THE JOURNAL OF BIOKXXCAJ CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 269, No. 19, Issue of May 13, pp. 14284-14289, 1994 Printed in U.S.A.

Purification and Biological Characterization of Epitaxin, a Fibroblast-derived Motility Factor for Epithelial Cells*

(Received for publication, December 27, 1993, and in revised form, March 14, 1994)

Motoyuki Shimonaka and Yu YamaguchiS From the Cancer Research Center, La Jolla Cancer Research Foundation, La Jolla, California 92037

We have identified a motility factor in conditioned me- dium of human fibroblasts that stimulates the migration of HepG2 cells in the Boyden chamber assay. This factor, termed epitaxin, was purified to homogeneity by ammo- nium sulfate precipitation, hydrophobic interaction chromatography on phenyl-Sepharose, and a series of reverse phase high performance liquid chromatogra- phy. Under a nonreducing condition, purified epitaxin migrated as a 36-kDa band, and the biological activity was recovered from the area in the gel coinciding with this band. Purified epitaxin stimulates the motility of HepG2 cells in the concentrations above 1 ng/ml with half-maximal activity at 4.2 ng/ml, and its mode of action is mainly chemotactic. Epitaxin slightly stimulates DNA synthesis of HepG2 cells, while scatter factorhepatocyte growth factor which also stimulates the motility of HepGZ cells showed a growth-inhibitory effect. Epitaxin increases the motility of a wide variety of epithelially derived tumor cell lines, but none of the tested fibro- blast lines responded to epitaxin. These results define epitaxin as a novel fibroblast-derived factor that affects the migration, and possibly the invasion, of epithelially derived tumor cells.

Cell migration is regulated by various extracellular signals at different molecular levels. Cell-substrate adhesion provides an anchorage from which moving cells can derive traction. Cell- cell adhesion and intercellular junctions act to restrict the movement of the cells by binding them to neighboring cells. A third type of regulation of cell migration is one by cytokines (see Ref. 1 for review). Various cytokines have been reported to affect not only cell proliferation but also cell motility. They include acidic and basic fibroblast growth factors (FGFs)’ (2,3), platelet-derived growth factor (PDGF) (4), insulin-like growth factor (1GF)-I and -11 (5,6), transforming growth factor (TGFI-p (71, tumor necrosis factor (TNF)-m (81, and interleukin (1L)-6 (9). These cytokines were originally isolated as mitogenic fac- tors, and their effects on cell motility were observed as a sec- ondarv Dhenomenon (1).

~~~~ ~

* This work was supported by National Institutes of Health Research Grant CA-57366. The costs of publication of this article were defrayed

hereby marked “aduertisement” in accordance with 18 U.S.C. Section in part by the payment of page charges. This article must therefore be

1734 solely to indicate this fact.

search Foundation, 10901 North Torrey Pines Rd., La Jolla, CA 92037. j: To whom correspondence should be addressed: La Jolla Cancer Re-

The abbreviations used in this paper are: FGF, fibroblast growth factor; AMF, autocrine motility factor; BSA, bovine serum albumin; FCS, fetal calf serum; IGF-I, insulin-like growth factor-I; IL-6, inter- leukin-6; MEM, minimum essential medium; MSF, migration stimulat- ing factor; PAGE, polyacrylamide gel electrophoresis; PDGF, platelet- derived growth factor; PVDF, poly(viny1idene fluoride); RP, reverse phase; HPLC, high performance liquid chromatography; SDMF, smooth muscle cell-derived migration factor; SF, scatter factor; HGF, hepato- cyte growth factor; TGF, transforming growth factor, TNF, tumor ne- crosis factor; MDCK, Madin-Darby canine kidney cells.

a l . : 619-455-6480; Fax: 619-455-0181.

Recent studies have revealed the existence of a new class of cytokines which predominantly stimulate the motility of vari- ous types of cells (1). These “motility factors” have been isolated based on assay systems that measure cell motility. The first evidence for motility factors was found by Yoshida et al. (lo), who demonstrated that certain tumor cell lines secrete a factor that enhances chemotaxis of the producer cells themselves. Thus far several motility factors have been purified. Autocrine motility factor (AMF) is a 55-kDa protein derived from mela- noma and mammary carcinoma cell lines that stimulates che- motaxis of the producer cells (11-13). Migration-stimulating factor (MSF), a 77-kDa protein, produced by human fetal and breast cancer patient fibroblasts, stimulates the migration of the producer cells into collagen matrix (14). Scatter factor (SF), which has recently been identified as same molecular species as hepatocyte growth factor (HGF), causes colonies of epithelial cells to scatter into single cells by enhancing cell motility (15- 18). Autotaxin is a 125-kDa protein which, like AMF, stimulates the motility of melanoma cells in an autocrine fashion (19). Another 58-kDa autocrine motility factor for vascular smooth muscle cells (SDMF) has recently been described (20).

