Wnt-5a/Ca²⁺-Induced NFAT Activity Is Counteracted by Wnt-5a/Yes-Cdc42-Casein Kinase 1α Signaling in Human Mammary Epithelial Cells

Cell culture.

We used the HB2 mammary epithelial cell line, which is a subclone of the MTSV-1.7 cell line originating from the laboratory of J. Taylor-Papadimitriou (ICRF, United Kingdom). We used the following clones, which had previously been produced in our laboratory (24): Wnt-5a-overexpressing cells stably transfected with a pLNCX Wnt-5a-HA vector (Wnt-5a^high), Wnt-5a-antisense cells obtained by transfecting with a pLNCX vector containing the Wnt-5a cDNA in an antisense direction (3′ to 5′; Wnt-5a^low), and Wnt-5a neo cells with an empty pLNCX vector (HB2 neo). We also used MCF-7 cells and MCF-7 cells stably transfected with the pLNCX Wnt-5a-HA vector (MCF-7 Wnt-5a^high) (24), SYF cells lacking all Src-kinase family members and HEK293 cells (both from the American Type Culture Collection). The cells were cultured in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum, and for HB2 cells 10 μg of bovine insulin and 5 μg of hydrocortisone/ml was added. The cells were maintained at 37°C in a humidified atmosphere with 5% carbon dioxide.

Chemicals and reagents.

Anti-CK1α and anti-NFAT1 antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), and antibodies to phosphorylated JNK, total JNK, phosphorylated GSK-3β, phosphorylated (Ser 144) Pak1, and total Pak1 came from Cell Signaling (Beverly, MA). We purchased anti-total Yes and GSK-3β from BD Transduction Laboratories (Franklin Lakes, NJ), JNK inhibitor II from Calbiochem (La Jolla, CA), Fura-2 AM from Molecular Probes (Eugene, OR), anti-Cdc42 from Signal Transduction (Lexington, KY), and the G410 anti-phosphotyrosine antibody from Upstate Biotech, Inc. (Lake Placid, NY). The polyclonal rabbit Wnt-5a antibody directed against the C terminus of human Wnt-5a was developed in our laboratory (25). The Src-family kinase inhibitor 4-amino-5-(4-methyl-phenyl)-7-(t-butyl) pyrozolo-d-3,4-pyrimidine (PP1) was from Alexis Biochemicals (San Diego, CA), the casein kinase 1 inhibitor IC261 was from Santa Cruz Biotechnology (Santa Cruz, CA), and the PKC inhibitor bisindolylmaleimide-I (GFX), the Src-family kinase specific inhibitor Su6656, ionomycin, cyclosporine (CS), phorbol myristate acetate (PMA), and cytochalasin D were from Sigma-Aldrich (St. Louis, MO). Recombinant mouse Wnt-5a (0.4 to 1.6 μg/ml) and Wnt-3a (100 ng/ml) was from R&D Systems (Rochester, MN). We have experienced variations in efficiency of recombinant Wnt-5a, and therefore the recombinant Wnt-5a concentrations used in the present study vary between 0.4 and 1.6 μg/ml. All other laboratory reagents were obtained from Sigma Chemicals (St. Louis, MO).

Western blotting and immunoprecipitation.

Adherent cells grown to subconfluency on tissue culture-coated plates or cells plated on collagen-coated plates were lysed in hot Laemmli buffer and dithiothreitol (DTT) for whole-cell lysates. For immunoprecipitations, the cells were serum and/or insulin starved overnight; left unstimulated or stimulated as described; and lysed in 1 ml of lysis buffer containing 100 mM Tris-HCl (pH 7.5), 1% Triton X-100, 5 mM EDTA, 5 mM EGTA, 50 mM NaCl, 4 mM Na₃VO₄, 20 μg of aprotinin/ml, 1 μg of leupeptin/ml, 2.5 mM benzamidine, and 2 mM Pefabloc (Roche). The supernatants were collected, and protein concentrations were measured. The lysates were precleared with protein-A Sepharose or protein-G Agarose (CK1α), and immunoprecipitations were performed with preconjugated beads, except for Yes immunoprecipitations that required preincubation with the antibody. The beads were washed five times in ice-cold wash buffer (50 mM HEPES [pH 7.4], 1% Triton X-100, 0.1% sodium dodecyl sulfate [SDS], 150 mM NaCl, 2 mM Na₃VO₄) and were boiled immediately in Laemmli buffer with DTT. The membranes were probed with antibodies to Wnt-5a (1:2,000), phospho-JNK (1:1,000), JNK (1:1,000), Cdc42 (1:1,000), phosphotyrosine (1:5,000), NFAT1 (1:200), Ack1 (1:1,000), phospho-GSK-3β (1:500), GSK-3β (1:1,000), phospho-Pak1 (1:1,000), Pak1 (1:1,000), or DDR1 (1:10,000), CK1α (1:500), and Yes (1:2,000) for 1 h at room temperature. The separated protein bands were visualized by incubating the membranes with horseradish peroxidase-conjugated secondary antibodies and with enhanced chemiluminescence. The chemicals used in the experiments were added at the times and concentrations indicated. For analysis of Wnt-5a protein levels in conditioned media, 15 ml of cell culture media from confluent Wnt-5a^high and Wnt-5a^low cells were centrifuged in Amicon Ultra Centrifugation Filter Devices (30,000 molecular weight cutoff; Millipore, Bedford, MA).

