Fax: (951) 827-5504
Human chorionic gonadotropin β (HCGβ) down-regulates E-cadherin and promotes human prostate carcinoma cell migration and invasion
Version of Record online: 1 DEC 2005
Copyright © 2005 American Cancer Society
Volume 106, Issue 1, pages 68–78, 1 January 2006
How to Cite
Wu, W. and Walker, A. M. (2006), Human chorionic gonadotropin β (HCGβ) down-regulates E-cadherin and promotes human prostate carcinoma cell migration and invasion. Cancer, 106: 68–78. doi: 10.1002/cncr.21549
- Issue online: 23 DEC 2005
- Version of Record online: 1 DEC 2005
- Manuscript Accepted: 19 JUL 2005
- Manuscript Revised: 23 JUN 2005
- Manuscript Received: 11 APR 2005
- National Institutes of Health (NIH). Grant Number: DK 61005
- human chorionic gonadotropin β (HCGβ);
- cell migration;
- prostate carcinoma;
- tumor invasion;
Membrane-associated human chorionic gonadotropin β (HCGβ) is correlated with a poor prognosis in localized prostate adenocarcinoma. The relationship between HCGβ and metastasis, however, is unclear.
To shed some light on the issue, two stable prostate carcinoma cell lines overexpressing HCGβ, designated DU145 HCGβ and PC3 HCGβ, were created and compared with empty vector stably transfected DU145 and PC3 cells (control cells).
HCGβ expression resulted in a change in morphology; the cells were more elongated and had multiple pseudopodia, while the control cells were more rounded. This change in morphology was duplicated by incubating control cells in conditioned medium from the DU145 HCGβ or PC3 HCGβ cells, or by adding purified HCGβ to control medium. The DU145 HCGβ and PC3 HCGβ cells were also less adherent than the controls, as assessed by the ease with which trypsin-EDTA could remove them from culture plates. Reduced adherence could be duplicated by incubation of control cells with either conditioned medium or purified HCGβ. Western blot analysis showed that DU145 HCGβ and PC3 HCGβ cells expressed less E-cadherin than control cells and that a change of medium increased expression of E-cadherin. Addition of conditioned medium, or purified HCGβ, to control cells down-regulated E-cadherin. Cell migration and invasion assays showed that DU145 HCGβ and PC3 HCGβ cells were more migratory and invasive than controls and that treatment of control cells with either conditioned medium or purified HCGβ increased their migratory/invasive capacity.
The data indicate that HCGβ is directly responsible for changes in prostate carcinoma cells associated with an increased metastatic phenotype. Cancer 2006. © 2005 American Cancer Society.
Human chorionic gonadotropin (HCG) is primarily a placental hormone composed of two unequal, noncovalently linked subunits, α and β.1 As is true for many hormones, however, several tissues express lower amounts of HCG, and HCG in the tissues and circulation of males and nonpregnant females has been shown to be a useful tumor marker for a variety of neoplasms.2–7 Correlations between phenotype and the presence of HCG specifically in prostate carcinoma cell lines showed that high expression of HCGβ was associated with a metastatic phenotype.8 Moreover, the presence of HCGβ in prostatic adenocarcinoma identifies a population of patients with a poor prognosis.9 What is missing from previous analyses, however, is demonstration that HCGβ directly affects the prostate carcinoma cells in a way that might increase their metastatic potential.
E-cadherin is a major player in epithelial cell–cell interactions. It is a Ca2+-dependent cell adhesion molecule that binds Ca2+ in its extracellular domain.10 This Ca is required for cadherin–cadherin interactions between cells. The cytoplasmic tail of E-cadherin interacts with intracellular molecules, such as the catenins, which in turn interact with the actin cytoskeleton.11–15 E-cadherin is thought to be an inhibitor of tumor migration and invasion by the promotion of cell–cell interactions.16 E-cadherin levels have been shown to be reduced in prostate carcinoma cells.16
In this study, we examined the hypothesis that HCGβ directly affects prostate carcinoma cells, increasing their invasive potential.
