Background
More than 20 metastasis-suppressor genes encoding products with specific metastasis-suppressing functions have been discovered [
1]. KAI1 (CD82) was first identified as a prostate cancer metastasis suppressor through genetic screening [
2]. Studies have shown that the expression of KAI1 is actually down-regulated in most metastatic cancers [
3]. Consistent with this, reduced KAI1 expression is associated with malignant progression of human prostate cancer [
4]. KAI1 is a member of the tetraspan transmembrane superfamily (TM4SF) of type III membrane proteins, specifically of the tetraspanin subgroup. It is ubiquitously expressed, especially in spleen, thymus, prostate and colon, and interacts with a large number of proteins, including integrins, epidermal growth factor receptor (EGFR), and other tetraspanins. KAI1 plays a role in organizing associated proteins into functional signaling networks that influence the metastatic potential of cancer as part of a tetraspanin web [
5].
Metastasis is the final step in the tumor progression processes and involves increased invasiveness, extravasation into secondary organs, and angiogenesis [
6]. Because cancer progression and metastasis involves multiple steps with a high degree of complexity, it requires the contribution of a variety of molecules. Hypoxia, which is the environmental factor best known to induce cancer metastasis, up-regulates the expression of vascular endothelial growth factor (VEGF), which, in turn, induces the formation of tumor-feeding vessels [
7]. It has been documented that the transcription factor, hypoxia-inducible factor 1 α (HIF-1α), is stabilized in the context of reduced oxygen availability and stimulates the expression of VEGF [
8]. An important step in HIF-1α induction is closely related to the activity of the tumor suppressor, von Hippel-Lindau (VHL) protein. VHL is viewed primarily as an ubiquitin ligase for HIF-1α that promotes the proteasome-dependent proteolytic degradation of HIF-1α [
9]. Generally, a loss of VHL expression in cancer leads to up-regulation of HIF-1α and increased VEGF expression [
10].
A previous proteomic analysis of the tetraspanin complex revealed CDCP1 as a new component of the tetraspanin web, showing that it was co-localized with tetraspanin proteins, including KAI1(CD82), CD81, and CD9 [
11]. CDCP1 acts as metastasis enhancer through tyrosine phosphorylation-dependent interactions with Src and PKCδ [
12,
13]. It is a transmembrane protein with three extracellular CUB domains, named for the first three identified proteins (complement factor C1r/C1s, embryonic sea urchin protein uEGF, and bone morphogenic protein-1) containing such a domain. Highest levels of CDCP1 expression are found in colon, skin, small intestine, and prostate [
14‐
16]. One tetraspanin protein, CD9, has been shown to be a CDCP1-binding protein [
12], but certain features of the interaction of CDCP1 with tetraspanin proteins have not been fully defined. Significantly, inhibiting CDCP1 with an anti-CDCP1 antibody was shown to block tumor growth and metastasis in prostate cancer [
16].
Recent studies have revealed that KAI1 attenuates EGFR, integrin, and c-Met signaling, confirming the involvement of KAI1 in the intracellular signaling pathways associated with these molecules [
3,
5]. Moreover, KAI1 expression is reported to inhibit activation of Src kinase in prostate cancer cells [
17‐
20]. The CDCP1-interacting protein Src--the most extensively characterized number of the family of nonreceptor tyrosine kinases--can dominantly contribute to cancer progression, angiogenesis, and metastasis [
21]. In particular, Elevation of Src is functionally linked to the development of prostate cancer [
22]. Thus, how KAI1 interrupts Src kinase activation--a step required for suppression of metastasis--becomes an essential question.
To better understand how KAI1 is involved in the negative regulation of Src and inhibition of metastasis, we postulated an interrelationship between KAI1 and CDCP1 in regulation of Src activity in TEM. To test this supposition, we here examined the function of KAI1 in CDCP1-mediated Src kinase activation and the consequences of this process, focusing on VEGF expression. Restoration of KAI1 expression in PC3 human prostate carcinoma cells blocked Src activation through negative regulation of CDCP1. Notably, HIF-1α and VEGF expression were significantly inhibited by KAI1-mediated Src inactivation and subsequent VHL up-regulation. Importantly, KAI1 inhibited the expression of CDCP1 and HIF-1α in an in vivo tumor xenograft model. These novel findings may indicate that KAI1 functions as a profoundly effective metastasis suppressor in the process of tumor malignancy and angiogenesis through inhibition of CDCP1-mediated Src activation and a subsequent dramatic reduction in HIF-1α and VEGF expression via functional activation of VHL. These studies may shed light on the mechanism by which KAI1 suppresses prostate cancer metastasis.
