Tumor metastasis is a critical event for cancer patients because it often results in death. During the progression of tumor cells to a metastatic phenotype, they undergo a series of changes that begins with loss of contact inhibition and increased cell motility, both of which allow the primary tumor cells to invade distant organs and induce neovascularization and thereby result in metastasis.1 Many of these changes involve the integrated action of numerous gene products, most of which culminate in the modulation of the actin cytoskeleton that is crucial for cell morphogenesis, motility, adhesion and cytokinesis.2, 3, 4
LIM kinase 1 (LIMK1) is a member of a novel class of serine–threonine protein kinases that contain 2 LIM motifs at the amino terminus and an unusual protein kinase domain at the carboxyl terminus.5 The LIMK1 gene is expressed predominantly in brain and developing neural tissues.6, 7 The potent actin-binding protein, cofilin, which regulates actin dynamics by depolymerizing F-actin, is the only known physiologic substrate of LIMK1.8, 9, 10 LIMK1 phosphorylates cofilin and inhibits its ability to bind and depolymerize actin, leading to accumulation of F-actin aggregates and filaments.11, 12 The level and activity of endogenous LIMK1 are higher in invasive breast and prostate cancer cell lines than in less invasive cells.13, 14
In addition to cytoskeleton reorganization, tumor progression requires degradation of the extracellular matrix by the serine protease urokinase-type plasminogen activator (uPA). uPA converts inactive plasminogen into plasmin and is therefore a kingpin in the initiation of a cascade of proteolytic steps that ends with the degradation of the extracellular matrix.15 uPA is found in cellular structures at the leading edge of migrating cells that are involved in adhesion, migration, invasion and intravasation.15 uPA receptor (uPAR) binds uPA and facilitates a proteolytic cascade at the cell surface. uPAR is a multifunctional protein that, through its interaction with integrins, initiates signaling events that alter cell adhesion, migration and proliferation.16 Numerous reports have documented the involvement of uPA/uPAR in cancer (for review, see ref.17). The use of antagonists of uPA and uPAR have been shown to prevent the growth, invasiveness and metastasis of tumors.18, 19 Further, increased levels of uPA/uPAR are strongly associated with poor prognosis and unfavorable clinical outcome.20, 21 The uPA system also has been implicated in the pathologic angiogenesis of a variety of tumors, including breast tumors.22 Although an important role of LIMK1 signaling in cell motility has been suggested, the signaling pathways induced by LIMK1 in metastatic process are unknown.
Here we report that LIMK1 overexpression in human breast cancer cells leads to enhanced tumor angiogenesis and metastasis. We also provide new evidence of upregulation of uPA and uPAR protein expression by LIMK1. Our results demonstrate an important role for LIMK1 signaling in breast cancer cell invasion via regulation of the uPA system.
Material and methods
Cell culture and transfections
MDA-MB-435 human breast cancer cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum (FCS). DNA constructs encoding mouse wild-type LIMK1 (WT-LIMK1) and the catalytically inactive mouse dominant-negative LIMK1 (D460N-LIMK1) are described elsewhere23 and were generously provided by Gordon Gill (Department of Endocrinology, University of California, San Diego). We transfected the MDA-MB-435 cells with the LIMK1 constructs or empty vector (pcDNA) as a control by using FUGENE reagent (Invitrogen, Carlsbad, CA). Stably transfected cell lines were selected in the presence of 500 μg/ml of the antibiotic G418 for 4 weeks.
Monoclonal anti-human uPA B-chain antibody and polyclonal rabbit anti-human uPAR immunoglobulin G were purchased from American diagnostica (Greenwich, CT). Polyclonal rabbit anti-LIMK1 antibody (C-19) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse anti-CD34 antibody was purchased from NeoMarkers (Fremont, CA).
Immunoprecipitation and kinase assay
Transfected cells were lysed in a buffer containing 20 mM Tris-HCl (pH 7.5), 10% glycerol, 1% Nonidet P-40, 10 mM NaF, 1 mM NaVO4 and protease inhibitor cocktail. LIMK1 was immunoprecipitated from the cell lysates with anti-LIMK1 antibody. Immunoprecipitated beads were washed 3 times with kinase buffer (20 mM HEPES, 10 mM MgCl2, 10 mM MnCl2, 1 mM dithiothreitol, 0.2 mM EGTA, 20 mM ATP and 10 mCi (γ- 32P)ATP), and the kinase reaction was done using cofilin as a substrate at 30°C for 40 min.