GST pull-down assay.

The cDNA of the Cdc42-binding domain from PAK1B (PAKcrib; amino acids 56 to 267) was cloned into bacterial expression vector pGEX-2T (10) and was expressed in Escherichia coli as a fusion protein with glutathione S-transferase (GST). For GST pull-down assays, the cells were lysed in a buffer composed of 50 mM Tris (pH 7.5), 1% Triton X-100, 100 mM NaCl, 10 mM MgCl₂, 5% glycerol, 2 mM Na₃VO₄, 20 μg of aprotinin/ml, 1 μg of leupeptin/ml, 2.5 mM benzamidine, and 2 mM pefabloc. The lysates were centrifuged for 30 s at 15,000 × g, the Triton X-100-soluble fractions were recovered, and protein concentrations were determined. The bacterial lysate containing the GST-PAK-crib fusion protein was added to the samples together with GST beads. After 1 h, the beads were collected and washed three times in 25 mM Tris (pH 7.5)-1% Triton X-100-100 mM NaCl-30 mM MgCl₂-1 mM DTT-1 mM Na₃VO₄. The beads were then resuspended in Laemmli buffer and boiled for 5 min under reducing conditions. The precipitated proteins were separated by electrophoresis, and the membranes were probed with antibodies against Rac1 or Cdc42. The blots were reprobed with anti-GST antibodies, and blots examining the total Cdc42 or Rac1 levels were run in parallel.

Transient transfections and luciferase assays.

Wnt-5a^high or Wnt-5a^low cells were transfected with oligofectamine reagent (Invitrogen) according to the manufacturer's instructions. Each transfection was performed with 10⁶ cells, 1 μg of reporter gene construct, and 0.2 μg of cytomegalovirus (CMV)-controlled Renilla reporter gene. The reporter genes used were human NFAT-pGL2 (AP1 independent) and a TATA-box cloned into pGL3 as a basic transcriptional background control (TATA-pGL3) or the TOP/FOPflash vectors. In cotransfection experiments, we used 1 μg of reporter gene construct, 1 μg of expression plasmid (dominant-negative Cdc42, Rac1 or wt hLRP6), and 0.2 μg of CMV-controlled Renilla. After 36 h, protein extracts were prepared, and luciferase assays were performed with a dual-luciferase reporter assay system (Promega). The obtained luciferase activity was normalized against the activity of the cotransfected Renilla reporter gene and the TATA background in the two different cell clones. Transfections for the JNK activation experiments were performed for 36 h, using 5 × 10⁶ cells and 3 μg of dominant-negative Cdc42, Rac1, or empty vector. The expression levels were analyzed with anti-Myc antibodies. With the exception of cytochalasin D, treatment with all chemicals used in the transfection experiments (PP1, CS, GFX, BAPTA, PMA, and JNKinh) occurred for 36 h at the concentrations indicated in the figure legends. Cytochalasin D was used as described elsewhere (51). For the invasion assays SYF cells were transfected for 18 h using Oligofectamine (Invitrogen), Optimem (Gibco), and a control vector (empty vector) or the NFAT blocking vector VIVIT-green fluorescent protein (GFP) (NFAT regulatory domain originally produced by A. Rao [2] and purchased from Addgene; plasmid 11106). The cells were subsequently cultured in normal medium for 48 h and then harvested for invasion assays.

