adhesion protein fibronectin. Both of these activities require the interaction of tenascin with the α₂β₁ integrins present on the surface of tumor cells.

The findings of specific integrin-mediated cell migration on tenascin has only been inferred previously by the very selective and regulated expression of tenascin during development, wound healing, and malignant tumor growth (Erickson and Bourdon, 1989). Tenascin is expressed developmentally during organogenesis of kidney, gut, breast, teeth, and brain (Grumet et al., 1985; Crossin et al., 1986; Chuong et al., 1987; Aufderheide et al., 1987, 1988; Thesleff et al., 1987). In each of these tissues, tenascin appears when multiple cell types must interact, reorganize, and differentiate. This pattern of expression is typified by the induction of tenascin in condensing mesenchyme resulting from mesenchymal-epithelial interactions during early development of breast and kidney as examples (Chiquet-Ehrisman et al., 1986; Aufderheide et al., 1987). Others have shown that the migration of neurons during brain development (Chuong et al., 1987; Husmann et al., 1992), as well as neural field patterning (Steindler et al., 1989) and neurite extension, are associated with or affected by tenascin expression. From these results and those demonstrating the association of tenascin expression with tumor vascular structures and the malignant tumor phenotype (Bourdon et al., 1983; Guarino et al., 1993; Sakai et al., 1993) it is clear that tenascin is associated with selective events in which cell signaling results in cell migration or tissue remodeling.

In two separate tumor cell migration assays we found that tenascin substrata promoted the rapid migration of glioma cells. Indeed, the rate of migration was greater on tenascin than on fibronectin. In Transwell assays glioma cells rapidly migrated through the membrane pores and accumulated on the tenascin-coated surface. The Transwell assays clearly demonstrated the high degree of motility and invasive potential of U251.3 glioma cells typical of malignant glioma in vivo. The 50% increase in numbers of cells transmigrating onto tenascin as compared to fibronectin provides direct evidence for the role of tenascin in promoting tumor cell migration, and suggests that the correlation between tenascin expression and the malignant tumor phenotype also correlates with increased cell migration. These observations were confirmed in a second model of cell migration.

Tumor cell migration in two dimensions was examined in the spheroid outgrowth assay. This assay allows for measurement of rates of migration in terms of distances traveled and serves as a model of microtumor cell migration. In these experiments observations were made of the radial distance that cells migrated, areas covered, and cell morphology. From the extensive data collected on approximately 400 spheroids for each matrix substratum it was confirmed that tenascin promoted an approximately 50% greater migration rate than fibronectin. In terms of the area covered, another measure of tumor migration, the area covered by migrating cell on tenascin was twice that of fibronectin. The estimate of cell migration rate is ∼25 µm/h; however, given the apparent random movement of tumor cells this represents only the average radial distance migrated. The migration rate, while likely a significant underestimate of absolute cell motility rates, does reflect the most significant parameter in regard to tumor migration and invasion: that is, the rate at which cells move out from the primary or metastatic tumor foci into surrounding tissues.

Rapid migration on tenascin was accomplished by striking changes in cell morphology. These changes in cell morphology as compared to fibronectin were indicative of highly motile cells. Tumor cells migrating on fibronectin remained highly spread throughout the 48 hours of observation, with cells maintaining close cell-to-cell contacts even at the migrating margin. In contrast, cells migrating on tenascin were dispersed with much less cell spreading and the appearance of ruffled edges and trailing cell extensions. Far from forming a monolayer, cells migrating on tenascin move freely over each other.

Having established that tenascin directly supports cell migration, we next examined whether integrins mediated this activity. The results shown here demonstrate that β₁ integrins and specifically α₂β₁ are critical for cell migration on tenascin. Blocking β₁ integrins results in the total blocking of the U251.3 migration on tenascin. Similar blocking studies directed against α integrin subunits show that α₂β₁ is a mediator of cell migration on tenascin. The partial inhibition with α₂ blocking antibody may indicate that additional β₁ integrins play a role in cell migration, although it should be considered that α₂ antibodies are less effective receptor blockers, even on collagen, than β₁ blocking antibodies. In addition, in tenascin-fibronectin cell migration to be discussed below, blocking α₂β₁ abrogates tenascin-mediated effects on fibronectin cell migration. The cell interaction of tenascin with α₂β₁ integrin is selective in that migration on fibronectin alone is mediated by α_vβ₅/α_vβ₃ integrins. This result suggests that independent cell signaling pathways exist for tenascin and fibronectin cell migration. Both α₂ and α_v integrins have been shown to participate in cell migration. Integrin α_vβ₅ has been shown to selectively mediate cell migration on vitronectin (Klemke et al., 1994), while α₂β₁ integrin mediates cell migration on collagen (Yamada et al., 1990; Grzesiak et al., 1992) and collagen matrix contraction (Klein et al., 1991; Schiro et al., 1991).

