The colonic epithelium is a dynamic tissue, undergoing continual renewal throughout adult life via a balance of cell proliferation and cell death. Tumours arising from the colonic mucosa remain a major cause of mortality globally. The majority of colorectal tumours acquire mutations in the Wnt signalling pathway as an early/initiating event in tumorigenesis.1 These mutations result in constitutive activation of canonical Wnt/β-catenin signalling and the subsequent upregulation of tumour-promoting genes such as c-myc.1 During embryonic development, the Wnt pathway frequently interacts in tightly regulated networks with other signalling pathways, such as those stimulated by the TGF-β superfamily and the Hh family of signalling proteins, to control growth and patterning.1
Expression of Hh proteins has been studied in the developing gut. Ramalho-Santos et al.2 showed, in the mouse, that by 18.5 dpc (birth occurs at around 19.5 dpc) Ihh is expressed throughout the colonic epithelium and Shh expression appears to be confined to crypt epithelial cells. Mice mutant for Ihh exhibit defective stem cell proliferation and differentiation in the intestinal epithelium,2 and the phenotype in these mice is similar to that observed for mice lacking Tcf-4.3
SHH and IHH expression in the embryonic mouse colon are confined to the epithelial component of the mucosa, whereas expression of Hh target genes, including the Ptch receptor, is located in the subepithelial mesenchyme and the smooth muscle layers.2, 4 Other embryologic studies have shown that Hh signalling from the intestinal epithelium is required for the correct specification and patterning of the underlying mesenchyme.5, 6, 7
A study in the adult human gut showed expression of SHH primarily in the fundic glands of the stomach.8Shh mRNA was detected in a small number of epithelial cells at the base of the crypts of the small and large intestine.8
Secreted SHH and IHH signal through binding to the PTCH receptor, which, in the absence of ligand, represses the activity of the transmembrane protein SMO. Release of this repression of SMO allows activation of downstream targets through the Gli transcriptional effectors (Gli1–3, mammalian homologues of the Drosophila gene Cubitus interruptus).9 Gli can then act as a transcriptional activator of the pathway target genes.9 The Hh signalling pathway controls epithelial stem cell proliferation in a variety of tissues in both Drosophila and vertebrates in a highly conserved manner.10 Mutations of PTCH and SMO (leading to constitutive activation of the pathway) have been implicated as causal factors in a number of human malignancies, including BCC, medulloblastoma, rhabdomyosarcoma, squamous cell oesophageal carcinoma and transitional cell carcinoma of the bladder.1, 11 In the developing cerebellum, SHH inhibits terminal differentiation and maintains a high proliferation rate in granule-cell precursors.1 These studies show that deregulation of the Hh signalling pathway can inhibit differentiation and promote tumorigenesis in a number of different tissues.
The effects of deregulated Hh signalling on tumour growth can be reversed by treatment with the steroidal alkaloid cyclopamine. Derived from the lily Veratrum californicum, cyclopamine inhibits cellular responses to SHH signalling.12, 13 Sheep grazing on this lily showed a high proportion of birth defects associated with disrupted Hh signalling in their offspring, without apparent harm to the adult animals. Cyclopamine blocks the abnormal cell growth associated with oncogenic mutations of Ptch and Smo in fibroblasts,14 inhibits the malignant growth of medulloblastoma cells lacking Ptch function15 and inhibits the SHH-dependent growth of small-cell lung cancer cells.16 The lack of adverse effects of cyclopamine exposure in adults thus far described makes it a possible therapeutic agent for tumours associated with deregulated Hh signalling.
Although Shh expression has been reported to be coincident with the stem cell compartment in the adult colonic crypt,8 previous studies have failed to show activating mutations of Shh, or inactivating mutations of Ptch in colorectal tumours. In the current study, we show expression of SHH and IHH in both adenoma- and carcinoma-derived cell lines and expression of PTCH, SMO and GLI1 in the same epithelial cells, suggesting the presence of an autocrine signalling mechanism. Using cyclopamine, we show that inhibition of Hh signalling leads to induction of apoptosis in colorectal adenoma and carcinoma cells, which can be partially rescued by further stimulation of the Hh pathway. These findings suggest that the Hh signalling pathway may represent a novel target for therapeutic intervention using cyclopamine.