Methods
Cell culture
PC3 cells (ATCC® CRL-1435) were maintained in RPMI medium 1640 supplemented with 10% heat-inactivated fetal bovine serum and antibiotics. Before exposure to hypoxia, culture medium was replaced by a thin layer of fresh medium (0.15 ml/cm2). Cells were transferred to a Bactron Anaerobic/Environmental Chamber (Sheldon Manufacturing, Inc.), which was flushed with 1% O2, 5% CO2, and 95% N2 at 37°C. In hypoxia-mimic experiments, cells were incubated with 100 μM of CoCl2 under normoxic conditions.
Plasmids and transfection
Plasmid of GFP tagged KAI1 was gifted by Dr. Kyung Keun Kim (Chonnam National University, Korea). VEGF-luc promoter-reporter construct (2.2 kb) and HA tagged VHL expresson vector were also gifted by Dr. Joohun Ha (Kyunghee University, Korea) and Dr. Hong-Duk Youn (Seoul National University, Korea) respectively. HIF-1α expression plasmid was provided by Addgene. Predesigned small interfering RNA (si-RNA) for KAI1, VHL, and PKCδ were purchased from Santa Cruz Biotechnology, Inc (Santa Cruz, CA, USA). Si-RNA for CDCP1 was purchased from Dharmacon (Lafayette, CO). For transient transfection, PC3 cells were plated on 6-well plate 1 day before transfection as a density of 105/well. After 24 h, cells were transfected with plasmids or siRNA using LipofectAMINE 2000 (Invitrogen, Carlsbad, CA, USA). For stable transfection of KAI1, PC3 (KAI1-/-) cells were transfected with GFP-KAI1 expression plasmid. Vector-transfected cells (PC3-GFP clone #8) and KAI1-expressing PC3 clones (PC3-KAI1 clone #5 and #6) were picked after selection in 800 μg/ml G418.
Immunoblotting and imuoprecipitation
For polyacrylamide gel electrophoresis (PAGE), the cells were washed with PBS twice and extracted in RIPA buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 1% NP-40, 0.25% sodium deoxycholate, 0.5% sodium lauryl sulfate, 1 mM EDTA). Cell lysates were centrifuged and supernatants were collected. For immunoprecpiation, cell extracts were incubated with antibody (1:1000 dilutions) with constant agitation at 4°C overnight. After electrophoresis, proteins were transferred to nitrocellulose membrane. Membranes were blocked with 1% bovine serum albumin. Then, protein levels were detected using the following commercial antibodies: anti-HIF-1 α, anti-integrin β1, anti-E-cadherin, anti-N-cadherin, and anti-FAK397 (BD Biosciences, Franklin Lakes, NJ, USA); anti-CDCP1, anti-phospho-p130CASY410, anti-phospho-SrcY416, anti-Src, and anti-VHL (Cell Signaling Technology, Danvers, MA, USA); anti-GFP, anti-PKC δ, anti-KAI1, anti-HA, anti-FAK, anti-snail, anti-VHL, anti-paxillin, and anti-β actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA).
In vitro migration assay
Cell migration assays were performed using Boyden chamber as previously described [
18]. Cells were plated on the upper side of a polycarbonate membrane separating two chambers of 6.5 mm Transwell culture plates (Costar, Corning, NY, USA). After 24 hours, cells on the upper face of the membrane were scraped using a cotton swab and cells that had migrated to the lower face of the membrane were stained with DiffQuick (Baxter Scientific, Deerfield, IL USA) Wright-Giemsa Solution.
In vitro invasion assay
Cell invasion assay was performed on a matrigel invasion chamber (BD Biocoat, BD biosciences, Bedford, MA, USA). This experiment was set up identically to the cell migration assay. After 72 hours, cells remaining upper face of the membrane were scraped using a cotton swab and cells that had invaded to the lower face of the membrane were stained with DiffQuick (Baxter Scientific, Deerfield, IL USA) Wright-Giemsa Solution.