Although physiological roles of various motility factors are largely unknown, their remarkable effects on cell motility sug- gested a role for them in tumor invasion and metastasis (see Refs. 1,21, and 22 for review). However, in spite of the growing number of motility factors described, most of them, with an exception of SF/HGF, appear to act mainly on nonepithelial cell types. As nearly 90% of human malignant tumors arise from epithelial tissues, studies on motility factors for epithelial cells are important. We have found an activity in conditioned media of human fibroblasts that stimulates the migration of HepG2 cells. This factor, named epitaxin, appears to act predominantly on epithelially derived tumor cells and functions in a paracrine fashion, as it is synthesized by mesenchymal cells but acts on epithelial cells. In the present report, we describe the purifica- tion and biological characterization of this novel motility factor.

EXPERIMENTAL PROCEDURES Materials-Recombinant human acidic FGF, basic FGF, PDGF (NA

form), IGF-I, IL-6, and y-interferon were purchased from Boehringer Mannheim. Human TGF-p1, porcine TGF-p2, recombinant human TNF-a, and goat anti-human HGF neutralizing antibodies were ob- tained from R & D Systems (Minneapolis, MN). Mouse monoclonal anti- human IL-6 neutralizing antibodies were purchased from Biosource International (Camarillo, CA), and rabbit anti-human serum albumin and mouse monoclonal anti-human serum albumin were from Sigma. Recombinant human HGF was a gift from Dr. Toshikazu Nakamura (Osaka University, Osaka, Japan), and human autotaxin was kindly provided by Dr. Mary Stracke (National Cancer Institute, Bethesda, MD).

Cell Culture-Chemically transformed human fibroblast lines HuT- 12, HUT-14, and HuT-l4T, all derived from human normal fibroblasts (23), were obtained from Dr. John Leavitt (ADEZA Biomedical, Sunny- vale, CA) and maintained in a-modified MEM ((Y-MEM) (GIBCO) supplemented with 9% FCS, 2 mM glutamine, and antibiotics (100 unitdml penicillin and 100 pg/ml streptomycin) (“maintenance me-

14284

Epitaxin Is Motility Factor for Epithelial Cells 14285

dium"). For collection of conditioned media, confluent monolayers of HUT-12 cells were washed extensively with a-MEM and cultured in a-MEM supplemented with 2 IMI glutamine and antibiotics ("serum- free medium") for 2 days. The collected serum-free conditioned media were filtered to remove cells, supplemented with 5 mM EDTA and 0.2 mM phenylmethylsulfonyl fluoride, and stored at -20" C until use. HepG2 cells were cultured in the maintenance medium. Rat hepatoma cell lines, H411, Fao, and C2, kindly provided by Dr. Mary Weiss (Insti- tut Pasteur, Paris, France), were cultured in a 1:l mixture of NCTC 135 (Sigma) and Ham F-12 (GIBCO) media supplemented with 5% FCS, 2 mM glutamine, and antibiotics. Canine kidney epithelial cell line MDCK, human lung adenocarcinoma cell line A549, human breast duc- tal carcinoma cell line ZR-75-1, and rat bladder carcinoma cell line NBT-11, are all obtained from the American Type Culture Collection and cultured in the maintenance medium, except for ZR-75-1 cells which are cultured in RPMI 1640 supplemented with 9% FCS, 2 m glutamine, and antibiotics.

Migration Assay-The modified Boyden chamber assay in a 48-well chemotaxis apparatus (Neuroprobe, Cabin John, MD) was performed according to Stracke et al. (5). Polycarbonate filters with 8-pm pores (Poretics Corp., Livermore, CA) were coated with 100 pg/ml type I collagen (Collagen Corp., Palo Alto, CA) diluted in 20 mM acetic acid, washed with phosphate-buffered saline and water, and placed on lower chambers of the chemotaxis apparatus which contained samples. Cells were detached from the culture dish with trypsin, washed, and incu- bated in the maintenance media for 2 h. Cells were then washed with a-MEM containing 1 mg/ml BSA (a-MEM/BSA), and 5 x lo4 cells sus- pended in 40 pl of a-MEMBSA were added to each upper chamber. After a 5-h incubation, the filter was removed from the apparatus, fixed with methanol, and stained with Diff-Quick (Baxter, Miami, FL). The number of cells on the lower surface of the filter was counted in four randomly chosen fields (x200 magnification). Random and directed mo- tility were analyzed by the checkerboard analysis (24) according to Stracke et al. (19).

Other Biological Assays-The effect of cytokines on the proliferation of cells was evaluated by the L3H1thymidine incorporation assay as described in Yamaguchi et al. (25), except that the assay was performed in the Corning 96-l/2 multiwell plates (0.16 cm2 culture surface per well; Corning, N Y ) instead of the standard 96-well plates. Scatter assay was performed according to Stoker et al. (15). Briefly, 5 x lo3 cells suspended in 100 pl of the maintenance medium were plated to the wells of the standard 96-well tissue culture plates which contain 100 pl of cytokines diluted in the serum-free medium. After an 18-h incuba- tion, the cells were fixed, stained with Diff-Quick, and analyzed for scattering of the cells.

