Five cancer cell lines were used in this study. MCF7, T47D, HEK293T and MDA-MB-231 cancer cells were cultured in DMEM medium and SKBR3 cells were grown in RPMI medium (Welgene Inc., Daegu, Korea). All media were supplemented with 10% FBS (Gibco-BRL, Carlsbad, CA, USA) and 1% antibiotic-anti-mycotic (Gibco-BRL). All cultures were grown at 37°C in a humidified atmosphere composed of 95% air and 5% CO₂. LY294004 (cat no. 440202) and Cpd22 (cat no. 407331) were purchased from Calbiochem (La Jolla, CA, USA).

Total RNA was extracted from MDA-MB-231 cells with the RNeasy mini kit (cat no. 74106; Qiagen, Hilden, Germany) following the manufacturer’s instructions. For reverse transcription, 1 μg RNA of each sample was subjected to cDNA synthesis using an oligo (dT) primer and the ImProm-II™ Reverse Transcription System (A3800; Promega, Madison, WI, USA). PCR amplification was performed using 10 ng cDNA, different sets of primers, and AccuPower PCR PreMix system (Bioneer, Daejeon, Korea). The amplification reaction was carried out using a PCR Thermal Cycler Dice (Applied Biosystems, Foster City, CA, USA). Each gene product was amplified using corresponding pairs of primers, in which β-actin gene product was used as an internal control. The oligonucleotide sequences for RT-PCR analysis were: pRT-p34^SEI-1-RT forward, 5′-AGGACCTCAGCCACAT TGAG-3′ and reverse, 5′-GGTGCCCAAAGTTCATTGTC-3′; pRT-HER2/neu-RT forward, 5′-CTGAACTGGTGTATGC AGAT-3′ and reverse, 5′-CCACACAGTCACACCATAA-3′; pRT-NEDD4-1 forward, 5′-TGGGACATCACTTTGT GATC-3′ and reverse, 5′-TGAGGCTTTTACTGGGGTC-3′; pRT-β-catenin forward, 5′-CATTTCCAATCTACTAATGC-3′ and 5′-CTGCATTCTGACTTTCAGTA-3′; pRT-c-MYC forward, 5′-ACCAGCAGCGACTCTGAGGA-3′ and reverse, 5′-TGACCCTCTTGGCAGCAGGATAGTCC-3′; pRT-ACTB forward, 5′-AGGTCGGAGTCAACGGATTTG-3′ and reverse, 5′-GTGATGGCATGGACTGTGGT-3′.

Cells were recovered from culture by centrifugation at 3,000 rpm for 1 min and washed twice in an ice-cold phosphate-buffered saline (PBS) buffer. The cells were then lysed in RIPA lysis buffer and the protein was quantified using a protein assay kit (Bio-Rad, Hercules, CA, USA). Approximately 25 μg of total protein per sample was subjected to 12% SDS-PAGE and the resolved proteins were transferred to Immobilon transfer membranes (cat no. IPVH00010, Millipore, Billerica, MA, USA). The filter was blocked in 5% non-fat dry milk/0.1% Tween-20/Tris-buffered saline (TBS) followed by incubation with each corresponding antibody and immune-detection was accomplished using the Power Opti-ECL Western blotting detection reagent (Bionote, Hwaseong, Korea). Antibodies used in this study were purchased as follows: p34^SEI-1 (ALX-804-645; Enzo Life Sciences, Farmingdale, NY, USA), NEDD4-1 (sc-25508) and PTEN (sc-7974; from Santa Cruz Biotechnology, Santa Cruz, CA, USA), E-cadherin (cat no. 610181) and N-cadherin (cat no. 610921; from BD Biosciences, Franklin Lakes, NJ, USA), vimentin (sc-7557), fibronectin (sc-8422), HER2/neu (sc-33684; from Santa Cruz Biotechnology), pAKT^ser473 (cat no. 9271; Cell Signaling, Danvers, MA, USA), pILK^thr178 (sc-130196; Santa Cruz Biotechnology), pGSK3β^ser9 (cat no. 2435-1; Epitomics, Burlingame, CA, USA) and γ-tubulin (sc-7396; Santa Cruz Biotechnology).