Reverse Transcription-polymerase Chain Reaction (RT-PCR)
Total RNA was extracted with Trizol reagent (Invitrogen, Carlsbad, CA, USA). The cDNA fragment was amplified by PCR using following specific primers: CDCP1, sense 5'-CTTCAACCTCTCCAACTGTG-3' and antisense 5'-TGGTCTGTGCAGCTTATGGT-3', HIF-1α, sense 5'-CTTGCTCATCAGTTGCCACTT-3', and antisense 5'-GCCATTTCTGTGTGTAAGCAT-3'; VEGF, sense 5'- CGAAGTGGTGAAGTTCATGGATG-3', and antisense 5'- TTCTGTATCAGTCTTTCCTGGTGAG -3'; GLUT1, sense 5'-CGGGCCAAGAGTGTGAA-3', and antisense 5'- TGACGATACCGGAGCCAATG-3', GAPDH, sense 5'-TGCTGAGTATGTCGTGGAGTCTA-3', and antisense 5'-AGTGGGAGTTGCTGTTGAAGTCG-3'; β-actin, sense 5'-GTGGGGGCGCCCAGGCACCA-3', and antisense 5'-CTCCTTAATGTCACGCACCATTTC-3'. PCR was initiated in a thermal cycle programmed at 95°C for 5 min, 94°C for 45 s, 60°C for 2 min, and 72°C for 2 min and amplified for 30 cycles. The amplified products were visualized on 1% agarose gels.
PC3 cells were plated onto 6-well plates (5 × 105 cells/well), and the following day cells were co-transfected with a VEGF-luc promoter-reporter construct and KAI1 expression vector. Forty-eight hours after transfection, reporter gene activity was measured by assaying luciferase activitiy (Promega, Madison, WI, USA), and normalized to β-galactosidase activities (Promega, Madison, WI, USA).
Tumor xenograft
Two million PC3 cells were resuspended in 100 μL PBS and injected subcutaneously in the flanks of six-week-old athymic nude mice (Balb/C, Charles River, Japan). Tumor measurements were done with precision calipers and animals were sacrificed after 35 days of injection. Experiments were conducted according to the guidelines for ethical use of animals of our Institution under an approved protocol (approved number of this experiment from ethical committee of animal experiment: KIRAMS 2011-5). Tumors were harvested and fixed overnight in 10% buffered formalin, embedded in paraffin, and sectioned. Primary tumor volumes were calculated with the formula: V = length × (width)2/2.
Immunohistochemistry
Deparaffinized sections were rehydrated and antigen retrieval was performed by 15 minutes incubation in warm trypsin followed by microwave in 10 mM citrate buffer for total of 10. Slides were then blocked with 3% H2O2 followed by blocking in goat serum and primary incubation at 4°C overnight. The slides were then washed in Tris-NaCl buffer and incubated at room temperature for 60 minutes with anti-CDCP1 antibody (R&D systems, Minneapolis, MN, USA) at a 1:50 dilution, anti-HIF-1 antibody (Thermo Fisher Scientific, MA, USA) at a 1:50 dilution, anti-VHL antibody (Thermo Fisher Scientific, MA, USA) at a 1:50 dilution. Slides were rinsed in Tris-NaCl buffer and incubated for 60 minutes with a biotinylated secondary antibody, biotin polyclonal anti-rabbit immunoglobulin (for CDCP1 and VHL) (BD Transduction Laboratories, Lexington, KY, USA) at a 1:200 dilution, or biotin polyclonal anti-mouse immunoglobulin (for HIF-1) (BD Transduction Laboratories, Lexington, KY, USA) (at a 1:100 dilution. Secondary staining was performed and colorized using Vectastain ABC Kit (Vector Laboratories, Burlingame, CA, USA) and 3,3'-diaminobenzidine (DAB)-H2O2 substrate (Sigma, St, Louis, MO). Slides were then counterstained with hematoxylin, dehydrated through graded alcohols and xylene and mounted. Slides were studied and imaged under brightfield microscopy.
Statistical analysis
Data are expressed as means ± standard deviations (SD) of at least three experiments. Statistical significance was determined using Student's t-test for comparisons between two means. The null hypothesis was rejected in cases where p- values were < of 0.05. Densitometry of immunoblots was performed by scanning of the exposed film and using The Quantity one 1-D analysis software (Bio-rad Laboratories, Inc, CA, USA)
Discussion
The purpose of this study was to establish the precise mechanism by which KAI1 (CD82), a metastasis-suppressor gene initially discovered in prostate cancer, inhibits Src and affects the metastasis of prostate carcinoma. Clinical cancer studies have linked a high level of KAI1 expression with a good prognosis. Conversely, the loss of KAI1 expression is correlated with metastasis in tumor progression [
3,
5]. However, how KAI1 functions to suppress metastasis has not yet been clearly established. Tetraspanins, including KAI1, may play a role in the plasma membrane as organizers of multi-molecular complexes that contain not only tetraspanins but also numerous cell membrane components. The TEM has been suggested to serve as a signal-networking platform that shifts the characteristics of a cancer between primary tumor and a metastatic phenotype [
30,
31]. Thus, it is of special interest to define the constituents of this special domain and their interrelationships.
