Protein Kinase Cϵ Mediates Polymeric Fibronectin Assembly on the Surface of Blood-borne Rat Breast Cancer Cells to Promote Pulmonary Metastasis*

Malignant breast cancer cells that have entered the blood circulation from primary mammary fat pad tumors or are grown in end-over-end suspension culture assemble a characteristic, multi-globular polymeric fibronectin (polyFn) coat on their surfaces. Surface polyFn is critical for pulmonary metastasis, presumably by facilitating lung vascular arrest via endothelial dipeptidylpeptidase IV (CD26). Here, we show that cell-surface polyFn assembly is initiated by the state of suspension, is dependent upon the synthesis and secretion of cellular Fn, and is augmented in a dose- and time-dependent manner by plasma Fn. PolyFn assembly is regulated by protein kinase Cϵ (PKCϵ), which translocates rapidly and in increasing amounts from the cytosol to the plasma membrane and is phosphorylated. PolyFn assembly is impeded by select inhibitors of this kinase, i.e. bisindolylmaleimide I, Ro-32-0432, Gö6983, and Rottlerin, by the phorbol 12-myristate 13-acetate-mediated and time-dependent loss of PKCϵ protein and decreased plasma membrane translocation, and more specifically, by stable transfection of lung-metastatic MTF7L breast cancer cells with small interfering RNA-PKCϵ and dominant-negative PKCϵ constructs (e.g. RD-PKCϵ). The inability to assemble a cell surface-associated polyFn coat by knockdown of endogenous Fn or PKCϵ impedes cancer cells from metastasis to the lungs. The present studies identify a novel regulatory mechanism for polyFn assembly on blood-borne breast cancer cells and depict its effect on pulmonary metastasis.

in cancer cell lines selected for enhanced lung colonization (8 -11). This pro-metastatic role of Fn is multifaceted, affecting several steps of the metastatic cascade by modulating cell adhesion, motility/invasion, cell cycle progression, and cell survival (for review, see Refs. [12][13][14][15][16]. By analyzing cancer cells that had entered the blood circulation from malignant breast cancers implanted into the mammary fat pad of rats or mice, we discovered that blood-borne tumor cells were decorated with a unique, multiglobular coat of polymeric Fn (polyFn) (9,17). PolyFn aggregates appeared to arise from focal accumulations of endogenous, cell surface-immobilized ("linearized") Fn, which served as scaffolds for further Fn-self-assembly from Fn recruited from blood plasma (pFn) (9). Aggregates typically became increasingly deoxycholate-insoluble as they increased in size with time of incubation of suspended cancer cells in serum-containing medium in vitro. Biochemically, aggregates impressed as prominent, insoluble (covalently bonded) Fn polymers sitting on top of the stacks of SDS-polyacrylamide gels and, immunocytochemically, as large globules randomly dispersed over the entire cancer cell surface (9). This "cluster arrangement" of polyFn was shown to have the following functional implications. First, the conversion of Fn from the globular state of soluble Fn to the linearized state of insoluble, surface-associated Fn aggregates is associated with exposure of a novel, cryptic binding domain for the lung endothelial cell address in dipeptidylpeptidase IV (DPP IV) (9,(17)(18)(19)(20)(21). This DPP IV binding domain is present as a consensus motif in each of the 13th, 14th, and 15th type III repeats of Fn (17). Second, the DPP IV binding specificity for linearized (polymeric), but not for globular (soluble) pFn, allows tumor cell adhesion to endothelial DPP IV in the presence of high pFn concentrations (9). Third, the large Fn aggregates on cancer cell surfaces allow multiple binding interactions with endothelial DPP IV molecules, thereby generating adhesion strengths between cancer cells and endothelial cells that are able to withstand the rigors of hemodynamic shear stresses (9). The importance of the Fn/DPP IV-docking mechanism is substantiated by our discovery that synthetic peptides directed against the DPP IV binding domain in the 13th, 14th, or 15th type III repeats of Fn as well as a polypeptide encompassing the bulk of the extracellular domain of DPP IV dramatically impeded pulmonary metastasis in a rat breast cancer model (19,20). These data are consistent with our finding that colonization of the lungs was greatly diminished in Fischer 344/CRJ rats, in which DPP IV is mutated causing a significantly decreased DPP IV protein expression in pulmonary endothelia (21) as well as in DPP IV Ϫ/Ϫ mice (19). DPP IV Ϫ/Ϫ mice injected with lung-metastatic cancer cells lived significantly longer than their wild-type counterparts, an outcome granted by the formation of significantly fewer and smaller lung colonies. 3 Although we have firmly established a critical dependence between the ability of the cancer cells to assemble an insoluble, globular polyFn-surface coat and lung colonization in an experimental metastasis model (9,17), we still do not know whether blood-borne cancer cells use their own cellular Fn (cFn) or rely on ubiquitous pFn to assemble their polyFn surface coat, how cancer cells regulate the polyFn build-up, and how Fn cell surface deposits are transformed into covalently bonded, insoluble aggregates (9). To answer some of these questions we examined polyFn genesis in lung-metastatic MTF7L rat breast cancer cells subjected to end-over-end (EoE) suspension culture in serum-or pFn-containing medium, which together induce and augment the build-up of a polyFn surface coat similar to that observed on tumor cells that have entered the blood circulation (9). The data presented here show that assembly of the prometastatic, multiglobular polyFn surface coat on MTF7L breast cancer cells depends upon the synthesis and secretion of endogenous, cellular Fn (Fn1: EDAϩEDBϩIIICS 120 -Fn) and is regulated by membrane-translocation and Ser/Thr phosphorylation of protein kinase C⑀ (PKC⑀). Inhibitors of protein synthesis, protein secretion, and novel PKC isoforms (nPKCs) as well as transfection of MTF7L cells with the PKC⑀ regulatory domain (RD) or PKC⑀ siRNA species all substantially decrease the ability of the cancer cells to assemble a polyFn surface coat. Functionally, inability to assemble polyFn is associated with failure to colonize the lungs. Together, our studies provide novel insights of the regulation of Fn in suspended (bloodborne) breast cancer cells and provide a renewed appreciation for the previously recognized role of Fn in metastasis (1-7).