To detect LIMK1 in transfected cells, we lysed them in RIPA buffer, separated the proteins in 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel and transferred the proteins to nitrocellulose membrane. The blots were probed with anti-LIMK1 antibody at a dilution of 1:3,000 (provided by Gordon Gill) as previously described,24 and then incubated with horseradish peroxidase-conjugated anti-rabbit antibody at a dilution of 1:2,000 (Amersham, Piscataway, NJ).
Cell proliferation assays
To determine cell proliferation levels, untransfected cells, which MDA-MB-435 cells and MDA-MB-435 overexpress LIMK1 were plated (2 × 104 cells) onto 24-well plates containing DMEM and 10% FCS and counted after 1, 3, 5 and 7 days of culture with a Coulter counter. For the MTT assay, cells were plated (5 × 103) onto 96-well plates containing DMEM and 10% FCS, and cell proliferation was determined after 3 days of culture using the 3-(4,5-dimethylthiazol-2-yl)-diphenyltetrazolium bromide dye method as previously described.25
Cell invasion assay
To assess the invasiveness of transfected cells, we used the Matrigel invasion assay. The upper surfaces of 8-μm filters were coated with 20 μg of Matrigel and dried overnight. The inserts were rehydrated for 1.5–2.0 hr with 0.5 ml of warm DMEM containing 0.1% bovine serum albumin. Cells (1 × 105) suspended in DMEM containing 0.1% bovine serum albumin plus 1.5 μg/ml anti-uPA antibody, anti-uPAR antibody or control IgG were placed in the upper compartment of a 24-well Boyden chamber. Conditioned medium from mouse fibroblast NIH-3T3 cells (ATCC) was used as a source of chemoattractant and placed in the lower compartment of the Boyden chamber. After 24 hr of incubation, nonmigrating cells were scraped off, and invaded cells were fixed with methanol and stained with hematoxylin and eosin. Migrated cells on the bottom side of the filters were counted on 6 randomly selected microscopic fields. The experiments were carried out in triplicate and repeated at least 3 times.
The uPA–CAT construct used here was described previously.26 Briefly, the construct included 2,345 bp of the 5′-flanking region of the uPA promoter fused with the CAT reporter. WT-LIMK1 and control (empty vector) transfectants were cultured in DMEM containing 0.1% serum for 24 hr. The serum-starved cells were then transiently transfected with both the uPA–CAT construct and the control pSV-b-Gal vector (Promega, Madison, WI) by using Lipofectamine (Life Technologies, Inc., Rockville, MD). CAT activity was measured at 24 hr after cotransfection. Each experiment was repeated with 3 independent transfections; transfection efficiency among different experiments varied between 40% and 50%.
To determine uPA and uPAR mRNA expression levels, total cytoplasmic RNA was isolated from WT-LIMK1, D460N-LIMK1 and control transfectants by using Trizol reagent (Life Technologies). Ten micrograms of RNA was analyzed in formaldehyde-agarose gel. 32P-labeled cDNA probes for uPA and uPAR (American Type Culture Collection, Manassas, VA) were used for hybridization.
We determined uPA activity by gel zymography. WT-LIMK1, D460N-LIMK1 and control transfectants were cultured until 60% confluency, washed with phosphate-buffered saline (PBS) and incubated with FCS-free DMEM medium. The conditioned medium was collected after 36 hr. Equal amounts of uPA (25 μg) were mixed with SDS gel-loading buffer and then loaded without reduction or heating onto 10% SDS-polyacrylamide gels containing plasminogen (20 μg/ml) as a substrate for uPA and nonfat dry milk (50 mg/gel) as a substrate for plasmin. After electrophoresis, the gels were washed with 2.5% (v/v) Triton-X 100 for 45 min to remove the SDS. Finally, we placed the gels in 0.1 M glycine buffer (pH 8.0) and incubated them overnight at 37°C to allow proteolysis of the substrates in the gels. The gels were stained with 0.1% (w/v) Coomassie Brilliant Blue-R250 [25% (v/v) methanol, 10% (v/v) acetic acid in water] and then destained [25% (v/v) methanol plus 10% (v/v) acetic acid in water without Coomassie blue]. Upon staining with Coomassie blue and destaining, the final gel had a uniform background except in regions to which uPA and plasmin had migrated and cleaved the substrate.