Immunofluorescence.

Wnt-5a^low cells were seeded on glass coverslips and grown overnight to allow them to adhere. Thereafter, the Wnt-5a^low cells were treated with recombinant mouse Wnt-5a (0.8 μg/ml; R&D Systems) for 1 h or were left untreated. Thereafter, the cells were washed with phosphate-buffered saline (PBS), fixed in freshly prepared 4% paraformaldehyde for 15 min, permeabilized in PBS-0.5% Triton X-100 for 15 min, and then blocked in 3% bovine serum albumin-PBS for 1 h at room temperature. Staining was performed with a mouse monoclonal anti-NFAT1 antibody (Affinity Bio Reagents) at 1:100 or isotype control antibodies in 1% bovine serum albumin-PBS for 1 h at room temperature, after which the cells were washed six times with PBS and then incubated with secondary anti-mouse Alexa 488 (1:500) antibodies. Next, the coverslips were washed six times in PBS, mounted in fluorescent mounting medium (Dako A/S), and examined and photographed in a Nikon Eclipse 800 microscope using a ×60 objective lens. Images were recorded with a scientific-grade, charge-coupled device camera (Hamamatsu, Japan) and were analyzed with HazeBuster deconvolution software (VayTek, Inc., Fairland, CT).

Calcium imaging.

HB2 or SYF cells were incubated with 4 μM fura-2/AM for 30 min, after which they were washed and placed in a chamber containing a physiologically balanced calcium medium (136 mM NaCl, 4.7 mM KCl, 1.2 mM MgSO₄, 1.1 mM CaCl₂, 1.2 mM KH₂PO₄, 5 mM NaHCO₃, 5.5 mM glucose, 20 mM HEPES [pH 7.4]) at 37°C. The chamber was placed in a Nikon Diaphot microscope equipped with a Photon Technology International imaging system, and the cells were allowed to rest for 10 min. Fluorescent signals of cells were recorded before and after stimulation with Wnt-5a (0.8 μg/ml) or Wnt-3a (100 ng/ml) using excitation wavelengths of 340 and 380 nm and an emission wavelength of 510 nm. At least 40 ratios were recorded per minute. For each experiment, the fluorescence intensity from all cells in the field of vision was monitored, after which the calcium response of at least 10 single cells was calculated as the change in fluorescence intensity ratio (340/380 nm) using PTI image master software. Each single trace shown is a representative of at least three independent experiments. The Wnt-5a-induced calcium response was detected in more than 95% of the cells that were imaged.

Matrigel invasion assays.

Invasion assays were carried out using Matrigel invasion chambers (BD) with 8.0-μm-pore-size pore membranes in 24-well plates. HB2 Wnt-5a^low and SYF cells were harvested, washed, and resuspended at a concentration of 10⁵ cells per ml in serum-free culture media. Serum-containing medium (10% fetal calf serum [FCS]) was added to the lower well, and 0.5 ml (5 × 10⁴ cells) of the cell suspension, containing recombinant Wnt-5a (0.8 μg/ml) where indicated, was added to the invasion chamber, and the cells were allowed to invade for 72 h (HB2 Wnt-5a^low cells) or 36 h (SYF cells). Cells that had invaded through the Matrigel were fixed (4% paraformaldehyde), stained with crystal violet (Sigma Aldrich), and counted. The lower wells were always checked for cells to ensure that the experiments were stopped at the appropriate time for each cell type. SYF-cell transfection procedures were as described above, and the transfection efficiency (ca. 50%) was evaluated by using fluorescence microscopy (GFP).

Wnt-5a induces a Ca²⁺ signal and slight but distinct activation of NFAT.