Cell Cultures-MTF7L cells were derived from a lung metastasis generated by tail-vein injection of MTF7 breast cancer cells (obtained from Dr. D. R. Welch, University of Alabama at Birmingham, Birmingham, AL) into Fischer 344 rats. At an intravenous inoculation dose of 2 ϫ 10 5 cells per rat, MTF7L cells consistently produce in excess of 400 lung colonies. Cells were grown in culture in Dulbecco's modified Eagle's medium (DMEM) containing 10% heat-inactivated FBS. For EoE suspension culture, MTF7L cells were grown to 80 -90% confluence, then removed from the growth surface by trypsinization (0.25% trypsin, 0.02% EDTA in phosphate-buffered saline (PBS) for 10 min at 37°C), washed twice in DMEM containing 10% FBS, and subjected to EoE suspension culture for 1 h (or as indicated) in 2-ml centrifuge tubes in DMEM plus 20% FBS at a concentration of 5 ϫ 10 6 cells/ml (9,17,20). Tumor cells were used for all experiments within 10 passages from frozen stocks that were tested for metastatic performance immediately before freezing.
For metabolic labeling, MTF7L cells in logarithmic growth phase were labeled with [ 35 S]methionine (0.33 mCi/3 ml) in methionine-free DMEM (both from MP Biomedicals, Solon, OH) containing 20 M methionine and 10% dialyzed, Fn-free FBS as previously described (9). For 32 P labeling, cells were serum-starved overnight, then incubated for 4 h in phosphatefree DMEM containing 100 Ci/ml [ 32 P]orthophosphate (ICN Biochemicals, Irvine, CA) and washed in 3 changes of PBS. Labeled cells were subjected to EoE suspension culture as describe above and immediately processed for biochemical analyses.
MTF7L cells grown to 70% confluence were transiently transfected with the above vector constructs or vector alone using Lipofectamine Plus as described by the manufacturer (Invitrogen). Transfection rates assessed by expression of GFP that is either tagged to the cDNA of interest or co-transfected at a ratio of 1:50 with the cDNA of interest were 20 -30%. Cells were used in the various assays 48 h after transfection unless otherwise stated. Stable clones were obtained by hygromycin selection (750 g/ml). In some cases hygromycin-selected clones (cl) were further selected by fluorescence-activated cell sorting (FACS) for optimal expression.
Flow Cytometry-FACS was used to quantify Fn expression on MTF7L breast cancer cell surfaces (9,17). Tumor cells that had been subjected to EoE suspension culture were washed twice in DMEM containing 1% bovine serum albumin (BSA), then incubated with rabbit anti-Fn[pan] antibody diluted 1:100 in PBS containing 1% BSA (PBS-BSA) for 1 h at 4°C. After washing in PBS-BSA, tumor cells were stained with PE-conjugated donkey anti-rabbit antiserum in PBS-BSA for 1 h at 4°C and fixed in 2% paraformaldehyde in PBS. In select experiments cells were stained with mouse anti-Fn[EDA] (diluted 1:50) and PE-conjugated goat antimouse antiserum. FACS analysis was performed on a Coulter Epics Profile (Coulter Electronics, Hialeah, FL). Nonspecific fluorescence was accounted for by incubating tumor cells with non-immune rabbit serum instead of primary antibody. To quantify the effect of overexpressed or knocked-down proteins on polyFn assembly, we generate bivariate distributions of red fluorescence (y axis: cells stained with anti-Fn antibodies and PE-conjugated secondary antibodies) and green fluorescence (x axis: same cells expressing GFP-tagged protein or co-transfected with GFP and cDNA of interest). The levels of polyFn expression in the cell population that emitted high GFP fluorescence were taken as a reflection of the effect of the transfected cDNA on polyFn assembly. To assess the effect of inhibitors of cell signaling, tumor cells were incubated with inhibitor 30 min before (adherent) and throughout EoE suspension culture, then subjected to polyFn quantification as described above. Controls were tumor cells incubated in equimolar inhibitor solvent concentration.