AA/C1 and RG/C2 are adenoma-derived cell lines.17 AA/C1 was cultured in conditioned medium as described by Williams et al.17 RG/C2 was cultured in DMEM supplemented with 20% (v/v) FBS (Life Technologies, Paisley, UK). Both cell lines are anchorage-dependent and nontumorigenic in athymic nude mice. JD/FIBS is a colonic fibroblastic cell line derived from the same patient as the AA/C1 epithelial adenoma line and is cultured in DMEM supplemented with 10% (v/v) FBS.17 The colon carcinoma-derived cell lines CaCo2, HT29 and SW480 were obtained from the ATCC (Rockville, MD) and Cancer Research UK and cultured in DMEM supplemented with 10% (v/v) FBS; all cell lines were cultured as adherent monolayers in 25 cm2 tissue culture flasks (Corning Costar, High Wycombe, UK). Care was taken in the routine culture of cell stocks, to avoid cells becoming confluent, which could result in differentiation.
RNA was extracted from 5 × 106 cells using the Qiagen (Chatsworth, CA) RNeasy mini-kit with “on-column” DNase digestion. Single-stranded cDNA was prepared from 10 μg RNA in a 50 μl volume containing 100 pmoles oligo-dT, 15 units AMV reverse transcriptase (Promega, Southampton, UK) and 0.4 mM each dNTP, in appropriate buffer. Primers were designed using Lasergene primer design software (DNAStar, Madison, WI) with the exception of positive control primers to the ribosomal protein 36B4, which were described by Lorentz et al.18 cDNA (5 μl) was amplified in 100 μl PCRs containing 50 pmoles of each primer, 200 μM dNTPs and 0.5 units of Taq DNA polymerase (Roche, Lewes, UK) in the buffer supplied. PCR was carried out for 35 cycles. To show a lack of genomic carryover, control reactions were carried out on RNA samples that had not been reverse-transcribed (representative sample shown for each primer set). The primers used for Ptch span exons 20–22 of the Ptch gene. PCR fragments were sequenced to confirm their identity. PCR primer sequences used were as follows: Shh forward 5′-CGCACGGGGACAGCTCGGAAGT-3′, Shh reverse 5′-CTGCGCGGCCCTCGTAGTGC-3′ (product = 477 bp); Ihh forward 5′-GGCCGGCTTTGACTGGGTGTATTA-3′, Ihh reverse 5′-CTTTGTGAGCGGGGCGTAGG-3′ (product = 487 bp); Ptch forward 5′-CGGCGTTCTCAATGGGCTGGTTTT-3′, Ptch reverse 5′-GTGGGGCTGCTGTTTCGGGTTCG-3′ (product = 376 bp); Smo forward 5′-ACCCCGGGCTGCTGAGTGAGAAG-3′, Smo reverse 5-TGGGCCCAGGCAGAGGAGACATC-3′ (product = 562 bp);Gli1 forward 5′-TCTGCCCCCATTGCCCACTTG-3′, Gli1 reverse 5′-TACATAGCCCCCAGCCCATACCTC-3′ (product = 480 bp).
Western blot analysis
Samples of 2 × 106 cells were prepared for Western blotting as described by Williams et al.19 Antisera were obtained from Santa Cruz Biotechnolgy (Santa Cruz, CA) raised against the carboxy terminals of SHH (C-18), IHH (C-15), PTCH (H-267) and SMO (H-300). The antibody for GLI1 was obtained from Abcam (Cambridge, UK). Blots were subsequently probed with anti α-tubulin (Sigma, Poole, UK) to show equal sample loading.