Purification of Epitaxin-Serum-free conditioned media from HUT-12 cells were first precipitated with 40% saturation of ammonium sulfate. The supernatants containing epitaxin activity were concen- trated by ultrafiltration through a 30-kDa cut-off membrane (P"K Minitan plate; Millipore, Bedford, MA). After removing aggregates by centrifugation, the concentrates were applied to a column (2.5 x 17 cm) of phenyl-Sepharose (Pharmacia Biotech Inc.) preequilibrated with 50 IMI TridHCl containing 1.6 M ammonium sulfate and eluted by simul- taneous linear gradients of increasing ethylene glycol (0-50%) and de- creasing ammonium sulfate (1.6-0 M) in 50 IMI TrisiHC1 (pH 7.5). The combined active fractions from the phenyl-Sepharose column were then fractionated by a series of RP-HPLC with a Vydac C4 column (4.6 x 250 mm; Separations Group, Hisperia, CA) on a Shimadzu LC-600 HPLC system. For preparation of bioassay samples, aliquots of each fraction were lyophilized by a Speed-Vac (Savant, Farmingdale, N Y ) in the pres- ence of BSA (Fraction V; Sigma) and dissolved in the serum-free me- dium. Final concentration of BSA was adjusted to 1 mg/ml.

Extraction of Epitaxin Activity from SDS-PAGE-The active sample from the second RP-HPLC was concentrated by lyophilization and re- solved by SDS-PAGE in a 4-20% gel (Novex, San Diego, CA) under a nonreducing condition. The proteins in the gel were electrotransferred to a P W F membrane (Novex), and the membrane was cut into 12 pieces along the length of the blotted membrane. Each piece was eluted by incubating first with 40% acetonitrile a t 37 "C for 3 h and then with 40% acetonitrile, 0.05% trifluoroacetic acid at 50 "C for 30 min (26). The combined eluents were lyophilized in the presence of 100 pg of BSA, dissolved in 100 p1 of a-MEM, and assayed for the epitaxin activity as described above.

RESULTS Purification of Epitaxin-We initially found that conditioned

media of human embryonic lung fibroblasts IMR-90 and WI-38

and chemically transformed human fibroblast line HUT-12 stimulate the migration of HepG2 cells through porous poly- carbonate membranes in the modified Boyden chamber assay. Conditioned medium from HepG2 cells does not contain such an activity. We tested cytokines that have been shown to have motility-stimulating activities in the same assay in order to examine whether this activity was due to one of the known cytokines. These cytokines included acidic FGF, basic FGF, TGF-P1, TGF-/32, PDGF, IGF-I, SF/HGF, TNF-a, IL-6, inter- feron-y, and autotaxin. Only SF/HGF and IL-6 showed any substantial activities. While IL-6 slightly stimulated the mi- gration of HepG2 cells, SF/HGF showed a strong activity (see Fig. 3). However, HUT-12-conditioned medium had little effect on MDCK cells in the scatter assay (not shown), suggesting that the activity in the conditioned media of these cells was distinct from SF/HGF. Based on these results, we used HUT-12 cells as the source of conditioned media for the isolation of the putative novel motility factor.

Ten liters of HUT-12 serum-free conditioned media were first precipitated with 40% saturation of ammonium sulfate. Both precipitated and supernatant fractions contained activities that enhance motility of HepG2 cells, but most of the activities in the precipitated fraction were shown to be due to adhesion proteins such as fibronectin and laminin present in the condi- tioned media. The supernatant fraction (16.7 liters) which con- tained the main portion of the migration-stimulating activity was concentrated to 1000 ml by ultrafiltration and then frac- tionated on a column of phenyl-Sepharose. The activity was eluted as a broad peak at the end of the dual gradients of increasing ethylene glycol and decreasing ammonium sulfate (not shown).

Combined active fractions were then directly loaded to a Vydac C4 column preequilibrated with 0.1% trifluoroacetic acid and eluted by a linear gradient of 1040% acetonitrile. The recovery of activity in this step was low (- 13%) at least partly due to the denaturing solvent used for the RP-HPLC. In a separate experiment, the exposure of the active fraction ob- tained from phenyl-Sepharose in 0.1% trifluoroacetic acid for 1 h has been shown to reduce the activity to less than 25% of the original. Nevertheless, a peak of activity was detected in the fractions eluted with -29% acetonitrile (Fig. lA). Finally, these fractions were applied to the same C4 column and eluted with a 16-36% linear gradient of acetonitrile. The activity was again eluted at an concentration of about 29% acetonitrile (Fig. 1B). SDS-PAGE revealed a single band at 36 kDa on a silver-stained gel in the active fractions. The density of the band correlated with the migration stimulating activity across the elution pro- file (compare Fig. 1, B and C ) . By these isolation steps, the activity was purified 1038-fold relative to the supernatant of the ammonium sulfate precipitates (Table I). Fold-purification from the original conditioned medium could not be determined, since it contained substantial amounts of fibronectin and lami- nin which stimulate haptotactic migration of the cells.