Cell Fractionation-MTF7L cells and transfectants thereof were incubated in EoE suspension culture in DMEM containing 20% FBS for the indicated periods of time. Cells were washed in PBS, collected by centrifugation, and resuspended in 0.5 ml ice-cold Tris buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM NaF, 0.1 mM NaVO 4 , 20 M leupeptin, 0.1% aprotinin, and 1 mM phenylmethylsulfonyl fluoride). Cells were disrupted by two 15-s cycles of sonication at 4°C using a microprobe sonicator at maximum power. After removal of unbroken cells and nuclei by centrifugation (500 ϫ g for 5 min at 4°C), supernatants were centrifuged at 16,300 ϫ g for 15 min at 4°C. The resulting supernatant was designated the cytosolic fraction. The pellet was solubilized in 0.5 ml of buffer A containing 1% Triton X-100 for 1 h at 4°C EoE, then centrifuged at 16,300 ϫ g for 15 min at 4°C. The detergent-solute was designated the membrane fraction. Twenty micrograms of protein from both the cytosolic and membrane fractions were separated by SDS-PAGE (10%), transferred to nitrocellulose membrane at 4°C, and probed by Western blotting as described (9).
Tumor Cell Proliferation Assay-MTF7L cells and clones thereof were seeded into 96-well microtitration plates (500 cells/well) and incubated in DMEM containing 10% FBS. Cell growth was monitored in daily intervals for up to 4 days. At the end of each incubation period, tumor cells were fixed with 2% paraformaldehyde in PBS and then stained with 0.5% crystal violet in 20% methanol as described (21). Absorbance was read on a microplate reader (Bio-Teck Instruments) at 562 nm and graphed as a function of time of incubation.
Isolation of Blood-borne Cancer Cells from Tumor-bearing Rats-MTF7L cancer cells (1 ϫ 10 6 cells/50 l of DMEM) were injected into the 4th (left ϩ right) mammary fat pads of six 6-week-old female Fischer 344 rats. At a tumor diameter of ϳ2 cm, rats were anesthetized by an intraperitoneal injection of sodium pentobarbital (65 mg/kg body weight), and blood was collected by cardiac puncture. Pooled, EDTA-treated blood was transferred to precooled 50-ml centrifuge tubes containing 15 ml of OncoQuick tumor enrichment medium below a porous barrier (Greiner Bio-One, Longwood, FL) and centrifuged at 1600 ϫ g for 20 min at 4°C in a swing-out rotor as described by Rosenberg et al. (28). After a second round of centrifugation of the fluid in the upper compartment, cells were washed with PBS-BSA and stained with anti-Fn antibodies as described above. Tumor cells were readily differentiated from contaminant blood mononuclear cells by size and intensity of anti-Fn staining in comparison with preparations from MTF7L cell-spiked blood.
Lung Colony Assays-To determine the effects of Fn or PKC⑀ knockdown on the lung colony efficiency, selected MTF7L clones as well as wt or vector-transfected MTF7 cells (2 ϫ 10 5 cells/0.3 ml DMEM/rat) were injected into the lateral tail vein of 5-week-old female Fischer 344 rats (5-9 rats/experiment) as described (9,21). Rats were sacrificed 30 days after tumor cell injection. Means and standard deviations of the lung weights, the number of lung colonies, and the colony diameters were determined for each cell variant (21).

Endogenous Fn Is Required for PolyFn Assembly on Suspended
Breast Cancer Cells-Breast cancer cells entering the blood circulation assemble a characteristic surface coat of globular polyFn as evidenced by anti-Fn[pan] staining of cancer cells isolated from the blood of mammary tumor-bearing rats (Fig.  1A). This phenomenon is mimicked when cancer cells are grown in EoE suspension culture in the presence of serum (9). To examine whether polyFn assembly is mediated by exogenous pFn alone or whether it critically depends on the synthesis of endogenous cFn, we incubated MTF7L breast cancer cells, which express high levels of Fn1 (EDAϩEDBϩCSIII 120 ) in EoE suspension culture in DMEM containing either 20% complete FBS or FFS. FFS-treated tumor cells stained with either anti-Fn[pan] (Fig. 1B) or anti-Fn[EDA] (Fig. 1B, inset) exhibited similar numbers of polyFn globules on their surfaces as FBStreated cells (Fig. 1C). However, the addition of complete FBS to the suspension medium significantly augmented the size of individual polyFn globules relative to those generated by FFS (Fig. 1, B and C). The serum effect on polyFn assembly appeared to be mediated only in part by pFn as shown by FACS analysis of MTF7 cells incubated in EoE suspension culture in the presence of (i) 20% FBS (Fig. 1, D and E), (ii) 6 g/ml pFn (amount present in 20% FBS; 22), or (iii) 20% FFS (Fig. 1E). Both 20% FFS and pFn (6 g/ml) were significantly less effective in promoting polyFn assembly than FBS (Fig. 1E). The potency of FFS to promote polyFn assembly could be restored to that of FBS by the addition of 6 g/ml pFn (Fig. 1E, inset). Thus, FFS appeared to promote surface deposition of cellular Fn (Fn1) on suspended MTF7L cells, whereas pFn in a time-and dose-dependent manner contributed to augmentation of the Fn polymers by Fn-Fn self-assembly (Fig. 1F).