Normal HB2 mammary epithelial cells express Wnt-5a endogenously. Our present (Fig. 1A) and previous (24) results show that the Wnt-5a protein levels are modest in empty vector-transfected (HB2; neo), low in Wnt-5a antisense-transfected (Wnt-5a^low), and high in Wnt-5a-overexpressing (Wnt-5a^high) clones of HB2 epithelial cells, all three of which contain the Wnt receptors Fz1/2 and Fz5 (data not shown). Also, the produced Wnt-5a protein is properly processed and secreted and can be found in the concentrated conditioned media of Wnt-5a^high cells but not of Wnt-5a^low cells (Fig. 1A right). To ascertain a distinct difference in the level of Wnt-5a protein expression, we have here chosen to compare Wnt-5a^low cells with Wnt-5a^high cells. It has previously been reported that Wnt-5a can activate NFAT through the Wnt/Ca²⁺ noncanonical pathway in Xenopus (53), whereas no significant NFAT activity was detected in mammalian cells (61). To investigate whether Wnt-5a can activate NFAT via a noncanonical pathway in HB2 human breast epithelial cell lines, we compared basal activation levels of NFAT in Wnt-5a^low and Wnt-5a^high HB2 cells. To this end, we transiently transfected Wnt-5a^low and Wnt-5a^high cells with luciferase reporter vectors encoding NFAT sites, and the results indicate a moderate, although statistically significant, increase in basal activation levels of NFAT in the latter cells (Fig. 1B, left; P < 0.05). The luciferase reporter vector used had no AP1 sites and can be activated by NFAT alone. The differences between the cell lines were not due to clonal variations, since stably transfected MCF-7 Wnt-5a^high cells also showed an increased NFAT activity compared to normal MCF-7 cells (Fig. 1B, right). We also detected an induced NFAT activation upon treatment of Wnt-5a^low cells with recombinant mouse Wnt-5a (rWnt-5a; 0.8 μg/ml) as measured by faster-migrating, dephosphorylated, active NFAT1 bands on 6% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (PAGE) gels (arrows; Fig. 1C), nuclear translocation of NFAT1, as measured by immunofluorescence (Fig. 1D) and by Luciferase reporter assays (see Fig. 3C). Ionomycin (labeled “i” in Fig. 1C) was used as a positive control (Fig. 1C and D), and the addition of CS inhibited the Wnt-5a-induced NFAT1 activation (Fig. 1C and D and 3C). We also analyzed the protein expression levels of other NFAT family members and found that NFAT5 was highly expressed, whereas NFAT2 is expressed at a moderate level but is surprisingly only weakly affected by ionomycin. NFAT4 is not expressed, and the expression of NFAT3 was not examined (data not shown).

Importantly, recombinant Wnt-5a induced a dose-dependent, distinct, and sustained Ca²⁺ signal in Wnt-5a^low cells, in contrast to recombinant Wnt-3a (Fig. 2A). However, recombinant Wnt-5a did not evoke a canonical signal in HB2 Wnt-5a^low cells as measured by TOP/FOPflash reporters, unless the coreceptor LRP6 was overexpressed (Fig. 2B). Although both recombinant Wnt-5a and Wnt-3a stimulation led to canonical Wnt signaling in HEK293 cells, Wnt-3a did not induce canonical Wnt signaling in HB2 Wnt-5a^low human breast epithelial cells (data not shown). Together, these data indicate that the recombinant Wnt-5a-induced Ca²⁺ signal in HB2 Wnt-5a^low cells is indeed Wnt-5a specific.

A member of the Src kinase family inhibits Wnt-5a-induced activation of NFAT.

Collagen-induced activation of the receptor tyrosine kinase DDR1 requires a Wnt-5a induced activation of a nonreceptor tyrosine Src kinase (8). To investigate whether Wnt-5a-induced stimulation of Src is an upstream event in NFAT activation, we treated the cells with the Src inhibitor PP1. Unexpectedly, this treatment resulted in increased NFAT activity in Wnt-5a^high but not in Wnt-5a^low cells (Fig. 3A and B). The experiments were repeated with recombinant Wnt-5a treatment of Wnt-5a^low cells. As shown in Fig. 3C, recombinant Wnt-5a indeed induced a slight, CS-sensitive activation of NFAT in Wnt-5a^low cells. Notably and in line with the previous experiments, recombinant Wnt-5a in combination with PP1 gave rise to a prominent NFAT activation (Fig. 3C) that was inhibited upon cotreatment with CS. In some cell types, Wnt-5a signaling has been linked to the activation of PKC (21, 30, 32, 65). However, in our experiments, the discrete NFAT activation was affected neither by the inhibition of PKC (with GFX; Fig. 3A and B) nor by the downregulation of that protein (prolonged treatment with PMA; data not shown), indicating that Wnt-5a activates NFAT through a PKC-independent pathway.

Wnt-5a-induced activation of Yes and Cdc42 counteracts its Ca²⁺-induced activation of NFAT.