To define further the molecule responsible for the migration stimulating activity, the active fractions from the second RP- HPLC were resolved in SDS-PAGE and electroblotted to a PVDF membrane, and eluents from slices of the membrane were assayed for the migration stimulating activity. The activ- ity was detected in a slice corresponding to 30-40 kDa which coincided with the 36-kDa band (Fig. 2). While it is possible that the activity might be associated with a very minor com- ponent other than the 36-kDa band present in the same gel slice, this result, together with the elution profile from the second RP-HPLC (Fig. E ) , supports the conclusion that the 36-kDa band on the SDS-PAGE represents the putative novel motility factor. We tentatively named this factor epitarin, based

14286 Epitaxin Is Motility Factor for Epithelial Cells

1 .o

0.5

0 Q: N

0.2

0.1

B I n

10 20 30 40

Fraction number

C Fraction number

21 23 25 27 29 31 33 35 37 39

94 - 67-

43 - 30 - m

20- I.h 14- ~ . .:._

+36

FIG. 1. Purification of epitaxin by RP-HPLC. A, combined active fractions of epitaxin from hydrophobic interaction chromatography on phenyl-Sepharose were first loaded to a Vydac C4 column (4.6 x 250 mm) preequilibrated with 0.1%' trifluoroacetic acid and eluted by a linear gradient of acetonitrile (10-60% in 100 min). B, active fractions from the first RP-HPLC (A) were reloaded on the same Vydac C4 col- umn and eluted by a shallow gradient of acetonitrile (16-36% in 100 min). In both cases, flow rate was 0.5 mVmin and 1 ml-fractions were collected. Each fraction was assayed for the epitaxin activity (bars) as described under "Experimental Procedures." C, SDS-PAGE analysis of fractions from the second RP-HPLC in a 4 2 0 % gradient gel under a nonreducing condition. Positions of molecular mass standards (Phar- macia) are shown on the left. The 36-kDa band coincided with the elution pattern of the epitaxin activity is indicated by an arrow on the right.

on its target cell specificity toward epithelial cell types (see below).

Epitaxin Is Distinct from Known Motility Factors and Cytokines-As described above, none of the tested cytokines, except for SF/HGF and IL-6, significantly stimulated the mi- gration of HepG2 cells. Although SFiHGF and IL-6 did show migration stimulating activities, biochemical data including the apparent molecular mass suggested that epitaxin is dis- tinct from these cytokines. To obtain further data confirming this, we examined the effects of neutralizing antibodies against

TABLE I Purification ofe~i tanin

Purification step Protein Total activity" Specific activity

mg units unitslflg HUT-12 conditioned medium 715' -' - 40% (NH,),SO,supernatants 338h 87,900 0.26 Phenyl-Sepharose 18h 70,300 3.9 Vydac C4 1 0.723" 9,300 12.9 Vydac C4 2 0.020" 5,400 270

" One unit of activity was defined as half maximal stimulation of HepG2 cell migration induced by the active fraction from phenyl-Sepha- rose chromatography.

Protein was determined by the BCAassay kit (Pierce, Rockford, IL). 'Activities in the original conditioned medium could not be deter-

mined due to the contaminating haptotactic activities derived from adhesion proteins present in the conditioned medium. " Protein was estimated by comparing the density of silver-stained bands with standard proteins on the SDS-gel.

A B

94- 67 - 43 - 30 - 20 -

350 260 190 135 100 74 54 40 30

..

14 - 16

12 9

- 1 1 I I

0 10 20 Number of migrated cells

FIG. 2. SDS-PAGE of purified epitaxin and recovery of epitaxin from gel slices. A, purified epitaxin was subjected to SDS-PAGE on a 4-20% gradient gel under a nonreducing condition and visualized by silver staining. Molecular mass standards (in kilodaltons) are shown on the left of the gel. B, an aliquot of purified epitaxin was concentrated, resolved in a 4 2 0 % gradient gel under nonreducing condition, and electrotransferred to a PVDF membrane, The membrane was cut into 12 slices, and each slice was eluted first with 40% acetonitrile at 37 "C for 3 h and then with 40% acetonitrile, 0.05% trifluoroacetic acid a t 50 "C for 30 min. Combined eluents were lyophilized in the presence of BSA and assayed for epitaxin activity with HepG2 cells. The numbers left of the graph indicate estimated molecular masses corresponding to each border of gel slices. The assay was performed twice with similar results.