To further test the hypothesis that surface deposition of a scaffold of endogenous cFn was important for the initiation of polyFn assembly, we examined whether inhibitors of protein synthesis (e.g. cycloheximide) (29) or secretion (e.g. monensin, brefeldin A) (30, 31) would affect polyFn assembly on sus- pended MTF7L breast cancer cells. Cycloheximide, monensin, and brefeldin A all dramatically impeded polyFn assembly on MTF7L breast cancer cells in the presence of 20% FBS (Fig. 2). Because these compounds might have also impaired surface expression of Fn receptors, we used RNA interference to knock down endogenous cFn in MTF7L (Fn1) cells and to examine whether the specific inhibition of MTF7L-cFn would affect polyFn assembly. Biochemically, stable clones derived from siRNA-Fn-transfected cells exhibited a significant decrease in the expression cFn protein as depicted by Western blotting (Fig. 3A, clones cl 1 and cl 3 ) and by a dramatic decrease in insoluble, metabolically labeled polyFn residing on top of the stack of the polyacrylamide gel (Fig. 3B, clone cl 1 ). Functionally, knockdown of endogenous, cellular Fn was associated with impaired surface polyFn assembly on MTF7L cells as shown by FACS (Fig. 3C) and immunocytochemistry (Fig. 3, D and E), albeit all tumor cells were incubated in the presence of 20% FBS containing ϳ6 g/ml pFn. Unspecific nucleotide sequences had no effect (Fig. 3, A-E, mcl us ). Stable siRNA-Fn clones were also used to test their lung metastatic potential in lung colony assays. In accordance with the level of knockdown of endoge-nous Fn, lung colonization, assessed by the averages of lung weights, colony numbers, and colony diameters, was significantly decreased relative to wtMTF7L cells or tumor cells that were stably transfected with an unspecific nucleotide sequence (Fig. 4, siFn-cl 1 ).
PolyFn Assembly Occurs in a PKC-dependent Manner-To identify the signaling cascade that regulates polyFn assembly, we screened a diverse group of inhibitors of cell signaling for their ability to impede polyFn assembly as determined by routine FACS analyses of anti-Fn[pan]-stained MTF7L breast cancer cells subjected for 1 h to EoE suspension culture in FBS (20%)-containing medium in the presence or absence of inhibitor. These inhibitors included pertussis toxin, PD98059, SU6656, PP2, wortmannin, LY294002, JAKI, and Y27632. None of these inhibitors had any effect on the assembly of the polyFn surface coat of MTF7L breast cancer cells (data not shown). Consistent with the negative result obtained with the ROCK1/2 inhibitor Y27632 is the finding that the Rho activity, associated FIGURE 2. Inhibitors of protein synthesis and secretion prevent polyFn assembly. Adherent MTF7L rat breast cancer cells were exposed for 30 min to 10 g/ml cycloheximide (A), 1 M monensin (B), or 5 M brefeldin A (C), then subjected to EoE suspension cultured in DMEM ϩ 20% FBS in the presence of the respective drug or drug solvent (S). After a 1-h incubation period, cells were stained with anti-Fn[pan] antibodies as described in Fig. 1A (control, MTF7L cell stained with non-immune rabbit IgG (rIgG)). Tumor cells treated with cycloheximide, monensin, and brefeldin A exhibit a dramatic reduction in the cell surface-associated polyFn, relative to those treated with drug solvent. with Fn-matrix assembly and stress fiber formation in adherent cells (32,33), steadily decreased during the polyFn assembly phase on suspended breast cancer cells. Moreover, transfection with constitutively active Rho (RhoAQ63L) suppressed polyFn assembly, whereas transfection with dominant negative Rho (RhoAT19N) appeared to promote polyFn assembly. 4 In contrast, polyFn assembly was dramatically reduced by U73122, an inhibitor of PLC (34), and the PKC inhibitor calphostin C, which competes with the diacylglycerol and phorbol ester binding site of conventional PKCs (cPKCs) and nPKCs (35) (Fig.  5A). Together, these findings suggest involvement of a cPKC or nPKC isoform but not an atypical PKC isoform, presumably acting downstream of PLC in polyFn assembly on MTF7L cancer cell surfaces (36).