To evaluate which Src-kinase is activated by Wnt-5a, we performed Wnt-5a-induced tyrosine phosphorylation studies of the Src-kinases Src, Fyn, and Yes. Neither Src nor Fyn was activated by recombinant Wnt-5a as measured by immunoprecipitation and immunoblotting for phosphotyrosine or by phospho Src-enzyme-linked immunosorbent assays (data not shown). However, the Src kinase Yes was phosphorylated by recombinant Wnt-5a already after 5 s and peaked within the first minute (Fig. 4A).

Small Rho GTPases have previously been implicated in noncanonical Wnt signaling (44). Furthermore, the small Rho GTPase Cdc42 can both be activated by Src-kinases (12, 62, 64) and activate many of the kinases that have previously been reported as regulators of NFAT activity (1, 11, 40, 62, 67). In light of this, we investigated the degree to which Rac1 and Cdc42 is activated in human breast epithelial cells. We used the GST-PAK-crib binding assay to measure the relative amounts of active Rac1 and Cdc42 upon treatment of Wnt-5a^low cells with recombinant Wnt-5a and found that while Rac1 might be slightly activated (Fig. 4B), Cdc42 is indeed activated (Fig. 4C). All blots were reprobed for GST, and the total levels of Rac1 and Cdc42 were analyzed in separate experiments (data not shown). It is of interest to note that the levels of actin-bound Cdc42 increased upon addition of Wnt-5a, as judged by analysis of the Cdc42 levels in the Triton-insoluble fractions (data not shown). The mouse embryonic fibroblast cell line SYF lacks all Src-kinase family members (26). We therefore analyzed whether Wnt-5a could activate Cdc42 in these cells. As shown in Fig. 4D, we did not detect an activation of Cdc42 upon Wnt-5a treatment in Src-family kinase-negative SYF cells. The cells expressed both Frizzled-2 and Frizzled-5, and Wnt-5a was able to induce a Ca²⁺ signal in these cells (data not shown and see also Fig. 8B). The Cdc42 activation levels (Fig. 4E, left) were also higher in HB2 breast epithelial cells with high Wnt-5a expression levels, as measured in resting subconfluent cells. Furthermore, the inhibition of Src-kinases (Yes) with PP1 led to decreased activation of Cdc42 in Wnt-5a^high-expressing cells (Fig. 4E, right). This suggests that Wnt-5a-induced signaling in mammary epithelial cells includes stimulation of both the nonreceptor tyrosine kinase Yes and Cdc42.

We next transfected Wnt-5a^high cells and Wnt-5a^low cells with dominant-negative Cdc42 (dnCdc42) and evaluated the levels of NFAT activation and, in support of our hypothesis, the results (Fig. 4F, left) were similar to those obtained upon the addition of PP1. Thus, the fact that PP1 augments the stimulation of NFAT suggests that the Wnt/Ca²⁺ pathway can activate NFAT in the absence of an antagonizing Wnt pathway.

Cytoplasmic localization of NFAT1 is regulated by Wnt-5a.

As previously mentioned NFAT activity is mainly regulated by phosphorylation of serine residues, thus leading to its cytoplasmic sequestration. Whether this process is an active nuclear export mechanism or an active inhibition of nuclear import is not fully understood (50). However, one way of measuring whether Wnt-5a indeed inhibits NFAT activity is to investigate the effects of Wnt-5a on the nuclear export of already-activated NFAT.

First, we investigated whether the addition of PP1 alone affected NFAT activity in Wnt-5a^high cells compared to Wnt-5a^low cells, as measured by Western blot analysis. The phosphorylated form of NFAT1 appears to have a molecular mass of approximately 135 kDa, whereas the active, dephosphorylated form, which is sequestered in the nucleus, is about 120 kDa (23, 52, 55). In line with our previous results we found that treatment with PP1 alone caused a slight activation of NFAT1 in both types of cells, an effect that was more pronounced in the Wnt-5a^high cells (Fig. 5A). Ionomycin activation of NFAT was used as a positive control. Western blotting indicated that the activation reached steady-state levels after approximately 2 to 3 h (data not shown). The degree of activation in the Wnt-5a^high cells cannot be directly compared to the strong activation observed in the luciferase assays, which measure the accumulated effects of Wnt-5a secretion, Wnt/Ca²⁺ signaling, and subsequent NFAT activation over a period of 36 h.