SFiHGF and IL-6 on the epitaxin-stimulated cell migration. As shown in Fig. 3, neutralizing antibodies against SFiHGF and IL-6 had no effect on epitaxin-stimulated migration of HepG2 cells, while they inhibited the migration stimulated by the re- spective cytokines. Taken together, these results indicate that epitaxin is a molecule distinct from these known cytokines.

Biological Characterization of Epitaxin-The dose-response analysis of epitaxin on HepG2 cells showed that epitaxin is active in the range above 1 ng/ml with half maximal activity a t 4.2 ng/ml (Fig. 4). This estimate does not take into account the fact that the activity of epitaxin was reduced as much as 10-fold by the exposure to 0.1% trifluoroacetic acid, which was used in the RP-HPLC as a solvent. Therefore, the activity of epitaxin in physiological conditions may be substantially higher than this estimate.

To determine whether the epitaxin-stimulated motility is chemotactic (directed motility) or chemokinetic (random motil- ity), we performed the checkerboard analysis (24). This analy- sis demonstrated that epitaxin stimulates both chemotactic and chemokinetic responses (Fig. 5). Such a mixed response is

Epitaxin Is Motility Factor for Epithelial Cells 14287

FIG. 3. Effects of neutralizing antibodies against SFMGF and

taxin (Em 30 nglml), SF/HGF (30 ng/ml), or IL-6 (6000 units/ml) was IL-6 on the migration of HepG2 cells induced by epitaxin. Epi-

mixed with polyclonal anti-human HGF neutralizing antibodies ( d F / HGF, 200 pg/ml), monoclonal anti-human IL-6 neutralizing antibodies ((UIL-6; 200 pg/ml), polyclonal anti-human serum albumin (RoLHSA; 200 pg/ml), or monoclonal anti-human serum albumin (MaHSA; 200 pglml). After incubating for 1 h at 37 “C, the mixtures were added to the lower chambers of the 48-well chemotaxis apparatus. The migration assay was performed with HepG2 cells as described under “Experimental Procedures.” Data represent the mean * 1 S.D. ( n = 3).

ETX (ng/ml) FIG. 4. Dose-dependent effect of epitaxin on the migration of

HepG2 cells. The purified epitaxin (Em) was lyophilized in the pres- ence of BSA, dissolved in serum-free medium, and assayed for the activity to stimulate the migration of HepG2 cells. Data represent the mean * 1 S.D. ( n = 3).

similar to the responses induced by autotaxin, SFMGF, and insulin (5, 19, 27). In contrast to these factors, the activity of IGF-I has been shown to be almost entirely chemotactic (5).

Like epitaxin, SF/HGF stimulates the migration of HepG2 cells (see Fig. 3). Moreover, SFMGF has an activity to scatter the colonies of a number of epithelial cell lines including HepG2 and MDCK cells (15, 17, 28). Comparison of the activities of epitaxin and SF/HGF in the scatter assay showed that, while SF/HGF scattered both HepG2 and MDCK cells as reported, epitaxin did not cause any discernible morphological changes in either cell types (not shown). These results indicate that, though both stimulate the motility of HepG2 cells, the mecha- nism of the action is different for SF/HGF and epitaxin.

Upper chamber (ng/ml)

- E

30 10 3 0 ETX . v 2 12.5f2.8 7.3f2.3 6.0f2.1 3.35 1.8 0

FIG. 5. Checkerboard analysis of the motility of HepG2 cells stimulated by epitaxin. Varying concentrations of epitaxin (ETX) ranging from 0 to 30 ng/ml were added to the upper and lower chambers of the 48-well chemotaxis apparatus to create different sets of concen- tration gradient. Numbers of migrated HepG2 cells were counted as described under “Experimental Procedures.” Data represent means * 1 S.D. for each point in the checkerboard ( n = 3).

The effects of epitaxin on cell proliferation were examined by L3H1thymidine incorporation assays. As illustrated in Fig. 6, epitaxin modestly (-50%) stimulated DNA synthesis of HepG2 cells. In contrast, SF/HGF showed a slight growth inhibitory effect on these cells. The growth-inhibitory activity of SFMGF on various hepatoma cells has been described previously (28, 29). This result again suggests that signaling pathways of these two factors may be distinct.

Epitaxin Has a Target Cell Specificity toward Epithelial Cells-Target cell specificity of epitaxin was studied with a series of rat hepatoma and several epithelial and fibroblast cell lines. Hepatoma cell lines included H411, Fao, and C2, all of which are descendants of Reuber hepatoma H35 (30). H41I and Fao are clonal cell lines with well differentiated phenotypes expressing a variety of liver-specific genes. C2 is a Fao-derived clone with a dedifferentiated phenotype lacking the expression of liver-specific genes including albumin (30). While well dif- ferentiated H4II cells did not respond to epitaxin, dedifferen- tiated C2 cells were stimulated by epitaxin more than HepG2 cells (Table 11). The motility of H4II and Fao cells was stimu- lated by epitaxin, but the magnitude of the response was less than that of C2 cells. Thus, the responsiveness toward epitaxin appears to inversely correlate with the differentiation of hepa- toma cells.