PKC⑀ Is the Mediator of PolyFn Assembly and Metastasis-Our attempts to identify the PKC isoform responsible for polyFn assembly were preceded by determining the c-and nPKC expression levels in MTF7L breast cancer cells. As reported for adherent, metastatic mammary tumor variants by Kiley et al. (37), suspended MTF7L cells strongly express PKC␣, PKC␦, and PKC⑀, both at the mRNA and protein levels (Fig.  5B). PKC mRNA is expressed moderately and protein weakly, and PKC mRNA is expressed weakly and protein non-detectably (Fig. 5B). Next, we used select PKC inhibitors to narrow the spectrum of PKC isoforms involved in polyFn assembly. Failure of high doses of the cPKC inhibitors Gö6976 and HBDDE (38) to prevent polyFn assembly ruled out participation of PKC␣ in polyFn assembly (Fig. 6A). In contrast, inhibitors that in addi-tion to cPKCs also inhibited nPKCs, including BIM I, Gö6983, BIM XI (Ro-32-0432), and rottlerin (39 -43), decreased polyFn assembly in dose-dependent manners (Fig. 6B). Albeit these inhibitor data do not provide conclusive evidence of the nPKC isoform involved in polyFn assembly, PKC␦ (preferentially, but not specifically, inhibited by rottlerin; Ref. 43) and PKC⑀ (preferentially inhibited by BIM XI) are our most likely candidates for involvement in polyFn assembly based on high protein expression levels and inhibitor activity.
To examine the roles of PKC␦, PKC⑀, and PKC (nPKC isoforms previously associated with breast cancer metastasis, see Refs. 44 -46) in polyFn assembly, we transfected MTF7L cells with the RDs of these isoforms or knocked down their expression by PKC isoform-specific siRNA oligonucleotides. The RDs of PKC␦ (co-transfected with GFP at a ratio of 1:50), PKC-GFP, and PKC⑀-GFP were transiently expressed in MTF7L breast cancer cells. Forty-eight hours after transfection, cancer cells were subjected to EoE suspension culture in serum-containing medium for 1 h, then stained with anti-Fn antibodies, and Fn was visualized by PE-conjugated secondary antibodies. Cells (2 ϫ 10 5 cells/0.3 ml DMEM/rat) derived from MTF7L clone mcl us (control clone), clone siFn-cl 1 (see Fig. 3), and MTF7L clone siPKC⑀-cl 1 (see Fig. 8) were injected into the lateral tail vein of female Fischer 344 rats as described under "Experimental Procedures." Numbers of lung colonies were significantly decreased in rats injected with MTF7L-siFn-cl 1 (*) and, most prominently, with MTF7L-siPKC⑀-cl 1 (*) relative to rats injected with MTF7L-mcl us cells. Colony diameters were reduced by ϳ5-fold in MTF7L-siFn-cl 1 (*) and 2-fold in MTF7L-siPKC⑀-cl 1 relative to MTF7-mcl us , in which the coalescence of individual colonies to large aggregates might have led to an inflated estimate of the colony diameter (results presented in this figure may reflect to some extent a reduced in vitro growth rate of 29% for MTF7L-siFn-cl 1 and 12% for MTF7L-siPKC⑀-cl 1 cells relative to MTF7-mcl us observed after a 4-day culture period in vitro) *, Student's t test, p Ͻ 0.01.

FIGURE 5. Inhibitors of PLC and PKC diminish polyFn assembly. A, MTF7L
breast cancer cells were incubated in the presence of U73122 (PLC inhibitor) and calphostin C (inhibitor of cPKCs and nPKCs) at the indicated concentrations or inhibitor solute (Me 2 SO, control) for 30 min, then subjected to EoE suspension culture for 1 h in DMEM ϩ 20% FBS ϩ inhibitor (or inhibitor solute (S). Cells were then stained for Fn surface expression as described in Fig. 1A and analyzed by FACS (negative control, cells stained with non-immune rabbit IgG). Strong, dose-dependent inhibition of polyFn assembly was observed with both U73122 and calphostin C. FL2-H, FACS phycoerythrin channel peak emission value. B, expression of c-and nPKC isoforms in MTF7L cells; MTF7L breast cancer cells were subjected to EoE suspension culture for 1 h in DMEM ϩ 20% FBS. One-half of the cells were extracted with Trizol and RNA-purified as described by the manufacturer (Invitrogen), and the other half was extracted with lysis buffer. One microgram of purified RNA was subjected to reverse transcription-PCR (35 cycles) using PKC isoform-specific primers (upper panel). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as reference message. Controls were samples run in the absence of reverse transcriptase (RTase). Cell lysates were subjected to SDS-PAGE and Western blotting with isoform-specific anti-PKC antibodies (lower panel). Strong message levels were recorded for PKC␣, -␦, and -⑀, moderate for PKC, and weak for PKC. Protein expression was strong for PKC␣, -␦, and -⑀, weak for PKC, and non-detectable for PKC. Cells were analyzed by bivariate, dual-color (red-green) FACS. Among the three nPKC isoforms, only RD-PKC⑀ inhibited polyFn assembly, as depicted by a significant downshift in the number of GFP-positive cells that strongly stained with anti-Fn antibody, i.e. only 21% of tumor cells that expressed high levels of GFP also expressed high levels of polyFn, relative to 56% in vector-transfected cells (Fig. 7A). The RD sequences of PKC had no effect on polyFn assembly, whereas the RD sequence of PKC␦ caused an increase in the number of strongly GFP-positive cells that expressed high levels of polyFn (Fig. 7A), suggesting that PKC␦ might exert a negative regulatory role in polyFn assembly.