To determine whether transfected cells could proliferate in vivo as well as in vitro, exponentially growing transfectants (3 × 106 cells) were injected into the number 4 inguinal mouse mammary fat pads of 4–6-week-old athymic female nude mice at both sides. (National Cancer Institute, Frederick, MD). Tumors were measured every week with calipers along 2 major axes, and tumor volume was calculated as 4/3πR12R2, where R1 is radius 1, R2 is radius 2 and R1 < R2. Primary tumors and selected organs were excised 10 weeks after the injections, and the numbers and locations of metastases were recorded. The organs were fixed with 10% phosphate-buffered formalin and embedded in paraffin. Paraffin sections were cut, stained with hematoxylin and eosin and observed microscopically for histopathologic characteristics. This experiment was repeated twice. Eight animals were used in each group WT-LIMK1 (WT-LIMK1 #1 and WT-LIMK1 #4), and control (empty vector) for a total of 24 mice and 48 tumors. In the second experiments, 8 animals were used in each group WT-LIMK1#1 and control for a total of 16 mice and 32 tumors.
Immunohistochemical and TUNEL staining of tumor tissue.
Tumor tissue sections were deparaffinized and rehydrated. Endogenous peroxidase was inactivated with 3% H2O2, washed in PBS, and preincubated in goat serum for 1 hr at room temperature. The sections were then incubated for 1 hr at room temperature with a 1:50 dilution of anti-CD34 antibody to detect vascular endothelial cells. Biotinylated rabbit anti-mouse immunoglobulin (Dako, Carpinteria, CA) diluted 1:100 in 3% normal goat serum was applied for 1 hr, and followed up with peroxidase-conjugated horseradish preformed streptavidin biotin complex for 10 min (Dako, Carpinteria, CA). Reaction sites were visualized using diaminobenzidine as the chromogen, and nuclei were counterstained with hematoxylin. Ten high-power fields (400×) from the tumor region were arbitrarily selected, and the number of stained blood vessels was counted in 10 random fields. For uPA staining, the sections were incubated for 1 hr at 37oC with a 1:25 dilution of anti-uPA antibody. For uPAR staining, the sections were boiled in 0.1 M citrate buffer and incubated for 1hr at room temperature with a 1:25 dilution of anti-uPAR followed by the protocol described above. We used the TUNEL method to detect DNA fragmentation, as previously described.27 Briefly, paraffin-embedded sections pretreated with protease were nick-end labeled with biotinylated poly(dU), using terminal deoxytransferase and stained using avidin-conjugated peroxidase.
Tube formation assay
A gel (Matrigel; 10 mg/ml) composed of reconstituted basement membrane proteins, to which endothelial cells attach rapidly was coated onto 24-well plates at 4oC. HUVEC endothelial cells (American Type Culture Colection) were seeded onto the Matrigel layers (5 × 104 cells/well) and treated with conditioned medium collected from transfected MDA-MB-435 cells and from WT-LIMK1 transfectants.
The differences between the groups were analyzed first by one-way analysis of variance. If the difference was significant (p < 0.05), a Dunnett's post hoc test was performed to determine the significance of the difference between the groups and control group. When comparing multiple groups, simultaneously a Student's t test was performed without correction. All analyses were performed with the SPSS (version 11.5) statistical package.
LIMK1 overexpression increases cell proliferation in vitro
We investigated whether changes in LIMK1 expression in MDA-MB-435 cells affected their proliferation. We generated clones of MDA-MB-435 cells stably expressing WT-LIMK1 and determined the level of LIMK1 expression in the transfected cells by immunoblotting (Fig. 1a). To investigate whether changes in the level of LIMK1 protein also leads to changes in its activity, we determined the level of phosphorylation of the endogenous substrate, cofilin, by in vitro kinase assay. The stable transfectants expressing WT-LIMK1 showed a 2- to 3-fold higher level of phosphorylated cofilin than control (empty vector) transfectants did (Fig. 1a). Using 2 different cell proliferation assays, we compared the growth rate of WT-LIMK1 transfectants with that of control transfectants. Quantification of these results revealed that LIMK1 overexpression in breast cancer cells increased the growth rate by 50–83% (Fig. 1b and 1c).