Second, we performed experiments to investigate the effects of Wnt-5a on nuclear export of already-activated NFAT1. Ionomycin is an ionophore that potently activates NFAT. Wnt-5a^high cells were preincubated in the absence or presence of PP1 and subsequently stimulated with 1 μM ionomycin for 5 min, after which the ionomycin-containing medium was replaced with the Wnt-5a conditioned medium lacking (−) or containing (+) PP1 (Fig. 5B), and the extent of nuclear export after withdrawal of ionomycin was analyzed by Western blotting. Indeed, we noticed that NFAT1 activation was prolonged in PP1-treated Wnt-5a^high cells (Fig. 5B), indicating that the noncanonical, Yes/Cdc42-associated Wnt-5a signal has the capacity to also counteract NFAT activation induced by other agonists. Notably, the ionomycin-induced NFAT activation was sustained in untreated Wnt-5a^low cells (Fig. 5C) compared to Wnt-5a^high cells (Fig. 5B). The Wnt-5a-induced inactivation of NFAT1 was also measured using recombinant Wnt-5a (Fig. 5C). Using the same set of experiments but without the addition of PP1, we could see that recombinant Wnt-5a indeed counteracted the ionomycin-induced NFAT activity in Wnt-5a^low cells. CS was used as a control.

Neither JNK, GSK-3β, Pak1, nor Akt is part of the counteracting Wnt-5a signaling.

Since serine/threonine kinases phosphorylate NFAT and hence regulate its activation, we sought to investigate potential Wnt-5a/Yes/Cdc42-activated serine/threonine kinases. It has previously been shown that JNK can phosphorylate and thereby inactivate NFAT (4, 15, 34). The RhoGTPase Cdc42 is a signaling molecule that has been shown to be dependent on Src for its activation (62) and also to act upstream of JNK (40). Furthermore, it has been reported that Cdc42 and the serine/threonine kinase JNK may be involved in noncanonical Wnt pathways (41), although existing data are contradictory (63). Nevertheless, we initially hypothesized that JNK is activated by Wnt-5a via Yes and Cdc42 and that it inhibits NFAT activity induced by an agonist-induced Ca²⁺ signal. First, we investigated the effects of inhibiting JNK on Wnt-5a-induced NFAT activation. The inhibition of JNK did not influence the activation of NFAT in Wnt-5a^high cells compared to Wnt-5a^low cells (Fig. 6A). The results also show that dnRac1 had a much smaller effect on NFAT activation compared to dnCdc42. More importantly, in dnRac1-transfected cells we only found a minute difference in NFAT activation levels between Wnt-5a^low and Wnt-5a^high cells compared to that found in dnCdc42-transfected cells. This indicates that other factors, unrelated to Wnt-5a, can affect NFAT activity via Rac1.

We also examined the possibility that JNK activation is related to the level of Cdc42 activity. To this end, we transfected HB2 neo cells with dnCdc42 or dnRac1 constructs and analyzed the levels of phosphorylation of JNK. The results revealed that inhibition of Cdc42 did not cause any significant changes in JNK activity (Fig. 6B). We analyzed the expression level of the myc-tagged dnCdc42 and dnRac1 in each experiment and measured JNK phosphorylation at different levels of myc expression and at different times after transfection. We also performed these experiments in Wnt-5a^high cells (data not shown). The results obtained using these approaches showed that the activation of JNK is not affected by Wnt-5a-induced stimulation of Cdc42. Using the same experimental strategy, we also found that neither p38 nor extracellular signal-regulated protein kinase phosphorylation/activation was affected by Wnt-5a-induced stimulation of Cdc42 (data not shown). This agrees well with a previous study concerning Dishevelled-induced stimulation of JNK in mammalian cells that showed that neither Cdc42 nor Rac1 was involved in Dishevelled-induced JNK activation (33).

It is well established that GSK-3β is a NFAT kinase, and there is evidence that this protein is stimulated by Cdc42 (11). Therefore, we chose to analyze the status of GSK-3β activation in cells treated with PP1 or the alternative Src-kinase inhibitor Su6656 (Su6). PP1 (Fig. 6C), but not Su6 (data not shown), caused a slight reduction of GSK-3β phosphorylation as measured in whole-cell lysates of Wnt-5a^high and Wnt-5a^low cells. More importantly, the PP1-induced decrease in GSK-3β phosphorylation on Ser-9 cannot explain the PP1-induced activation of NFAT because that would have required inactive GSK-3β (phosphorylated). Another NFAT kinase is Pak1 (67). Pak1 is activated by both Cdc42 and Rac1 (42). However, neither PP1 (Fig. 6D) nor Su6 (data not shown) affected the phosphorylation-status of Pak1 (Ser 144 in the CRIB domain). We also analyzed the Akt phosphorylation levels in these sets of experiments but found no differences (data not shown).