Several epithelial cell lines, including MDCK, ZR-75-1, A549, and NBT-11, have also been tested. While MDCK cells are de- rived from normal kidney epithelia and retain a highly differ- entiated phenotype, the other cell lines are derived from tu- mors. Epitaxin stimulated the migration of all but the MDCK cells (Table 11). Thus, consistent with the case of hepatoma cells, it appears that highly differentiated cells are less respon- sive to epitaxin than tumor-derived epithelial cells. The migra- tion of the MDCK cells was not stimulated by epitaxin even at concentrations as high as 100 ng/ml. In contrast, SF/HGF stimulated the migration of MDCK cells in the concentration range of 10-30 ng/ml (Table II), which is consistent with pub- lished data (15). ZR-75-1 cells have been shown to respond to IL-6 by cell scattering (9). In the Boyden chamber assay, how- ever, IL-6 did not stimulate the migration of ZR-75-1 cells, whereas epitaxin did.

Responsiveness of fibroblastic cells to epitaxin has been stud- ied with a series of human fibroblast lines HUT-12, HuT14, and HUT-14T. While HUT-12 is nontumorigenic in the nude mouse assay, HUT-14 is tumorigenic and HUT-14T is highly tumori- genic (31). Although each of these cell lines showed significant levels of unstimulated motility, in none of them was motility significantly enhanced by epitaxin (Table 11).

Epitaxin Is Motility Factor for Epithelial Cells

2-1

0 1 10

ETX or SF/HGF (ng/ml) FIG. 6. Effects of epitaxin and SF/HGF on the DNAsynthesis of

HepG2 cells. [3HlThymidine incorporation assay was performed as described by Yamaguchi et al. (25). Confluent monolayers of HepG2 cells in wells of the Corning 96-1/2 well tissue culture plates (0.16 cm2 cul- ture surface) were treated for 24 h with varying concentrations of epi- taxin ( E m 0) or SFLHGF (A) diluted in wMEM containing 1% FCS. Cells were then pulsed with [3H]thymidine for 2 h, and the incorpora- tion into DNA was measured after precipitation with 10% trichloroace- tic acid. The experiment was performed in duplicate and repeated twice with similar results.

TABLE I1 Effect of epitaxin on various cell lines

Number of cells migrated at:”

0 nglml 30 nglrnl Cell line Cytokine

Rat hepatoma lines HI11 Fao c 2

MDCK MDCK

Epithelial lines

ZR-75-1 ZR-75-1 A549 NBT-I1

Fibroblast lines HUT-12 HUT-14 HUT-14T

Epitaxin Epitaxin Epitaxin

Epitaxin SF/HGF Epitaxin

Epitaxin Epitaxin

Epitaxin Epitaxin Epitaxin

IL-6

0 0

8.3 t 1.3

0 0 0 0

53.0 t 10.2 19.8 f 6.2

118.4 t 22.0 39.6 t 6.5 90.6 .c 15.5

13.0 f 2.2 10.7 f 1.2

224.3 f 11.3

0

23.8 t 2.0 36.3 t 8.5

0 182.8 t 17.2 279.3 t 25.5

127.2 t 14.3 38.0 t 6.2 83.1 f 7.3

mental Procedures” with the cell lines indicated in the table. Suspen- The migration assay was performed as described under “Experi-

sions of 5 x lo4 cells in serum-free media containing 1 mg/ml of BSA were added to each well and incubated for 5 h except for H4II and Fao cells which were incubated for 18 h. Data represent the mean t l S.D. (n = 3).

DISCUSSION

In the present study, we described the purification and bio- logical characterization of epitaxin, a 36-kDa motility factor for epithelial cells. Epitaxin appears to be distinct from any known motility factors and other cytokines that have been reported to stimulate cell motility. With the exception of SFMGF and IL-6, various purified cytokines failed to significantly change the migration of HepG2 cells in a Boyden chamber assay, whereas epitaxin had a strong effect on these cells. Furthermore, none of these cytokines has an apparent molecular mass around 36 kDa.