LIMK1 increases tumor growth and induces lung and liver metastasis in vivo
To determine whether the data from the cell proliferation assays reflected the ability of the transfected cells to proliferate in vivo, we tested the ability of breast cancer cells to form tumors and to metastasize in nude mice. The mean tumor volume was 1.7- to 2.7-fold higher in WT-LIMK1 tumors than in control (empty vector) tumors (p < 0.001) (Fig. 2a and 2b). Histopathologic examination of diverse organs stained with hematoxylin and eosin revealed the presence of multiple micrometastases in the livers and lungs from 3 of 8 (38 %) mice in the WT-LIMK1 group for both experiments (Fig. 2c). In contrast, no metastasis was observed in any of the 8 mice in the control group in 2 independent experiments. Western blot analysis of the tumor samples revealed higher expression of uPAR in WT-LIMK1 tumors than in control tumors (Fig. 2d). To investigate the mechanism of the observed invasive properties of WT-LIMK1 tumors, we examined expression levels of uPA and uPAR, known prognosis markers of cancer metastasis, by immunostaining. The results revealed higher levels of uPA (Fig. 2e) and uPAR (Fig. 2f) expression in tumor cells of WT-LIMK1 tumors than those of control tumors.
LIMK1 regulates the uPA system in breast cancer cells
To explore the role of LIMK1 in the regulation of uPA, WT-LIMK1 and control transfectants were transfected with the uPA–CAT construct. We found that WT-LIMK1 transfectant were able to induce 8 times more uPA than transfectants control were (Fig. 3a). We next generated stable clones expressing kinase-inactive mutant D460N-LIMK1 protein. The level of D460-LIMK1 protein was determined by immunoblotting (Fig. 3b). Using a cDNA probe, we observed that uPA and uPAR mRNA expression levels were 5–7 times and 3–5 times higher, respectively, in WT-LIMK1 transfectants than in control transfectants (Fig. 3c). We also observed a marked decrease in the basal levels of uPA and uPAR mRNA expression in D460N-LIMK1 transfectant (Fig. 3c). In addition, western blot analysis showed that uPAR protein expression was upregulated in WT-LIMK1 cells (Fig. 3d). Using gel zymography, we next determined whether LIMK1 induces uPA activity. Compared with control transfectants, WT-LIMK1 transfectants produced a large amount of uPA, whereas D460N-LIMK1 transfectants secreted only a slight amount of uPA (Fig. 3e).
To determine whether LIMK1 overexpression affects cell invasiveness, we performed a Matrigel invasion assay. Compared with control transfectants, WT-LIMK1 transfectants were 2–3 times more motile, whereas D460N-LIMK1 transfectants exhibited lower invasiveness. The results indicated that endogenous LIMK1 activity is required for the invasiveness of MDA-MB-435 cells (Fig. 3f). The Matrigel invasion assay also revealed that the anti-uPA and anti-uPAR antibodies completely blocked the invasiveness of both WT-LIMK1 transfectants and D460N-LIMK1 transfectants (Fig. 3g). The IgG control did not affect the invasiveness of MDA-MB435 cells (data not shown). Taken together, these observations suggested that uPA and uPAR contribute substantially toward the LIMK1-induced invasive pathways.
LIMK1 promotes in vitro and in vivo angiogenesis
The animal studies showed that LIMK1 overexpression in breast cancer cells resulted in increased tumor growth and metastasis and corresponded with high expression levels of uPA and uPAR. It has been reported that uPA promotes the progression of breast cancer by enhancing angiogenesis and tumor cell invasion, and the relationship between tumor metastasis and tumor angiogenesis has been shown. To determine whether the highly metastatic behavior of WT-LIMK1-transfected cells could be related to the increased angiogenic property of these cells, we tested the ability of conditioned medium collected from untransfected MDA-MB-435 cells and from WT-LIMK1 clones to stimulate tube formation of endothelial cells. Because basement membrane can stimulate differentiation, HUVEC cells were plated onto Matrigel-coated plates. After 24 hr, elongated processes appeared in the HUVEC cells that had been cultured in the presence of conditioned medium collected from untransfected MDA-MB-435 cells (Fig. 4a). In contrast, the conditioned medium collected from WT-LIMK1 clones promoted the formation of networks of branching and anastomosing cords of cells (commonly known as the tube formations) (Fig. 4b). WT-LIMK1 tumors appeared red compared with the pale untransfected tumors (Fig. 4c and 4d). Hematoxylin and eosin staining of the tumors revealed that WT-LIMK1 tumors (Fig. 4f) were more angiogenic and less necrotic than untransfected tumors were (Fig. 4e). To determine whether LIMK1 can induce tumor angiogenesis, we also stained the prepared tumor sections with an anti-CD34 antibody, which is a marker of vascular endothelial cells. WT-LIMK1 tumors (Fig. 4i and 3g) had 3–4 times more blood vessels than untransfected tumors did (Fig. 4g and 3h). TUNEL assays also revealed that the centers of the tumors (30–50% of the total tumor volume) formed by untransfected cells were necrotic and that the periphery of the tumors contained viable cells (Fig. 4j). In contrast, the tumors formed by WT-LIMK1 cells had only sparse patches of necrosis (5–10% of the total tumor volume) (Fig. 4k).