Treatment with cyclopamine
Cyclopamine was obtained from V. californicum as previously described and dissolved in 95% ethanol.20 Cyclopamine treatments were carried out using DMEM-F12 (Invitrogen, Paisley, UK) supplemented with 2% FBS, and the doses were previously described.14, 21
Cells (2 × 106/flask) were seeded in triplicate flasks and allowed to grow to approximately 50% confluence prior to treatment. Cells were treated for 48 hr and trypsinized for counting and analysis. Corresponding vehicle controls were prepared by addition of 0.1% (v/v) 95% ethanol. Initial experiments, carried out with 0.1% (v/v) ethanol, showed no negative effect on cell growth in any of the cell lines used. Preliminary experiments showed that apoptosis was induced by 48 hr; therefore, this time point was used.
Assessment of apoptosis
The level of apoptosis in cultured colon cells can be assessed by measuring the proportion of cells that detach from the flask and float in the medium.22, 23 Apoptosis was confirmed in these floating cells by morphology (following acridine orange staining) and proteolytic cleavage of PARP. Acridine orange staining was performed as previously described with at least 300 cells scored for each sample.24 The proportion of cells exhibiting apoptotic morphology was >90% in the floating cell population, and this proportion remained constant following cyclopamine treatment. Therefore, the proportion of cells floating in the medium can be used as a measure of the extent of apoptosis in the culture. To further confirm the apoptotic nature of the floating cells, Western blotting was performed on samples prepared from floating cells, as described above, and blots were probed for PARP (antibody from Alexis, Geneva, Switzerland). The 116 kDa form of PARP is cleaved during apoptosis by caspase-1 and caspase-3 to yield an 85 kDa cleavage product.24
Production of recombinant SHH-N protein and luciferase reporter assays
The pSHH-N and pGli-Luc plasmids were a kind gift from Dr. P. Beachy (Johns Hopkins, Baltimore, MD). Active amino-terminal SHH (SHH-N) was produced by transient transfection of 293 cells as described previously,25 and the medium was assayed for SHH content by Western blotting.
For the Gli-luciferase reporter assay,14 8 × 105 SW480 cells were seeded per well in 6-well plates and treated with cyclopamine as above. These cells were transfected with 1.8 μg of pGli-Luc and 0.2 μg of pCMV-Renilla (Promega). Cells were also transfected with the appropriate vector control. Transfection was carried out using Transfast transfection reagent (Promega) following the manufacturer's guidelines. Following 24 hr of transfection/cyclopamine treatment, cells were lysed in “passive lysis buffer” and the dual luciferase assay (Promega) was performed. Values were corrected for background luminescence and are shown as ratios of Firefly to Renilla luciferase, to control for transfection efficiency and cell number. No significant background was observed in the empty vector controls.
Both colorectal adenoma- and carcinoma-derived cells express not only Shh and Ihh but also the essential Hh signalling pathway components Ptch, Smo and Gli1
Five human colorectal tumour and one colonic fibroblast (JD/FIBS) cell line were screened for expression of Shh and Ihh using RT-PCR. The identity of all PCR products shown was confirmed by sequencing and BLAST analysis (NCBI, NIH). Two of these lines were derived from nonmalignant adenomatous polyps (AA/C1 and RG/C2) and 3 from adenocarcinomas (CaCo2, HT29 and SW480). mRNAs for both Shh and Ihh were found in all 5 tumour cell lines but not in the colonic fibroblasts or the “no-RT” controls (Fig. 1a). Western blots using antisera raised against the carboxy terminals of SHH and IHH show that the 45 kDa precursor proteins of both proteins were expressed in both the adenoma- and the carcinoma-derived colorectal epithelial cells (Fig. 1b).