Although SF/HGF has a similar effect on the motility o f HepG2 cells, we concluded that SFMGF and epitaxin are dis- tinct molecular entities for the following reasons. First, SF/ HGF is a much larger molecule than epitaxin. Active SF/HGF

preparations isolated from various sources migrate as a single band of 62-75 kDa in nonreducing SDS-PAGE (16, 18, 32). A naturally occurring 28-kDa variant of HGF which acts as an antagonist for HGF does not have the biological activity by itself (33). Second, while epitaxin appears to be an acidic pro- tein as it binds to anion exchange column but not to heparin column (not shown), both the active and the 28-kDa forms of SFMGF are basic, heparin-binding proteins (16,32-34). Third, the neutralizing experiments clearly demonstrate that the ac- tivity of epitaxin is not mediated by SFMGF (Fig. 3). Finally, epitaxin and SFMGF are distinct with respect to their biologi- cal activities; SFMGF and epitaxin have opposite effects on the growth of HepG2 cells (Fig. 6), and epitaxin does not scatter HepG2 or MDCK cells, while SF/HGF scatters both (see text). These results suggest that, although both epitaxin and SF/ HGF stimulate the migration of HepG2 cells, the underlying signaling mechanisms of this effect may be distinct.

Our biological analyses suggest that epitaxin has a target cell specificity toward epithelial cell types. Among epithelial cell lines tested, transformed cell lines derived from malignant tumors responded to epitaxin, while MDCK cells derived from normal kidney epithelia did not. A similar tendency is observed in rat hepatoma cell lines, where dedifferentiated cells are stimulated more highly by epitaxin than are well differentiated cells. These results may suggest that epitaxin does not over- come or down-regulate strong cell-cell adhesions of highly dif- ferentiated epithelial cells. In contrast, SF/HGF dissociates colonies of epithelial cells which are bound together by cell-cell adhesion (15, 27). Thus, the difference in the biological activi- ties of epitaxin and SFMGF might derive from their different effects on the regulation of cell-cell adhesion.

Several normal and transformed fibroblast lines produce epi- taxin, but they do not respond to epitaxin. On the other hand, conditioned medium from HepG2 cells themselves contained no epitaxin activity, whereas their migration is stimulated by epi- taxin. Thus, although more cell lines should be examined be- fore drawing definitive conclusions, it is likely that epitaxin acts in a paracrine fashion, as does SF/HGF (15). In this re- spect, epitaxin is distinct from most of other motility factors that are thought to operate in an autocrine fashion, including AMF, MSF, autotaxin, and SDMF (11, 14, 19, 20).

Although it is dificult t o predict the physiological role of epitaxin at present, it is conceivable that epitaxin may be in- volved in the invasion and metastasis of epithelial tumors. Considering that epitaxin is produced by mesenchymal cells, tumor cells which have invaded into connective tissues beneath the basement membranes might be more strongly stimulated by epitaxin. Alternatively, paracrine motility factors such as epitaxin and SF/HGF may be produced inside the tumor by stromal cells, thereby enhancing the dissociation of tumor cells from primary lesion. Thus it will be important to study distri- bution of epitaxin in tumor tissues in order to define its patho- physiological role in tumor invasion and metastasis. It will also be necessary to clone and sequence epitaxin cDNA. Determi- nation of its complete primary structure will establish its mo- lecular identity and relationship with other motility factors and cytokines.

Acknowledgments-We thank Drs. Toshikazu Nakamura (Osaka University) and Mary Stracke (National Cancer Institute) for their generous gifts of SFLHGF and autotaxin, respectively, and Dr. Erkki Ruoslahti for helpful discussions and critical reading of the manuscript.

Note Added in Proof-Recently, we obtained a single N-terminal amino acid sequence by microsequencing analysis of an HPLC-purified epitaxin preparation. This 20-amino acid sequence showed no signifi- cant homology to any protein in protein sequence data banks.

