Background
Invasive growth is a critical step in the progression of tumorigenesis as it is what distinguishes a malignant from a benign tumor [
1]. A tumor’s ability to disseminate, invade and migrate to distant tissues correlates with worse prognosis [
2]. The edge of an invasive tumor is characterized by the loss of apico-basal polarity along with a loss of cell-cell junctions and decreased E-cadherin expression. The actin cytoskeleton is reorganized with the formation of F-actin rich protrusions (FRP) at the leading edge of an invasive tumor, where the cell changes from a cuboidal shape to a motile spindle shape [
3]. The cell motility pathways such as those controlled by the integrin receptors, the focal adhesion kinase (FAK), the Rho and Rac family of small G-proteins, and the metalloproteases (MMPs) are also activated in the invasive tumor cells [
4]. Histology of colon tumor samples has shown that some of these characteristics, i.e. change in shape and loss of E-cadherin, are found only at the leading edge of the tumor in cells that have direct contact with the ECM, while cells fully encased in the solid tumor maintain expression of E-cadherin [
5]. It thus appears that the invasive phenotype might occur in individual cells responding to the external cues rather than the entire tumor mass undergoing global changes.
The recent TCGA (The Cancer Genome Atlas) analysis of human colorectal cancer (CRC) has established that the Wnt and the TGF-β (BMP) pathways are consistently up or down regulated, respectively, by genetic and epigenetic mechanisms in 97% and 87% of CRC in the hypermutated group [
6]. The Wnt pathway is also upregulated in 92% of CRC in the non-hypermutated group [
6]. This finding is consistent with the fact that maintenance of the intestinal crypt stem cells requires full activation of the Wnt pathway and inactivation of the BMP pathway by the anti-BMP ligand Noggin [
7]. In the intestinal crypt compartment, binding of locally produced Wnt and R-Spondin to their respective seven transmembrane-serpetine receptors, Frizzled and Lgr4/5, leads to the assembly of a Wnt signaling complex involving the recruitment of another membrane receptor, LRP, and the stabilization of cytoplasmic β-catenin [
8]. The accumulation of cytoplasmic β-catenin is a pre-request for its nuclear translocation, which is regulated by a variety of factors, as β-catenin itself does not contain any nuclear localization signals [
9]. Nuclear β-catenin associates with the TCF-family of transcription factors to stimulate gene expression that promotes cell cycle progression and inhibits apoptosis [
8]. In the normal regenerating intestinal tissue, Wnt and Noggin levels are high at the base of the crypt to stimulate proliferation and inhibit differentiation. The concentrations of these factors are reduced in the villi, where Wnt and Noggin levels are low and BMP levels are high, promoting differentiation [
10]. With the constitutive activation of the Wnt and the receptor tyrosine kinase (RTK) pathways as well as the downregulation of the TGF-β pathway, colon cancer cells do not require this complement of factors to proliferate.
In this study, we show that established colon cancer cells remain responsive to the stimulation of a complement of crypt growth factors to undergo a reversible and localized invasive phenotype but only in 3-D cultures. This invasive response requires activation of β-catenin and EGFR and can be inhibited by drugs that interfere with the function of downstream effectors such as ABL or AKT.
Methods
Antibodies and reagents
Anti-β-catenin (610153), and anti-EGFR (610016) were from BD Biosciences. Anti-GAPDH (MAB374), anti-active-β-catenin (05–665), and anti-phospho-FAK (44625G) were from Millipore. Anti-Akt (9272), anti-phospho-Akt (9271), anti-E-cadherin (3195), anti-phospho-Abl (2861), anti-phospho-EGFR (4407), and horseradish peroxidase (HRP)-conjugated secondary antibodies were purchased from Cell Signaling Technology. Anti-FAK (05537) and TRITC conjugated phalloidin (12381) were purchased from Invitrogen. Anti-vimentin (01191) was purchased from GenScript. Anti-Abl 8E9 was generated in our laboratory. The peptides EGF (100–15) and Noggin (250–38) were purchased from Peprotech. Conditioned media was collected from 293 cells stably overexpressing either Wnt3a or R-Spondin1 (a generous gift from Dr. Karl Willert at UCSD) according to [
11] using serum free media.
Cell culture
The human colon cancer cell lines HCT-116 and HT29 (ATCC) were maintained in DMEM medium supplemented with 10% FBS (HyClone). The cell lines LIM1215, 1899, and 2551 were maintained in RPMI 1640 medium (Invitrogen) supplemented with 10% FBS and Additives (10 μM Thioglycerol (Sigma), 2.5 ug/ml Insulin (Sigma) and 0.5 mg Hydrocortisone) [
12]. All cell lines were initially maintained in 2-D plastic tissue culture dishes at 37°C with 5% CO
2. For seeding in 3-D, cells were washed with PBS and trypsonized to detach from each other and the plate. Between 500–1000 cells were seeded in a 24-well plate embedded in 50 μl of 100% matrigel (BD Biosciences). Each well then received 500 μl of DMEM/F12 media supplemented with 1% Pen/Strep (Cellgro), 1M HEPES (Gibco), and Glutamax (Gibco). Growth factors (EGF, Noggin, Wnt3a condition media, and R-Spondin1 condition media) were then individually added to each well. Media was changed every 2 days for a total of 6 days, at which time colonies were passaged.
Immunofluorescence and confocal microscopy
Cells were grown as described above, fixed with 3% paraformaldehyde for 20 mins. at room temperature and stained according to [
13]. Images were captured using an Olympus FV1000 scanning laser confocal microscope.