MATERIALS AND METHODS
Human prostate carcinoma cell lines (DU145, PC3) were purchased from the American Type Culture Collection (Rockville, MD). The cells were cultured in RPMI 1640 medium (Invitrogen, Carlsbad, CA) supplemented with 10% FBS (Invitrogen), 100 units/mL penicillin, and 100 μg/mL streptomycin (Invitrogen). pVSneo-HCG β construct was purchased from Stratagene (La Jolla, CA). Restriction enzymes XhoI and SalI and T4 DNA ligase were purchased from Invitrogen. An HCGβ detection kit, F-HCG β Accubind ELISA, was purchased from Monobind (Costa Mesa, CA). Geneticin (G418), used for establishing stable cell lines, and crystal violet for staining cells were obtained from Sigma (St. Louis, MO). Transwells, used for cell migration and invasion assays, were from Costar (Cambridge, MA). Matrigel Basement Membrane Matrix for the cell invasion assay was a product of BD Biosciences (Bedford, MA). The monoclonal antibody against HCGβ was purchased from BioDesign International (Saco, ME) and that against E-cadherin was purchased from BD Biosciences.
Establishment of Stable Cell Lines
The control vector was produced by using XhoI and Sal I to remove HCGβ cDNA from the pVSneo–HCGβ construct, followed by autoligation. This control construct, pVSneo Vector, was 5445 basepairs (bp) in size. The constructs pVSneo–HCGβ or pVSneo–Vector, were amplified in E. coli (DH5α) and used for transfection. In a six-well tissue culture plate (Invitrogen), 2 × 105 cells were seeded per well in 5 mL of RPMI 1640 medium. The cells were incubated at 37 °C in a CO2 incubator until 50–70% confluency. Cells were washed twice with serum-free RPMI 1640 medium. A prepared mixture of pVSneo–HCGβ or pVSneo–Vector DNA (1 μg) and lipofectamine (6 μL) (Invitrogen) in 200 μL of serum-free medium was diluted to 1 mL by adding 800 μL serum-free medium to the mixture. The final solution was overlaid onto the rinsed cells. The cells were incubated at 37 °C for 72 hours. Cells were then passaged 1:10 into the selection medium containing geneticin (1 mg/mL). Cells were grown continually for 2 weeks and single-cell colonies were picked up. The screened cells were cultured in selection medium (400 μg/mL geneticin) for 2 more weeks until no additional dead cells were found. Cells were maintained in medium containing 400 μg/mL geneticin.
Cell Dispersal Time
In a 100-mm tissue culture plate, 2 × 105 DU145 HCGβ, DU145 vector, PC3 HCGβ, and PC3 vector cells were grown for 3 days alongside DU145 vector and PC3 vector cells treated with previously collected 3-day conditioned medium from DU145 HCGβ or PC3 HCGβ cells, and DU145 and PC3 vector cells treated with 200 ng/mL HCGβ standard. Cells were grown in RPMI 1640 medium (10 mL) containing 10% FBS until confluency. After the cells were washed with Dulbecco phosphate-buffered saline (DPBS) (Invitrogen) three times, they were incubated in 3 mL Trypsin-EDTA at 37 °C (same batch in each dish) (0.25% trypsin, 1 mM EDTA) (Invitrogen) and the time for cells to form a single-cell suspension without vibration or shaking was monitored. Final cell numbers were very similar in all dishes (1 × 107).
Cells exposed to endogenous HCGβ, conditioned medium, or purified HCGβ and appropriate controls were viewed and photographed using a Nikon Phase Contrast microscope linked to a 3CCD camera (DAGE-MTI, Michigan City, IN) with the PAX-it 5.0 program (LECO, St. Joseph, MI). Cell morphology was observed before confluency so that cell shape and pseudopodia could be clearly distinguished.