For overexpression of p34^SEI-1, cells were plated at 1×10⁶ cells in a 60-mm-diameter culture dish and transfected with 4 μg of either the control empty vector (pEF-BOS-EX) or the C-terminally Flag-tagged human p34^SEI-1 protein (pEF-p34^SEI-1-Flag) for 6 h in serum free medium using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). For knockdown of p34^SEI-1 expression, MDA-MB231 cells were plated as above and transfected with 4 μg of either empty control vector (pLKO.1) or p34^SEI-1 directed shRNA expression vector [pLKO.1/p34^SEI-1-short hairpin (sh) RNA] for 6 h in serum free medium using Lipofectamine 2000. After replacing the DNA-Lipofectamine complex-containing medium with complete, antibiotic-free growth medium, transfected cells were incubated for 48 h. For knockdown of endogenous HER2/neu, SKBR3 cells were plated 4×10⁵ in a 60-mm-diameter culture dish and transiently transfected with 200 pmol of either scramble control RNA (scRNA) or a HER2/neu silencing siRNA (siHER2/neu) for 2 h in serum free medium using Lipofectamine 2000. After replacing the RNA-Lipofectamine complex-containing medium with complete growth medium, transfected cells were incubated for 48 h. pEF-BOS-EX and pEF-p34^SEI-1-Flag plasmids were kindly provided by Dr Rikiro Fukunaga (Osaka University, Osaka, Japan) and pLKO.1/p34^SEI-1-shRNA plasmid was obtained from OriGene (http://www.origene.com). HER2/neu oligonucleotides were chemically synthesized by ST Pharm Co. Ltd (Seoul, Korea).

Cell migration was estimated by a wound-healing migration assay and monitored by microscopy. MCF7 and MDA-MB231 cells were transfected with p34^SEI-1 overexpressing pEF-p34^SEI-1-Flag or p34^SEI-1 suppressing pLKO.1/p34^SEI-1-shRNA vector with each corresponding control vector, respectively. Each cell line was fully cultured in a 60-mm-diameter culture dish and a scratch was made on the monolayer using a sterile white tip. The distance of migration by cancer cells was measured after 48 h.

In vitro invasion assay was performed using a Transwell membrane apparatus (Corning Life Sciences, Tewksbury, MA, USA) and Matrigel (BD Biosciences, Seoul, Korea). SKBR3, MCF7 and MDA-MB-231 breast cancer cells (2×10⁵ cells) were placed in the Matrigel-coated upper chamber of the apparatus. Medium to each cell line was placed in the lower chamber as a source of chemo-attractants. Incubation was carried out for 24 h at 37°C. The capacity of these cells to invade through the semi-solid Matrigel was estimated by fluorescence.

IHC data using human tissue samples were kindly provided by Dr Chang-Jin Kim at Soonchunhyang University Hospital (Chonan, Korea) and it was obtained as previously described (28).

To investigate whether p34^SEI-1 oncoprotein is involved in the development of metastasis, the effect of p34^SEI-1 on migration and invasion was tested by wound healing migration and Matrigel invasion assays. In the wound healing assay, cell mobility was enhanced in MCF7 cells transfected with p34^SEI-1 overexpressing pEF-p34^SEI-1-Flag vector, while it was reduced in MDA-MB-231 cells transfected with p34^SEI-1 suppressing pLKO.1/p34^SEI-1-shRNA vector compared to control cells (Fig. 1A). The Matrigel invasion assay showed that overexpression of p34^SEI-1 in SKBR3 and MCF7 cells increased the invasiveness compared to control cells, while knockdown of p34^SEI-1 in MDA-MB 231 cells decreased invasiveness (Fig. 1B). These data strongly suggest that p34^SEI-1 exerts a positive effect on the cell migration and invasion in vitro implying the involvement of p34^SEI-1 in metastasis. During the EMT process, epithelial markers like E-cadherin are diminished, while expression of mesenchymal markers including N-cadherin, vimentin and fibronectin increase due to activation of matrix metalloproteinases (MMPs) inducing metastasis (38). Accordingly, the expression levels of EMT-related proteins were checked using western blot analysis. The representative epithelial marker E-cadherin was decreased but its antagonist N-cadherin was increased in SKBR3 and MCF7 cells after transfection with pEF-p34^SEI-1-Flag vector (Fig. 1C). Furthermore, vimentin another mesenchymal marker, was also increased in SKBR3 and MCF7 cells. The opposite result was obtained in MDA-MB-231 cells transfected with pLKO.1/p34^SEI-1-shRNA vector compared to control (Fig. 1C). Taken together, the findings indicate that p34^SEI-1 promotes metastasis by enhancing migration and invasion of cancer cells.