Immunoblotting
Proteins from the cell lines were extracted in RIPA buffer (25 mM Tris–HCl pH 7.4, 1 mM EDTA, 0.1% SDS, 150 mM NaCl, 1% NP-40, 1% Sodium Deoxycholate, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitor cocktail) and measured by Lowry protein assay. Equal amounts (50 μg) of total proteins were loaded on SDS-PAGE, transferred onto a nitrocellulose membrane, and probed with primary antibodies overnight at 4°C. HRP conjugated secondary antibodies were incubated for 1 hour at room temperature. Proteins were visualized by chemiluminescence as recommended by the manufacturer (Thermo).
Luciferase assay
Cells were transfected with β-galactosidase and either BRE (gift from Peter ten Dijke) or TopFlash luciferase plasmid using Genetran (Biomiga) according to manufacturer’s protocol. Twenty four hours later, cells were lysed with Cell Culture Lysis Reagent (Promega) and the luciferase substrate (Promega) was added at a 5:1 dilution. Luminescent values were determined by Monolight 3010 Luminometer. β-galactosidase assay was performed on 96 well plate using ONPG substrate solution (Sigma) and 10μL of cell lysate in each well. Absorbance values were read at 420nm. Luciferase assay was the normalized to β-galactosidase readings.
Statistical analysis
Data are represented as mean and SEM (Standard Error of the Mean). Two-tailed unpaired t-test was used to determine statistical significance of the differences between data sets. p < 0.05 was considered statistically significant.
Conclusions
Our results show that established colon cancer cell lines can be cultured in 3-D matrigel, and like their original source, human CRC, do not require the presence of EGF, R-Spondin1, Wnt3a, or Noggin for proliferation and long term expansion. However these cells did remain responsive to crypt growth factors by taking on a disc-like morphology when grown in the presence of EGF, Wnt, R-spondin1, and Noggin (Figures
1,
2). Chemical or genetic perturbation of the EGFR or the β-catenin pathway revealed that RNEW not only activated these oncoproteins (Figure
3), but that both were required for disc formation. We found that these growth factors also activated ABL and AKT kinases and inhibition of either pathway could prevent disc colony formation (Figure
4). This growth factor-induced disc morphology correlated with localized and reversible invasive characteristics, however it was only seen when cells were grown in 3-D conditions (Figures
5,
6). Given that the amount of disc colony formation with RNEW was an additive but not a synergistic effect indicates that the activation of all three pathways (β-catenin/Wnt, EGFR, and BMP) may function independently, as opposed to converging on a common target.
Interestingly, RNEW was not able to increase proliferation or stimulate actin re-organization when HCT-116 cells were grown in 2-D. This observation is consistent with breast cell lines like MCF10A, which, when grown in 2-D conditions remain as unpolarized epithelial cells; however under 3-D conditions form acini and undergo differentiation and polarization [
13]. Additional file
2: Figure S2 shows that when HCT-116 cells are plated on matrigel coated coverslips RNEW is still not able to induce an invasive phenotype. This suggests that matrigel per se is not the collaborating factor for RNEW to induce the cytoskeletal transformation. An obvious difference between 2-D and 3-D tissue culture conditions is the stiffness of the culture substratum. The stiffness of healthy intestinal tissue is between 20–40 kPa [
26,
27], while standard plastic tissue culture dishes are around 10,000 kPa [
28]. Another difference between 2-D and 3-D growth conditions is the area of cell contact with the ECM. When cells were grown on either plastic or matrigel coated slides only a fraction of the cell membrane was in direct contact with the matrix. However, in 3-D a significantly larger portion of the cell surface was engaged in ECM interactions. Increased cell surface contact with the ECM might alter the signaling activation of growth factors receptors as has been reported [
29], and stimulate invasion of cells on the edges of the colonies.
Colon cancer presents as two different morphologies; the of stalked (pedunculated) type, and the flattened disc type [
31]. While the majority of colon cancers are of the pedunculated morphology which are easy to detect by colonoscopy [
30], between 10-20% of colon cancers are of the disc type [
31,
32]. This form of cancer is very difficult to detect, corresponds with a worse prognosis [
32,
33], and has been postulated to be up to 10 times more malignant than the pedunculated type [
33]. The genetics for these tumor sub-types are not yet completely understood as some groups have found that disc tumors have increased nuclear β-catenin and no KRAS or BRAF mutations [
32], while other groups have found the opposite [
33]. What is known is that these tumors remain disc throughout the entirety of the tumorigenesis process and do not follow the traditional Vogelstein model of colon carcinogenesis [
32]. Our finding suggests that the tumor phenotypes can be influenced not only by the genetic mutations but also by the availability and the tumor response to crypt growth factors. Using the method of 3-D culture with RNEW developed in this study, we were able to identify a previously unreported role for imatinib and MK-2206 in inhibition of growth factor-induced invasive phenotype. This suggests that this methodology may be used to study the effects of other inducers and/or inhibitors of tumor invasion.
Author contributions
KL carried out the studies, and wrote the manuscript, EST carried out the luciferase assays, and JYJW conceived of the study, participated in its design, and revised the manuscript. All authors read and approved the final manuscript.
Acknowledgments
We thank Dr. Tony Burgess at the Ludwig Cancer Research Institute in Melbourne, Australia for the generous gifts of the LIM1215, 1899, and 2551 cells [
12], Dr. Karl Willert for the 293 Wnt3a and R-Spondin1 cells [
34], and the UCSD Neuroscience Microscopy Facility. KL was supported by the NIH Cancer Cell Biology Training Grant (2T32CA067754-16A1). This work was supported by an NIH RO1 (CA058320) to JYJW.
Competing interests
The authors declare no conflicts of interest.