A similar effect on polyFn assembly to that of RD-PKC␦ and RD-PKC⑀ was observed when MTF7L cells were co-transfected transiently with GFP and siRNA-PKC␦ or siRNA-PKC⑀ and subjected to bivariate FACS analyses 48 or 72 h after transfection. Again, knockdown of PKC⑀ suppressed polyFn assembly, whereas knockdown of PKC␦ resulted in increased polyFn assembly on MTF7L cell surfaces (Fig. 7B). These results were confirmed by the analysis of stable MTF7L clones with "siRNA knockdown" of PKC⑀. PKC⑀ knockdown resulted in a significant decrease in polyFn assembly relative to a control clone transfected with an unspecific siRNA nucleotide sequence (Fig.   8A). As expected, the FACS-measured decrease in polyFn paralleled decreased PKC⑀ protein expression (Fig. 8B) as well as decreased amounts of polyFn residing on top of the stack of polyacrylamide gels (Fig. 8C, top), generated from anti-Fn[pan] immunoprecipitates of metabolically labeled siRNA-PKC⑀ and control clones. The biochemically identified decrease in polyFn was reflected in decreased surface association of polyFn on MTF7L cells stained with anti-Fn[pan] (Fig. 8C,  bottom). Functionally, impaired polyFn assembly in siRNA-PKC⑀ clones is associated with a dramatic decrease in lung colonization (Fig. 4).
Plasma Membrane Translocation and Activation of PKC⑀ Promote and PMA Suppresses PolyFn Assembly-The process of polyFn assembly was associated with a time-dependent translocation of PKC⑀ to the plasma membrane. In adherent MTF7L cells, PKC⑀ was present predominantly in the cytosolic fraction (Fig. 9A, time 0) but translocated in increasing amounts to the plasma membrane when MTF7L cells were subjected to EoE suspension culture in the presence of pFn-containing serum (Fig. 9A, time 15-60 min). Peak values for membrane-associated PKC⑀ were observed at 30 min of incubation in EoE suspension culture and remained high for the remainder of the 1-h incubation period. Translocation of PKC⑀ to the plasma membrane was associated with a noticeable gain of its molecular weight, presumably caused by the observed serine and threonine phosphorylation of PKC⑀. PKC⑀ phosphorylation, again, was most prominent at 30 min of EoE suspension culture, then rapidly declined to barely detectable levels at 60 min (Fig. 9B). This decline was accompanied with a backward shift to the lower molecular weight form of PKC⑀ (Fig. 9A). The PKC inhibitor BIM XI totally blocked PKC⑀ phosphorylation (Fig. 9B). This result was confirmed when 32 P-labeled MTF7L were examined for PKC⑀ phosphorylation (Fig. 9C). Control experiments conducted with PKC␦ yielded no phosphorylation of this PKC isoform (Fig. 9D). Interestingly, PMA (1 M) caused a time-dependent loss of PKC⑀ protein from both the cytosolic and membrane fractions of MTF7L cells (Fig. 10A). The PMAmediated loss of PKC⑀ protein and to a slightly lesser extent of PKC␦ protein was most striking after an 18-h incubation period of tumor cells with PMA. No effect was observed for the PMAinsensitive atypical PKC isoform PKC (Fig. 10B). Concomitantly, loss of PKC⑀ protein and decreased plasma membrane translocation in PMA-treated cells resulted in a dose-dependent, diminished polyFn assembly (Fig. 10C).

DISCUSSION
The role of Fn in cancer progression is controversial. Early studies suggest that Fn may act as a tumor suppressor gene, promoting differentiation and suppressing proliferation, migration, invasion, and metastasis (for review, see Refs. [13][14][15][16]. In accordance, expression of pFn in Fn-negative LMM3 murine mammary cancer cells reduced the rate of migration as well as spontaneous and experimental metastasis (47) and forced expression of the ␣5␤1 integrin, which was thought to capture secreted cFn at the cell surface, convert it into fibrils, and then deposit it in the extracellular matrix, reduced motility, and tumorigenicity of transfected tumor cells (48). These findings are contrasted by cDNA or oligonucleotide microarray FIGURE 6. Effect of various PKC inhibitors on polyFn assembly. A, inhibitors of cPKCs such as Gö 6976 and HBDDE have no effect on polyFn assembly. FL2-H, FACS phycoerythrin channel peak emission value. B, other PKC inhibitors such as BIM I, BIM XI, Gö 6983, and Rottlerin, however, inhibit polyFn assembly in a dose-dependent manner. Experiments were conducted as described in Fig. 5A. Significant differences in polyFn assembly by BIM XI at 100 nM and 500 nM and rottlerin at 3 M and, more notably, at 100 M provide strong clues of a possible involvement of PKC⑀ or PKC␦ given the fact that these isoforms are also strongly expressed in suspended MTF7L cells. S, drug solvent.