Small Rho GTPases, such as Cdc42, are all well-known modulators of the actin cytoskeleton (35). It is difficult to imagine that Cdc42 can have a direct effect on NFAT1 serine-phosphorylation and activity. However, it was recently suggested that actin polymerization has a direct impact on the nuclear sequestration of NFAT (51). Therefore, we also studied whether using cytochalasin D, to disrupt filamentous actin, would affect Wnt-5a-induced levels of NFAT activity. We found no effect of cytochalasin D in either the Wnt-5a^low or the Wnt-5a^high cells (Fig. 6E).

CK1α binds to NFAT1 upon Wnt-5a stimulation.

As mentioned above, NFAT proteins have specific serine-rich motifs that are phosphorylated upon inactivation. It has been reported that the different motifs are phosphorylated by distinct kinases and the motif required for nuclear import is regulated primarily through phosphorylation by CK1 (45). The CK-family of proteins is constitutively active serine/threonine kinases that act through complex formation, rather than being activated by phosphorylation (27). Thus, in resting T cells, it was shown that CK1 and NFAT formed a cytoplasmic complex that upon activation dissociated, resulting in the subsequent nuclear localization of NFAT. Interestingly, the involvement of CK proteins has been implicated in both canonical and noncanonical Wnt pathways (16, 18, 19, 27, 59, 60). However, it has not yet been studied whether Wnt proteins can induce complex formation between NFAT and CK1 and thereby inhibit the activation of NFAT. To analyze whether CK1α is a downstream effector of Wnt-5a signaling, we stimulated Wnt-5a^low cells with recombinant Wnt-5a and immunoprecipitated NFAT1. As shown in Fig. 7A, we could detect a modest but consistent increase in CK1α binding to NFAT1 upon stimulation with recombinant Wnt-5a. We simultaneously detected a clear activation of NFAT1 as shown by an increase of the faster-migrating band (Fig. 7A, right). We confirmed this association through immunoprecipitations of CK1α and immunoblotting for NFAT1 (Fig. 7B) before and after recombinant Wnt-5a stimulation. Treatment with ionomycin has previously been shown to both activate NFAT but, more importantly, to disrupt the CK1α/NFAT1 binding (45). As shown in Fig. 7B, we demonstrate that upon the addition of recombinant Wnt-5a, the ionomycin-induced disruption of the CK1α/NFAT complex was avoided. In accordance with this, when we inhibited Yes we did not detect the Wnt-5a induced increase in complex formation between CK1α and NFAT1 (Fig. 7B), again indicating that the Wnt-5a-induced Yes/Cdc42 pathway is responsible for the inhibition of NFAT activity. Finally, we examined the ability of recombinant Wnt-5a to counteract ionomycin-induced NFAT activity in Wnt-5a^low cells treated with a CK1α inhibitor (Fig. 7C). Indeed, addition of the CK1 inhibitor prevented Wnt-5a from inactivating NFAT1, as measured by nuclear export experiments and using Wnt-5a as a positive control.

Wnt-5a decreases the invasive capacity of human breast epithelial cells.

In order to examine the biological significance of Wnt-5a-induced effects on NFAT, we used Matrigel invasion assays. Wnt-5a has previously been shown to inhibit tumor cell migration (7). As shown in Fig. 8A, the addition of recombinant Wnt-5a to Wnt-5a^low HB2 cells indeed suppressed the capacity of these cells to invade Matrigels as well. However, when the same set of experiments was performed using the Src-family kinase-negative SYF cells, the addition of Wnt-5a rather induced the invasive capacity (Fig. 8B). To explore whether this possibly could be caused by Wnt-5a-induced NFAT activation, we transfected the cells with an NFAT regulatory domain (VIVIT-GFP) that should disrupt NFAT activation. In agreement with our hypothesis, VIVIT both decreased the basal invasive capacity of SYF cells and disrupted the Wnt-5a induced increase in Matrigel invasion (Fig. 8B). The control cells (ctrl) were transfected with an empty vector, and the transfection efficiency was ca. 50%.