Cells were rinsed with DPBS and lysed with a buffer containing 20 mM Tris-HCL, pH 7.4, 140 mM NaCl, 0.05 mM EDTA, 10 μg/mL leupeptin, 10 μg/mL aprotinin, 25 μg/mL pepstatin, 1 mM PMSF, 1 mM Na3VO4, 10 nM NaF, 1 mM EGTA, and 1% NP-40. After centrifugation at 12,000g for 10 minutes the supernatant was collected (cell lysate). Protein concentration was measured by the Bradford method. Thirty μg of protein was loaded on a 12% reducing SDS-PAGE gel. After electrophoresis, protein was transferred to a nitrocellulose membrane in transfer buffer containing 48 mM Tris, 39 mM glycine, 0.1% SDS, and 20% methanol (pH 8.3). The membrane was blocked with 5% nonfat milk in wash buffer (DPBS containing 0.1% Tween 20). Filters were then incubated with monoclonal antibodies against HCGβ or E-cadherin (dilution 1:1000) overnight at 4 °C. After washing 3 × 10 minutes each, the blot was incubated in secondary antimouse immunoglobulin G (IgG) conjugated to horseradish peroxidase (Sigma) at a dilution of 1:10,000 for 45 minutes at room temperature. After three washes, filters were exposed to ECL reagent (Amersham Biosciences, Piscataway, NJ) followed by autoradiography and image analysis. Filters were reprobed after stripping in wash buffer with 0.1 M β-mercaptoethanol and 2% SDS for 30 minutes at room temperature with agitation. β-Actin was used to normalize for loading and transfer.
Migration of prostate carcinoma cells through 8-μm pores was assessed using Transwell culture chambers (6.5 mm diameter). Cells (2 × 105) were seeded in the upper chambers in 500 μL of serum-free RPMI 1640 medium. One mL of the same medium was placed in the lower chambers. The cells were incubated for 6 hours. The medium was discarded and the cells were washed with DPBS twice. Then the cells were fixed with 100% methanol for 10 minutes at –20C. After aspirating the methanol, the cells were stained with 0.5% crystal violet solution (made in 25% methanol) at room temperature for 10 minutes. The crystal violet solution was poured off and the cells were rinsed with distilled water until excess dye was removed. The cells and dye in the upper chamber were removed using a cotton swab. The cells that had migrated to the underside of the filter were viewed and photographed. The relative number of cells that had migrated was determined by counting five random fields at a magnification of ×200.
Invasion of prostate carcinoma cells through Matrigel and the 8-μm pores was studied using Matrigel-coated Transwells. The filters were coated with Matrigel Basement Membrane Matrix (160 μg/6.5 mm well, the concentration given in the BD Biosciences user's manual). Matrigel is a solubilized basement membrane from the Engelbreth-Holm-Swarm mouse sarcoma. Cells (2 × 105) were seeded as before and then incubated for 24 hours, after which time processing was as described above.
Western blot analysis of cell lysates showed bands recognized by anti-HCGβ in the DU145 HCGβ and PC3 HCGβ cells, while the control, empty vector-transfected cells showed no band (Fig. 1). Purified HCGβ (200 ng) gave two major bands. It is interesting to note that the band in DU145 HCGβ cells migrated to the same point as the upper of the two bands in the purified material, while the band in the PC3 HCGβ cells had a slightly lower molecular weight and migrated between the two bands on the standard. HCGβ is a highly glycosylated molecule,1 and these differences in migration are most likely related to the relative degree of glycosylation.
HCGβ was also detected by ELISA in cell lysates and conditioned medium from cells 3 days postconfluency. Conditioned medium from DU145 HCGβ cells and PC3 HCGβ cells contained 70 and 160 mIU/mL/106 cells and the cell lysates contained 15 and 30 mIU/mL/106 cells, respectively. HCGβ was therefore shown to accumulate in the medium relative to the cell content. As was evident from the Western analysis, it is also clear that the PC3 HCGβ cells made more HCGβ than the DU145 HCGβ. One mIU/mL is equivalent to 1 ng/mL HCGβ.