analyses that attempted to identify genes whose expression was associated with malignant behavior of breast cancers. In several of these studies, Fn (Fn1, EDAϩEDBϩ IIICS 120 ) was found to be consistently overexpressed in invasive/ metastatic breast tumors in toto as well as in tumor cells isolated by laser capture microdissection from a primary invasive ductal carcinoma and an auxiliary node harboring metastatic breast cancer (5-7). Interestingly, ErbB2 overexpression, observed in 25-30% of invasive ductal breast carcinomas and correlated with a poor clinical prognosis, was associated with Fn1 overexpression in 11 of 36 cases (49). Overexpression of Fn was also noticed in cancer cell lines that exhibited an invasive/metastatic phenotype, including MDA-MB-231, MeWo-70W, PC3-ML, 4T1, LLC1, B16-F10, K7M2, MTF7, and RPC-2 (17) and was most dramatic in cancer cell lines that were subjected to a selection process for enhanced lung metastatic performance (9,11,17) or in highly metastatic clones derived from a rabdomyosarcoma (8) or B16 melanoma (10) but not in non-metastatic clones (8). Moreover, expression of the metastasis suppressor gene nm23 in MDA-MB-435 breast cancer cells resulted in a down-regulation of genes associated with adhesion and motility, including Fn1 (50). On the other hand, neither the poor prognosis signature obtained by large-scale transcript profiling of human breast cancers (51,52) nor the gene expression profile that was associated with lung metastasis of selected MDA-MB-231 clones (i.e. lung-metastatic signature) contained Fn (53). A possible explanation for this result may be related to the multifunctional role of Fn in tumor progression, requiring high levels of Fn expression throughout tumor progression. For example, in the early stages of progression Fn promotes survival, cell cycle progression, and angiogenesis, and in later stages Fn promotes motility/invasiveness and dissemination and implantation in sec- ondary organs (for review, see Refs. [12][13][14][15][16]. Support for this notion can be found in the observation that tumor cell lines that do not express Fn such as MCF-7 cells exhibit poor tumorigenicity, growth, and angiogenesis (54), whereas overexpression of c-Jun in MCF-7 cells stimulated Fn synthesis and, concomitantly, increased tumorigenicity and invasiveness (55,56).
In recent years our laboratory has been interested in the role of Fn in breast cancer lung metastasis. This interest was triggered by the fact that a survey of all our highly lung-metastatic breast cancer cell lines, including MTF7, MTF7L, MDA-MD-231, and 4T1 cells, revealed high levels of Fn expression (17) and by the observation that tumor cells isolated from the blood of tumor-bearing animals exhibited a characteristic cell surface coat of multiglobular polyFn. This phenotype could be reproduced in vitro, when cancer cells were subjected to EoE suspension culture in the presence of serum (9,17). As indicated in the introduction, this unique surface expression of Fn was responsible for the docking of MTF7 to the lung endothelial address in DPP IV (9, 17-21) and for facilitating lung colonization (9,17,20,21). Interference with the polyFn/DPP IV adhesion by monoclonal antibodies (9), peptides directed against the DPP IV binding domains in the 13th, 14th, and 15th type III repeats of Fn (17), and a polypeptide representing the extracellular domain of DPP IV (20) abrogated metastatic colonization of the lungs. Here, we have presented some of the molecular underpinnings that regulate polyFn assembly in a rat model of hematogenous dissemination of breast cancer cells. We show that polyFn assembly critically depends upon the expression of endogenous cFn (Fn1) and is regulated by PKC⑀. This is evidenced by a significantly decreased polyFn assembly in cells transfected with either siRNA-PKC⑀ or dominant-negative RD-PKC⑀ and in cells treated with select PKC inhibitors. During a 1-h polyFn assembly phase, PKC⑀ translocates in increas- FIGURE 8. Analysis of stable clones selected from siRNA-PKC⑀-transfected MTF7L cells. siRNA-PKC⑀ clones (cl 1 and cl 2 ) were hygromycin-selected from MTF7L cells transfected with nt sequence 3, siRNA-PKC⑀-cl 3 from MTF7L cells transfected with the nt sequence 4, and control clone mcl us from MTF7L cells transfected with the unspecific nt sequence 5 (see "Experimental Procedures"). A, clones PKC⑀-cl 1 and -cl 2 and control clone mcl us subjected to EoE suspension culture in DMEM ϩ 20% FBS for 1 h were stained with anti-Fn[pan] and analyzed by FACS as described under "Experimental Procedures," then analyzed by FACS. PolyFn assembly by MTF7L clones PKC⑀-cl 1 and PKC⑀-cl 2 was significantly