LIMK1 expression is increased in human breast tumors
Western blot analysis revealed that LIMK1 was overexpressed in human breast tumor tissues (Fig. 5a). In 9 of 12 sample pairs, LIMK1 was expressed at a substantially higher level in the tumor tissues than in the normal tissues (Fig. 5a). Further analysis of the same paired specimens suggested that in 7 of 12 sample pairs, uPAR expression was substantially higher in tumor tissues than in normal tissues (Fig. 5b). This observation suggests that LIMK1 expression and uPAR expression corresponded in human breast tumors.
Evidence for involvement of the uPA system in cancer metastasis has increased greatly, since it was demonstrated that inhibition of uPA and uPAR binding with their respective antibodies or with other specific binding antagonists prevented tumor cell invasion.18, 28 Here we showed that expression of LIMK1, one of the proteins that modulate actin dynamics, in MDA-MB-435 breast cancer cells enhanced cell invasion by upregulating the uPA system, increased uPA promoter activity, induced uPA and uPAR mRNA and protein expression and induced uPA secretion. In contrast, expression of the catalytically inactive LIMK1 mutant D460N-LIMK1 did not have these effects. The expression of D460N-LIMK1 may decrease cancer cell motility and the levels of uPA, and uPAR, and serum-soluble uPAR protein expression because of competion with endogenous LIMK1 protein, which is required for the invasiveness of MDA-MB 435 cells. Taken together, our data provide evidence that upregulation of the uPA system contributes to the invasive function of LIMK1 in MDA-MB-435 breast cancer cells and suggest a signaling pathway connecting LIMK1 and the uPA system to actin reorganization and increased cell invasiveness. It has been previously reported that the effect of LIMK1 in facilitating cell invasion is not mediated through the inactivating phophorylation of cofilin.13 This observation suggests an unknown mechanism involved in the induction of cell motility by LIMK1.
It is known that the uPAR–vitronectin interaction induces cytoskeletal rearrangement and increased cell motility via activation of Rac signaling pathway.29 We speculate that LIMK1-induced cell motility and invasion function is mediated by uPAR regulation.
We also showed that LIMK1 overexpression increased cell proliferation and tumor growth in vitro. A role for LIMK1 in the regulation of mitosis has been suggested by others on the basis of that showed LIMK1 activity fluctuated with cell cycle progression and attained a maximum level during mitosis when LIMK1 became hyperphosphorylated, presumably by mitotic Cdks.30, 31 Antisense RNA-mediated reduced expression of LIMK1 provided evidence of the involvement of LIMK1 in the regulation of the growth of prostate cancer cells, especially at the G2/M phase of the cell cycle.13
In addition, our study revealed that WT-LIMK1 tumors exhibited more angiogenesis and a lower death rate than did control tumors, which exhibited large necrotic areas as a direct result of cell death. Immunohistochemical analysis of tumor sections showed a much higher uPA expression level in WT-LIMK1 tumors than in control tumors. uPA, which is expressed only by highly invasive cancer cells, has been implicated in the invasion, metastasis and angiogenesis of several malignancies, including breast cancer.32 We suggest that high uPA expression in WT-LIMK1 tumors induces tumor neovascularization and angioinvasion and causes tumor progression and metastasis.
Finally, we observed that LIMK1 was highly expressed in human breast cancer tumors. The level of endogenous LIMK1 corresponded with uPAR protein expression in these tumors and may also be associated with the invasiveness and therefore metastatic potential of the tumors. uPAR appears to be a prognostic marker in certain cancers. In breast cancer, high uPAR levels are associated with a shorter disease-free interval and a shorter overall survival time.33, 34 Our findings indicate the potential for treating invasive breast cancers with compounds that inhibit LIMK1 activity. Thus, measuring LIMK1 levels could be important to the management of breast cancer and might be used as a reliable marker of the invasiveness of the disease.
This work was supported by National Institutes of Health grants CA098823 and CA080066 to R.K.