Current studies have proposed that KAI1 functions to inhibit Src kinase activity, but the identified mechanisms have been limited to associations with integrins, EGFR, and c-Met [
5]. It has been proposed that KAI1 attenuates EGFR signaling and inhibits cell motility in breast cancer [
32]. However, other study reported no alteration of EGFR signaling in response to KAI1 expression in DU145 and PC3 prostate cancer cell lines [
18]. It is generally accepted that c-Met can play a role in the metastasis-suppression activity of KAI1 in prostate cancer [
33,
34]; however, inhibition of this pathway and Src kinase by KAI1 expression appear to be independent of one another [
18]. Similarly, and consistent with a previous report [
35], we found no connection between integrin complexes and Src inhibition, showing that depletion of integrin β1, the major component of integrin complexes, had no specific effect on Src phosphorylation status but decrease of phosphorylated FAK and p130CAS, downstream effector of integrin complexes, in PC3 cells (Figure
2A). Therefore, the absence of a connection between these KAI1 targets and inhibition of Src kinase activity prompted us to investigate another target of KAI1.
Together with KAI1, CDCP1 has been reported to be a component of the TEM in colon cancer, but its function in this context has remained poorly characterized [
11]. CDCP1 is a transmembrane protein that primarily associates with Src and PKCδ in a phosphorylation-dependent manner and may act as a scaffold for various interacting proteins in the plasma membrane [
12]. Generally, CDCP1 is highly phosphorylated and functionally activated in metastatic cancer; dysregulated expression of CDCP1 is associated with tumor malignancy [
14,
36‐
38]. Tyrosine phosphorylated CDCP1 has been linked to cell detachment from the extracellular matrix and subsequent cell migration [
39]. The kinase activity of Src is essential for the initial CDCP1 phosphorylation; Src-activated CDCP1, in turn, further potentiates Src kinase activity [
25,
40], although the mechanism for this latter effect is not clear. Our data showed that KAI1-induced inhibition of CDCP1 protein resulted in an almost complete elimination of Src phosphorylation (Figure
2F). This observation suggests that the function of CDCP1 protein, which acts as a positive regulator of Src kinase, may be lost due to KAI1 expression in prostate cancer, raising the question of the possible existence of a feedback control mechanism between CDCP1 and Src.
CDCP1 gene regulation has been investigated in various cancer cell lines. PC3 cells show abundant expression of CDCP1 protein and a low frequency of methylation in transcription-initiation sites [
41]. We found that both HMW and LMW forms of CDCP1 are expressed in PC3 cells (Figure
2). Although the precise mechanism by which CDCP1 is cleaved has not been defined, proteolytic processing by a serine protease results in the generation of the LMW form of CDCP1 and a subsequent increase in LMW phosphorylation [
24]. Our data clearly showed that KAI1 expression decreased the level of both HMW and LMW forms without altering CDCP1 mRNA levels (Figure
2), suggesting that the KAI 1-induced decrease in CDCP1 in PC3 cells reflects the operation of a posttranscriptional mechanism. To test this, we initially treated PC3 cells with cytochalasin D, which blocks cytoskeletal movement, with the goal of determining whether KAI1 affected the endocytic trafficking of CDCP1. We found no effect of this actin-depolymerizing reagent on the KAI1-induced decrease in CDCP1 (data not shown). As demonstrated in previously reported proteomic analyses, CDCP1 and KAI1 are colocalized with various proteases in the TEM [
11], raising the possible involvement of KAI1 in protease activity. Since serine protease-induced proteolysis of CDCP1 has been documented [
24], we treated PC3 cells with protease inhibitors and monitored the levels of CDCP1 under conditions of KAI1 over-expression, but found no significant change in either HMW or LMW forms of CDCP1 (data not shown). Finally, to test the involvement of proteasome-dependent degradation in the process of CDCP1 down-regulation, we treated PC3 cells with the proteasome inhibitor, MG132, and found that inhibiting the proteasome prevented the KAI1-dependent reduction in CDCP1 levels (Figure
3A). Because CDCP1 is reported to localize in the TEM of colon cancer by proteomic analysis, we could not exclude a possible interaction between KAI1 and CDCP1. In experiments designed to determine whether these two membrane proteins were co-localized in PC3 cells, it was hard to detect an interaction of CDCP1 with KAI1 due to KAI1-induced loss of CDCP1 protein level. Our observations indicate that KAI1-induced down-regulation of CDCP1 reflects effects on CDCP1 protein stability, but unknown mechanisms may also participate in this phenomenon.