Susceptibility of Cultures to Trypsin-EDTA
As shown in Figure 2A, cells expressing HCGβ are more susceptible to trypsin-EDTA dispersal than control cells. Figure 2B shows that incubation in conditioned medium containing HCGβ has a similar, if lesser, effect, as does incubation in the HCGβ standard (200 ng/mL) (Fig. 2C). The dose of 200 ng/mL was determined empirically as an amount that altered cell morphology in a 3-day incubation postconfluency.
HCGβ Changes the Morphology of Both DU145 and PC3 Cells
Control DU145 cells have an epithelioid morphology, with most cells being somewhat round and having few cellular processes (Fig. 3A). The DU145 HCGβ cells, by comparison, are less rounded and have more cellular processes (Fig. 3B). Treatment with conditioned medium from HCGβ cells or purified HCGβ has the same effect, once again producing more elongated cells with long processes (Fig. 3C,D). The morphologic changes are more difficult to discern in PC3 cells because control PC3 cells have cellular processes (Fig. 3E,F).
When medium was changed on the DU145 HCGβ cells, or DU145 cells incubated with HCGβ, so that HCGβ was reduced or eliminated, the elongated morphology reverted to control morphology within 30–60 minutes (not shown). It then took 4 hours of HCGβ accumulation to reestablish the long cellular processes in DU145 HCGβ cells (not shown).
HCGβ Decreased E-cadherin Expression in DU145 and PC3 Cells
E-cadherin is a Ca-dependent protein that plays a major role in cell adhesion. We therefore investigated whether HCGβ regulated E-cadherin levels in these cell lines. The results show that DU145 HCGβ and PC3 HCGβ cells had lower levels of E-cadherin than their control counterparts (Fig. 4A). The effect of HCGβ expression was greater in the PC3 HCGβ cells, presumably because they express and release larger amounts of HCGβ. In addition to decreasing the main E-cadherin band, elimination of other immunopositive bands was evident when the control cells were compared with the DU145 HCGβ cells. Bands at 125 kDa, 70 kDa, and 50 kDa were eliminated. Conditioned medium containing HCGβ or HCGβ (200 ng/mL) produced similar, although less dramatic results (Fig. 4B,C). The β-actin bands demonstrate that HCGβ had no general negative effect on protein expression. When the expression of E-cadherin was followed as a function of time after a change of medium in either the DU145 HCGβ or DU145 vector cells, Figure 5 shows that removal of medium resulted in induction of E-cadherin protein by 60 minutes. The increased expression of E-cadherin was maintained at 90 minutes, but had reverted to the regular culture level by 120 minutes. This correlates with the removal and then subsequent accumulation of HCGβ and the changes in morphology previously described. Changing the medium had no effect on the levels of E-cadherin in the DU145 vector cells.
HCGβ Promotes Cell Migration
The results show that more of the DU145 HCGβ and PC3 HCGβ cells migrated through the pores in 6 hours when compared with control cells (Fig. 6A). In addition, HCGβ present in conditioned medium, or purified HCGβ, increased the migration rate of control cells (Fig. 6B,C). Corresponding representative images of migrated cells are given in Figure 7A–H.
HCGβ Promotes Cell Invasion
Basement membranes are thin extracellular matrices underlying epithelial cells. Matrigel is a commercial product extracted from a mouse sarcoma rich in extracellular matrix proteins. The major component is laminin, followed by collagen IV and heparan sulfate proteoglycans. After a 24-hour incubation on Matrigel-coated Transwells, more DU145 HCGβ and PC3 HCGβ cells had invaded the basement membrane material and passed through the pores to reach the underside of the filter (Fig. 8A). Treatment of control cells with conditioned medium or HCGβ increased their invasive capacity (Fig. 8B,C). Corresponding representative images of invading cells are given in Figure 9A–H.