decreased relative to the mcl us control clone. B, Western blot analysis for PKC⑀ and vinculin expression in lysates from siPKC⑀-cl 1 , -cl 2 , and -cl 3 and the mcl us clone; PKC⑀ protein is decreased in siPKC⑀ clones relative to the unspecific siRNA-mcl us control clone. C, lysates from MTF7L clones mcl us and siPKC⑀-cl 1 labeled with [ 35 S]methionine and processed as described in Fig. 3B were subjected to anti-Fn[pan] immunoprecipitation followed by SDS-PAGE and autoradiography. There is a significant decreased in polyFn residing on top of the stacking gel in siPKC⑀-cl 1 relative to the mcl us control, mimicking the degree of tumor cell surface-associated polyFn of these clones stained with anti-Fn[pan] (C, bottom). ing amounts from the cytosol to the plasma membrane. This translocation is accompanied by PKC⑀ Ser/Thr phosphorylation and is crucial for successful polyFn assembly, since overexpression of dominant-negative RD-PKC⑀, which blocks diacylglycerol-mediated plasma membrane-docking and activation of endogenous PKC⑀ (57,58), prevents polyFn assembly. In contrast, both RD-PKC␦ and siRNA knockdown of PKC␦ promoted polyFn assembly, mimicking the previously recognized opposing effects of PKC⑀ and PKC␦ in ischemic injury of the heart (59) and suggesting a dominant-negative regulatory effect of PKC␦ over PKC⑀ in MTF7L cells. A PKC⑀ regulatory effect was also observed for PMA, which in a dose-and timedependent manner inhibited PKC⑀ protein synthesis and, in accordance, impaired polyFn assembly on MTF7L cells in a manner similar to that reported for other cell types (60,61). Together, PMA/diacylglycerol competition for plasma membrane-docking, overall loss of PKC⑀ protein, and competition between related PKC isoforms appears to be primarily responsible for the observed decrease in polyFn assembly on cancer cell surfaces.
The role of PKC⑀ in polyFn assembly appears to be dualistic in that it may facilitate surface expression of endogenous, cellular Fn by promoting Fn exocytosis (62)(63)(64)(65)(66) and protein synthesis (67). In accordance, inhibitors of protein synthesis (e.g. cycloheximide) and exocytosis (e.g. monensin and brefeldin B) strongly impeded polyFn assembly (Fig. 2), albeit cells were kept in suspension culture in the presence of high concentrations of pFn. However, at this writing we have not identified any of the PKC⑀ partner proteins that are required in regulating these processes. Nonetheless, inhibition of PKC⑀ has been shown to trap integrin ␤1, a possible receptor for cellular Fn, in a CD81-positive intracellular compartment, and electron microscopy demonstrated the co-localization of PKC⑀ and integrin ␤1 on vesicular membranes (68). Translocation of the integrin ␤1 from cytosolic vesicles to the plasma membrane appears to involve the PKC⑀-mediated phosphorylation of the cytoplasmic tail of ␤1-integrin at threonine 788/789 (69) or cytoskeletal intermediaries such as vimentin (70) and actin (62). Accordingly, interaction of PKC⑀ with actin has been linked to the formation of invadopodia-like structures, increased pericellular metalloproteinase activity, and ultimately, invasiveness and metastasis in PKC⑀-transformed NIH3T3 fibroblasts (71). Whether the concomitant down-regulation of RhoC in MDA-MB-231 cells, in which PKC⑀ expression was knocked down by RNA interference resulting in a cell phenotype that was significantly less proliferative, invasive, and metastatic (45), is intertwined in PKC⑀-actin, remodeling is unclear and needs further investigation. Supporting evidence for such a role may be deduced from the thrombin-induced PKC⑀-Rho-actin complex formation in actin reorganization in myofibroblasts (72). However, in our system of suspended (blood-borne) breast cancer cells, the Rho activity is significantly suppressed, and expression of constitutively active Rho suppresses polyFn assembly. 4 In conclusion, our studies suggest that surface expression of polyFn occurs in a PKC⑀-dependent manner presumably involving the transport of Fn-receptor complexes from cytosolic vesicles to the plasma membrane and that such complexes serve as scaffolds for the subsequent augmentation of polyFn aggregates on tumor cells during their journey in Fn-rich blood plasma. This notion is supported by a dramatic inhibition of polyFn assembly by inhibitors of exocytosis and by siRNA knockdown of PKC⑀, which occurs even when cancer cells are exposed to high exogenous Fn concentrations and express several integrin and syndecan Fn receptors on their surfaces. 3 Thus, we have disclosed here a novel regulatory principle for the recognized pro-metastatic role of Fn that may spur the design of new anti-metastatic therapies.