To support the conclusion that p34^SEI-1 has metastatic potential, immunohistochemistry was performed to test whether p34^SEI-1 expression is clinically related with the metastasis of breast cancer. Indicated tissue samples were stained with p34^SEI-1 antibody and the degree of positive signals was estimated. Normal lobular and ductal hyperplastic cells showed no strong positive signal (0 of 20 samples) while 8 of 20 (40%) ductal carcinoma in situ and 13 of 20 (70%) invasive ductal carcinoma samples displayed strong positive signals (Fig. 2). The invasive carcinoma showed more strong expression than non-invasive ductal carcinoma in situ. The representative immunohistochemistry data are shown in Fig. 2, in which p34^SEI-1 was not or very weakly expressed in normal lobular epithelial or ductal hyperplasia breast cancer tissues, but was more strongly expressed in invasive ductal carcinoma tissue samples than in ductal carcinoma in situ. The data indicate that p34^SEI-1 expression increases as the tumor invasiveness progresses in human breast tissues, strongly relating p34^SEI-1 with breast cancer metastasis.

To elucidate the mechanism of how p34^SEI-1 promotes metastasis during cancer cell tumorigenesis, western blot analysis was employed to check the expression levels of the main components of the PI3K-AKT pathway, NEDD4-1, PTEN, phosphorylation of AKT on serine 473 residue (pAKT^ser473) and phosphorylation of GSK3β on serine 9 residue (pGSK3β^ser9) after overexpression or suppression of p34^SEI-1. In MCF7, T47D and HEK293T cells transfected with pEF-p34^SEI-1-Flag, p34^SEI-1 induced increased pAKT^ser473 and pGSK3β^ser9 protein levels at least partly by inducing NEDD4-1-mediated PTEN degradation as we previously reported (20). In MDA-MB231 cells transfected with pLKO.1/p34^SEI-1-shRNA vector, suppression of p34^SEI-1 resulted in a decrease of NEDD4-1 and an increase of PTEN compared to the control. Consequently, the protein levels of pAKT^ser473 and pGSK3β^ser9 decreased (Fig. 3A). However, SKBR3 cells transfected with pEF-p34^SEI-1-Flag revealed a different expression pattern, in which pAKT^ser473 was unexpectedly diminished. More interestingly, expression levels of GSK3β phosphorylation and β-catenin, downstream target of GSK3β, were not affected by the AKT inactivation (Fig. 3A). This fact implies the presence of another upstream kinase regulating GSK3β phosphorylation regardless of AKT inactivation. This speculation was supported by RT-PCR results showing that both classes had very similar expression pattern to the NEDD4-1, β-catenin, and its downstream target, c-MYC, at the transcriptional level. Overexpression of p34^SEI-1 produced an increase of NEDD4-1, β-catenin and c-MYC in SKBR3, MCF7 and HEK293T cells, whereas, in MDA-MB231 cells p34^SEI-1 was suppressive (Fig. 3B). The data implicated an unknown kinase in this pathway downstream of NEDD4-1 and upstream of GSK3β. Collectively, the data indicate that p34^SEI-1 may promote cancer metastasis using distinct signaling pathways in two different types of cancer cell lines with different genetic background.