Previous work in this field had demonstrated a strong correlation between the ability of human cancer cells to metastasize and the expression of membrane-associated HCGβ.8 This work was performed using human tumors grown in nude mice. In addition, others have shown that cell membrane-associated HCGβ identifies a subgroup of prostate adenocarcinoma patients with a poor prognosis, regardless of the histologic grade of the tumor.9 Both of these studies strongly suggest a connection between HCGβ expression and invasive characteristics, but it is not clear whether the HCGβ is responsible for the increased metastatic potential or is just a useful diagnostic side effect of other changes. In this study, we directly addressed the role of HCGβ in the production of a more metastatic phenotype by creating stable cell lines overexpressing HCGβ. These stable cell lines are therefore continually exposed to locally produced HCGβ. In addition, we compared the ability of conditioned medium from these cell lines to produce noted effects in control cells, thereby establishing that it is the secreted HCGβ and not an artifact of protein overexpression that is responsible for the effects observed. Finally, we compared conditioned medium to effects of purified HCGβ to ensure that each effect is a function of the HCGβ itself and not another component of conditioned medium produced secondarily to HCGβ overexpression. For each assay, autocrine expression of HCGβ usually produced a larger effect than conditioned medium or purified HCGβ, most likely because the concentration at the cell surface was higher than was achieved by just adding HCGβ to the medium. In all but the digestion time assay, however, at least one cell line showed the same degree of response to purified HCGβ as to autocrine expression.
With this experimental approach, we examined several cellular parameters associated with increased metastasis. In order for a cell to metastasize from a primary tumor, it needs to reduce cell–cell interactions,17 increase its migratory capacity, and show an ability to digest its way through extracellular matrix. Reduced cell–cell interactions have been demonstrated by the increased ease with which trypsin-EDTA solutions were able to bring about cell dispersal in the DU145 HCGβ and PC3 HCGβ versus control cells. This increased ease of dispersal was accompanied by a decreased intracellular level of E-cadherin. Studies from several laboratories have shown that E-cadherin suppresses tumor invasion and metastasis.10, 16, 18 Furthermore, lower expression of E-cadherin is associated with tumor invasion and metastasis,19 an effect that can be reversed by transfecting invasive cells with E-cadherin.20 Conversely, transfection with E-cadherin antisense RNA increases cellular invasiveness.18 In addition to making homophilic interactions, E-cadherin also interacts with integrins—in particular, integrin αEβ7,21 and hence reduced E-cadherin may also contribute to reduced interactions between cells and the extracellular matrix. Through its cytoplasmic tail, E-cadherin interacts via the catenins22 with the cytoskeleton, the arrangement of which is responsible for normal cell shape. A reduction in E-cadherin can therefore contribute to a change in cell shape. In this study, we observed an altered cell shape in the DU145 HCGβ cells versus the controls and that removal of HCGβ from the culture medium reverted the cell shape and upregulated expression of E-cadherin. The altered shape of the DU145 HCGβ cells was consistent with the idea that the cells were becoming more mobile. The importance of the E-cadherin–catenin interaction to cell shape and tumor metastasis has been well established.16, 22–28
For HCGβ to exert a biologic effect equated with HCG, it has to be combined with an alpha subunit before interaction with its receptor.1 In fact, HCGβ alone has been considered free of biologic activity to the point where it has been suggested for use as an inert experimental marker of tumor growth (Stratagene, literature). The current studies make it clear that HCGβ alone has marked effects on prostate carcinoma cells. While it was possible that the tumor cells could cosynthesize an alpha subunit, the rapid duplication of effects by purified HCGβ on control cells in the migration assay argues against this possibility. An effect of HCGβ alone has been previously described for bladder carcinoma cells where the HCGβ was shown to have antiapoptotic effects.29
While we have confined our study to two prostate carcinoma cells lines, evidence for tissue expression of HCGβ exists for bladder, renal, gastrointestinal, neuroendocrine, lung, breast, head and neck, hematologic, and nontrophoblastic gynecologic tumors.1 Our findings may therefore have broad relevance.
In conclusion, we have demonstrated that HCGβ has a direct effect on prostate carcinoma cells, causing them to alter their shape, adhesiveness, and migratory and invasive ability. These results suggest that it is the expression of HCGβ itself that accounts for the increased metastatic potential of prostate adenocarcinomas that express HCGβ.