Angiogenesis is an indispensible step in the progression of prostate cancer, and VEGF has emerged as a critical proangiogenic growth factor in prostate carcinogenesis [
22,
42,
43]. Hypoxia is the environmental factor best known for its ability to induce cancer metastasis; it also stabilizes HIF-1α and up-regulates the expression of VEGF, which, in turn, induces the formation of tumor-feeding vessels [
6]. Moreover, it is known that Src kinase not only affects cell proliferation and migration, it also control angiogenesis via up-regulation of VEGF expression [
44]. However, few studies have focused on the role of KAI1 activity in VEGF expression and vasculature formation in tumors. Thus, given the critical role of KAI1 in inhibiting Src kinase and our motivation to identify a plausible mechanism to account for KAI1 effects in carcinogenesis, we examined HIF-1α and VEGF expression in the context of KAI1 expression. We found that HIF-1α and VEGF expression were dramatically inhibited upon restoration of KAI1 expression (Figures
4 and
5). These findings were supported by the results obtained using an
in vivo xenograft model, which showed that KAI1-expressing tumor volumes were significantly reduced compared with those of controls (Figure
6). Moreover, whereas CDCP1 and HIF-1α expression were reduced in KAI1-expressing tumor tissue, VHL expression was clearly augmented. Thus, our data provide a mechanistic basis for the clear correlation between the loss of KAI1 in cancer and poor prognosis. It has recently been demonstrated that Src can promote oncogenesis through destabilization of the VHL tumor suppressor [
26]. Importantly, functional inactivation of VHL, including through germline mutations, has been well documented in highly vascularized tumors such as renal cell carcinomas, hemangiosarcomas, and pheochromocytomas [
9]. The tumor-suppressive function of VHL is best viewed in the context of its role as an E3 ubiquitin ligase that targets various substrates, including HIF-1α and atypical PKC [
8]. Two forms of VHL, approximately 30 and 18 kDa, exist, but the significance of the two forms is unknown [
45]. It has been suggested that VHL is extremely unstable when not complexed with components of ubiquitin ligase complexes, such as elongin B and elongin C [
9]. This observation may account for the difficulty in detecting VHL protein expression and may help to explain why most VHL functions have been examined by overexpressing or immunoprecipitating VHL. We observed that VHL protein levels were increased by KAI1 expression (Figure
3B, C). How KAI1 negatively regulates CDCP1 remains uncertain, but our data clearly suggest that this effect of KAI1 involves the regulation of CDCP1 protein stability. Intriguingly, VHL over-expression decreased CDCP1 levels; this result was confirmed by treating cells with siRNA against VHL, which had the opposite effect on CDCP1 levels (Figure
3D). VHL protein is expressed in various normal and cancer tissues, where it is localized in the cytosol or membrane [
46]. In human renal cell carcinoma, its presence in the membrane is significantly associated with a missense mutation [
10,
47]. These reports prompted us to further consider changes in subcellular localization of VHL in the process of tumorigenesis. In addition, we observed the existence of VHL in both of cytosol and membrane fractions in VHL overexpressing PC3 cells (data not shown). This observation raises the possibility of recruitment of VHL from cytosol to plasma membrane and ubiquitination of unknown substrates located in the membrane. Although we could not observe co-localization of CDCP1, VHL and KAI1 in cell membrane or TEM, these proteins would be expected to coexist in the plasma membrane during the process of prostate carcinogenesis.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JJP performed experiments; YBJ did tumor xenograft and immunohistochemistry; YJL helped performing experiments. JSL helped designing the research; YSL helped designing the research; YGK designed, analyzed data, and wrote paper; ML designed, analyzed data, and wrote paper. All authors have read and approved this manuscript