Epitaxin Is Motility Factor for Epithelial Cells 14289

REFERENCES Rieder, H., Fonatsch, C., Tsubouchi, H., Hishida, T., Daikuhara, Y., and Birchmeier, W. (1991) Proc. Natl. Acad. Sci. U. S . A. 88, 7001-7005

2. Valles, A. M., Boyer, B., Badet, J., Tucker, G. C., Barritault, D., and Thiery, J.-P. 1. Stoker, M., and Gherardi, E. (1991) Biochim. Biophys. Acta 1072, 81-102 18. Naldini, L., Weidner, K. M., Vigna, E., Gaudino, G., Bardelli, A., Ponzetto, C.,

Narsimhan, R. P., Hartmann, G., Zarnegar, R., Michalopoulos, G. K., Birch-

3. Moscatelli, D., Presta, M., and Rifkin, D. B. (1986) proc. Natl. Acad. sei. 19. Stracke, M. L., Krutzsch, H. C., Unsworth, E. J., kes t ad , A,, Cioce, V., Schiff- meier, W., and Comoglio, P. M. (1991) EMBO J. 10, 2867-2878

4. Grotendorst, G. R., SePP% H. E. J., meinman, H. K., and Martin, G. R. (1981) 20. Koyama, N., Harada, K.,Yamamoto,A., Morisaki, N., Saito, Y., andYoshida, S. mann, E., and Liotta, L. A. (1992) J. Biol. Chem. 267,2524-2529

(1993) J. B i d . Chem. 268, 13301-13308

(1990) Proc. Natl. Acad. Sci. U. S . A. 87,1124-1128

U. S. A. 83, 2091-2095

Proc. Natl. Acad. Sci. U. S. A. 78, 3669-3672

Res. Commun. 153, 10761083 Krutzsch, H. C., Liotta, L. A., and Schiffmann, E. (1988) Biochem. Biophys. 65-84

and Helman, L. J. (1990) Cell Growth & Differ. 1, 325-331

Med. 165,251-256

Reiss, M., and Kacinski, B. M. (1991) Cancer Res. 51, 5315-5321

B. (1989) J. Exp. Med. 170, 1649-1669

123-132 28. Tajima, H., Matsumoto, K., and Nakamura, T. (1992) Exp. Cell Res. 202,

and Schiffmann, E. (1986) proc. Natl. Acad. sei. u. s. A. 83,33023306 29. Shiota, G., Rhoads, D. B., Waw, T. c., Nakamura, T., and Schmidt, E. (1992)

chem. Biophys. Res. Commun. 146.99E-1002 30. Ott, M.-O., Sperling, L., Herbomel, P., Yaniv, M., and Weiss, M. C. (1984)

51,3507-3511 31. Leavitt, J., Bushar, G., Kakunaga, T., Hamada, H., Hirakawa, T., Goldman, D.,

Natl. Acad. Sci. U. S. A. 86, 2438-2442 32. Gherardi, E., Gray, J., Stoker, M., Perryman, M., and Furlong, R. (1989) Proc.

239-242 33. Chan, A. M.-L., Rubin, J. S., Bottaro, D. P., Hirschfield, D. W., Chedid, M., and

Cell B i d . 111, 2097-2108 34. Nakamura, T., Teramoto, H., and Ichihara, A. (1986) Proc. Natl. Acad. Sci.

5. Stracke, M. L., Kahn, E. c., hnavoofian, s. A.3 Wilson, L. L., SalomOn, D.2 21. Hart, 1. R,, Goode, N. T,, and Wilson, R. E. (1989) &whim. Biophys. Acta 989,

6. El-Badry, 0. M., Minniti, C., Kohn, E. C., Houghton, P. J., Daughaday, W. H.,

7. Postlethwaite,A. E., Keski-Oja, J., Moses, H. L., and Kang,A. H. (198715. Exp. ;:: ~ ~ ~ ~ ~ ~ T ~ , ' t " ~ ~ ~ ~ ~ ~ ~ ~ 7 ~ ~ ,

8. Rosen, E. M., Goldberg, I . D., Liu, D., Setter, E., Donovan, M. A,, Bhargava, M., 25' Yamaguchi' D' M'' and RuOsl*ti' E' (1990) Nature 346' 281-284

9. Tamm, I., Cardinale, I., Krueger, J., Murphy, J. S., May, L. T., and Sehgal, P.

22. Liotta, L. A,, and Stetler-Stevenson W. G. (1991) Cancer Res. 61, (suppl.) 5054s-5059s

26. Stone, K. L., LoPresti, M. B., Crawford, J. M., DeAngelis, R., and Williams, K. R. (1989) A Practical Guide to Protein and Peptide Purification for Micro- sequencing, Academic Press, New York

10. Yoshida, K., Ozaki, T., Ushijima, K., and Hayashi, H. (1970) Znt. J. Cancer 6, 27. Stoker, M. (1989) J. physioz. 139, 565-569

11. Liotta, L. A,, Mandler, R., Murano, G., Katz, D. A., Gordon, R. K., Chiang, P. K., 423431

12. Atnip, K. D., Carter, L. M., Nicolson, G. L., and Dabbous, M. K. (1987) Bio- Proc. Natl. Acad. Sci. U. s. A. 89, 373-377

13. Silletti, S., Watanabe, H., Hogan, V., Nabi, I. R., and Raz,A. (1991) Cancer Res. EMBO J. 3,2505-2510

14. Grey, A,"., Schor, A. M., Rushton, G., Ellis, I . , and Schor, S. L. (1989) Proc. and Merril, C. (1982) Cell 28,259-268

15. Stoker, M., Gherardi, E., Perryman, M., and Gray, J. (1987) Nature 327, Natl. Acad. Sci. U. S . A. 86, 5844-5848

16. Weidner, K. M., Behrens, J., Vandekerckhove, J., and Birchmeier, W. (1990) J. Aaronson, S. A. (1991) Science 264, 1382-1385

17. Weidner, K. M., Arakaki, N., Hartmann, G., Vandekerckhove, J., Weingart, S., U. S . A. 83, 64894493