Cell adhesion to tenascin for several primary cells and cell lines including HUVEC and MG-63 is mediated by at least two sites, the SRRGDMS site, which is known to bind both α_vβ₃ and α_vβ₆(Sriramarao et al., 1993; Prieto et al., 1993), and a second non-RGD site, binding α₂β₁(Sriramarao et al., 1993). However, the SRRGDMS site does not appear to play a critical role in U251.3 migration. The evidence from cell migration and adhesion assays with recombinant tenascin fragments indicates that the first 13 fibronectin type III repeats, including the rTNfnIII1-5 containing the RGD site, do not support U251.3 cell attachment or cell migration. The data obtained with HUVEC, MG-63 and DU54 cell lines demonstrate that at least the rTNfnIII1-5 fragment, which includes the SRRGDMS site, is functionally active in adhesion assays. In addition, a mAb to tenascin, 9F8, which blocks MG-63 and HUVEC cell adhesion to the SRRGDMS site of tenascin and rTNfnIII1-5, does not inhibit tenascin-mediated migration of U251.3 cells. For both tenascin and fibronectin it appears that integrins critical to cell migration represent a subset of integrins capable of mediating adhesion. This likely reflects the differences in cell function that relate to adhesion alone as compared to cell motility, which involves both adhesive and anti-adhesive events.

The most intriguing effect of tenascin on the migration of U251.3 cells is its modulating effect on fibronectin-mediated cell migration. When tenascin was added to the media of tumor cells migrating on fibronectin substrata, migration was enhanced and cell contacts reduced. The enhancement of cell migration was directly related to the concentration of tenascin added over the concentration range of 3 to 100 µg/ml. These effects were not the result of tenascin blocking fibronectin sites but were specifically α₂β₁ receptor mediated, since blocking antibodies to α₂ and β₁ integrins blocked the migrationenhancing effects of tenascin and left cell migration on fibronectin unaffected. Blocking α₂β₁ over a wide range of tenascin concentrations demonstrates that cell receptor occupancy and not ligand concentration is critical to mediating the effects of tenascin on fibronectin cell migration. The results demonstrate that α₂β₁ binding to tenascin is capable of signaling enhanced migration on what are otherwise α_v-mediated cell migration substrata. In these experiments we have not determined whether the effects of tenascin on fibronectin cell migration were due to tenascin adsorption to the coated wells. There is some indication that this can occur, in which case cells presented with two cell adhesion and migration substrata have selectively responded to α₂β₁ signaling to enhance cell migration. These effects could be mediated by a dominant specific receptor signaling mechanism that inhibits fibronectin/integrin-mediated strong adhesion and/or stimulates cell motility. The use of separate receptors to mediate migration and the quantitative differences in migration on tenascin and fibronectin may indicate distinct signal pathways for migration on each substratum. The cell signaling mechanisms by which tenascin mediates cell migration and elicits a dominant effect on fibronectin-supported cell migration at least requires the critical interaction with α₂β₁ integrin.

The finding that a given cell line can distinguish between tenascin and fibronectin and respond differently via different integrins suggests that cells have the potential to recognize alterations in the extracellular matrix and as a consequence to alter their adhesive and motility characteristics. This is a novel aspect of α₂β₁-tenascin regulation in keeping with the presumed role of tenascin in modulating cell adhesion and the growing evidence of integrin cross-talk and signal regulation by ligand and intracellular regulatory interactions (Hynes,1992). In our model of tenascin function, the up-regulation of tenascin in the tissue matrix results in a new receptormediated matrix signal that down-regulates strong adhesion and provides an enhanced matrix for migration. In this model tenascin mediates integrin cell adhesion but not strong attachment, allowing cells to attach and detach from the matrix rapidly. The effect of tenascin in enchancing cell migration on fibronectin also suggests that tenascin interactions with integrins or other receptors may provide a second signal modulating cell motility and cell adhesion on tissue matrices.

The results described here suggest that tenascin and α₂β₁ could play a significant role in tumor growth and metastasis in vivo. The findings that tenascin supports cell migration and exerts a dominant effect on cell migration in the presence of fibronectin suggest that the up-regulation of tenascin in malignant tumors may contribute to tumor cell invasiveness (Bourdon et al., 1983; Shriramarao and Bourdon, 1995; Zagzag et al., 1995) by promoting tumor cell migration. In addition, as endothelial cells bind tenascin through α₂β₁ and α_vβ₃(Sriramarao et al., 1993), tenascin may be an important matrix protein in the tumor-endothelial interactions that result in angiogenesis (Shriramarao and Bourdon, 1995). The results suggest that down-regulation of tenascin or blocking functional sites on the tenascin molecule could be a new approach to control of tumor and endothelial cell migration.

We thank Boris I. Ratnikov and Clare Feigl for their excellent technical assistance, and Dr Alexey V. Cherepakhin for development of the statistical analysis program. We also thank Dr P. Sriramarao for comments on the manuscript. This work was supported by a grant from the NCI, CA 52879 (to Mario A. Bourdon).