p34^SEI-1 seems to promote metastasis by activating the PI3K/AKT signaling pathway. During this process, overexpression of p34^SEI-1 decreased the phosphorylation of AKT on 473 serine residue in SKBR3 cells but increased phosphorylation in MCF7, T47D and HEK293T cells. Considering that SKBR3 is HER2/neu-positive cell line, but MCF7, T47D and HEK293T are HER2/neu-negative cell lines, it was assumed that HER2/neu might be responsible for the decrease of pAKT^ser473 protein level in p34^SEI-1 overexpressing SKBR3 cells, since HER2/neu is a positive regulator of the PI3K/AKT signaling pathway and its expression was significantly decreased by p34^SEI-1 overexpression (Fig. 4A). This result was consistent with the view that p34^SEI-1 suppresses HER2/neu expression and downregulated HER2/neu inhibits the phosphorylation of AKT at the 473 serine residue. Unexpectedly, the phosphorylation level of GSK3β on 9 serine residue was increased despite AKT inactivation (Fig. 3A). This finding indicated that GSK3β might be phosphorylated by another factor rather than AKT in SKBR3 cells. To elucidate the mechanism, ILK was at first suspected to be responsible for GSK3β phosphorylation because ILK is known to affect PI3K/AKT signaling pathway by directly phosphorylating the 9 serine residue of GSK3β (23). The phosphorylation levels of ILK on the 178 threonine residue were checked in both HER2/neu-positive and -negative cell lines. pILK^thr178 protein level was significantly increased in HER2/neu-positive SKBR3 cells while no change was found in HER2/neu-negative MCF7, HEK293 and T47D cells (Fig. 4A). The protein level of pILK^thr178 was significantly induced when phosphorylation of pAKT^ser473 was inhibited by decreased HER2/neu. However, it was not changed in HER2/neu-negative cells, in which AKT phosphorylation was induced (Fig. 4A). This observation indicates that p34^SEI-1 overexpression promotes cancer metastasis by inducing ILK instead of AKT when HER2/neu is diminished or depleted. Further RT-PCR analysis showed that the negative effect of p34^SEI-1 on HER2/neu expression occurred both at the transcriptional and the translational levels (Fig. 4B).

Taken together, the data demonstrates that p34^SEI-1 activates the PI3K/AKT signaling pathway using at least two different types of signaling pathways depending on HER2/neu expression status.

Decreased HER2/neu activity by p34^SEI-1 overexpression might be responsible for the activation of ILK signaling in SKBR3 cells. This idea prompted the assessment of the dependence of the activation of ILK on HER2/neu. The phosphorylation of AKT, ILK and GSK3β was checked at the protein level after SKBR3 and MCF7 cells were transfected with HER2/neu specific siRNA or overexpressing pHER2/neu vector. In HER2/neu silenced SKBR3 cells, pAKT^ser473 protein level was decreased, probably due to HER2/neu mediated inhibition of PI3K, a direct downstream target of HER2/neu. Surprisingly, both pILK^thr178 and pGSK3β^ser9 protein levels were increased after treatment of HER2/neu silencing siRNA, in which GSK3β was thought to be phosphorylated by the ILK rather than AKT (Fig. 5A). On the other hand, the expression levels of same proteins were also checked after MCF7 was transfected with HER2/neu overexpressing vector. Overexpression of HER2/neu highly promoted the phosphorylation of AKT on 473 serine residue but reduced that of ILK on the 178 threonine residue. PTEN was not affected by neither HER2/neu inhibition or overexpression (Fig. 5A). In an extended experiment, the switching relationship between AKT and ILK was investigated using Cpd22 of ILK inhibitor and LY29002 of PI3K/AKT inhibitor. In both groups, PTEN was decreased whenever p34^SEI-1 was overexpressed probably due to p34^SEI-1 mediated NEDD4-1 activation as we showed before (28). Very importantly, both groups showed an inverse relationship between pAKT^ser473 and pILK^thr178 expression levels. When HER2/neu-positive SKBR3 cells were treated with Cpd22, pAKT^ser473 protein level increased even under p34^SEI-1 overexpression (Fig. 5B). When HER2/neu-negative MCF7 cells were treated with LY29002, the ratio of ILK phosphorylation was significantly elevated (Fig. 5B). LY29002 treatment after p34^SEI-1 overexpression produced an even higher level of pILK^thr178 expression (Fig. 5B). It may be explained by the strong switching relationship between AKT and ILK. In both cases, p34^SEI-1 overexpression and treatment of Cpd22 and LY29002 tremendously increased the phosphorylation of GSK3β on the 9 serine residue in both HER2/neu-positive and -negative cell lines. This may be the basis for progression of metastasis because either AKT or ILK phosphorylation could promote the metastasis under the circumstance that one of them is inhibited (Fig. 5B).

Taken together, the data demonstrate that p34^SEI-1 induces the activation of either AKT or ILK signaling in a HER2/neu-dependent manner (Fig. 6).