RT-PCR was then used to assay expression of the major downstream components of the Hh signalling pathway, namely, Ptch, Smo and Gli1, in these cell lines. Again, the identity of all PCR products was confirmed by sequencing and BLAST analysis. Expression of the Hh receptor Ptch and the downstream effectors Smo and Gli1 was detected in both adenoma- and carcinoma-derived cell lines as well as in colonic fibroblasts (Fig. 1a). Western blotting was again used to confirm these findings (Fig. 1b) and showed expression of all components in all of the lines assayed. Interestingly, the adenoma-derived cell lines exhibited higher levels of IHH than the carcinoma-derived cell lines but lower levels of PTCH, SMO and GLI1.
Cyclopamine treatment results in reduced cell yield and increased apoptosis in human colorectal tumour cells
Expression of both the Hh ligands and the downstream signalling components Ptch, Smo and Gli1 in colorectal tumour cells suggests autocrine Hh signalling in these epithelial cells. To determine the consequences of blocking this pathway, cells were treated with cyclopamine, a known inhibitor of Hh signalling. Treatments were carried out using 5, 10 and 20 μM cyclopamine (doses previously described14, 21) to determine whether this inhibition reduced cell growth and induced apoptotic cell death.
We and others have shown that in colorectal epithelial cell culture the majority of cells that detach from the adherent monolayer and float in the medium have undergone apoptosis (see Material and Methods).22, 23 Because of rapid detachment of cells entering apoptosis, very few apoptotic cells were seen in the attached population. The proportion of cells floating in the medium that are apoptotic remains constant and can be used as a measure of the extent of apoptosis in culture.
Cyclopamine treatment resulted in decreased cell yields and induction of apoptosis in each of the cell lines investigated at all of the tested doses (Fig. 2). Thus, cells derived from both colorectal adenomas and carcinomas are sensitive to the growth-inhibitory effects of this agent.
Following cyclopamine treatment, apoptosis was confirmed in floating cells by studying the proteolytic cleavage of PARP (as shown by Western blotting, Fig. 2b) and identification of apoptotic morphology following acridine orange staining, as described previously24 (data not shown). Figure 2b clearly shows the presence of full-length (116 kDa) PARP in the attached cells but only the 85 kDa cleaved form in the detached, apoptotic population. Following acridine orange staining, a minimum of 90% of floating cells in control cultures were identified as apoptotic and this proportion remained unchanged following cyclopamine treatment. No increase in apoptosis was observed in the attached cell population (<3%), and no evidence of necrotic cell death was observed, suggesting that the cyclopamine doses used were not toxic. The basal level of apoptosis varies between cell lines, and RG/C2 in particular demonstrates higher spontaneous levels than the other cell lines used, as reported previously.24
We also tested whether adding an excess of ligand to stimulate the Hh signalling pathway could reduce cyclopamine-induced apoptosis. HEK293 cells were transiently transfected with a vector coding for the active amino-terminal SHH protein (SHH-N) as previously described, to produce SHH-N-rich conditioned medium.25 Culturing cells in SHH-N-containing medium did not result in stimulation of cell growth under the conditions used (data not shown). However, the presence of SHH-N resulted in a significant reduction in cyclopamine-induced apoptosis compared to cells cultured in medium from empty vector-transfected HEK293 (Fig. 3a). This finding suggests that cyclopamine-induced apoptosis could be rescued, at least in part, through stimulation of Hh signalling.
To confirm the inhibition of autocrine Hh signalling by cyclopamine in colorectal cancer cells, we used a Gli-dependent luciferase reporter. Figure 3b shows a dose-dependent reduction in Gli reporter activity in response to cyclopamine treatment in SW480. Furthermore, as PTCH is a target of Hh signalling, we would expect it to be downregulated following cyclopamine treatment. Figure 3c shows a clear downregulation of PTCH in the 3 carcinoma cell lines. Regulation of PTCH in the adenoma cell lines AA/C1 and RG/C2 (which have much lower endogenous levels of PTCH) was small but reproducible in 3 separate experiments. Expression of the other components of the Hh signalling pathway was unaffected by cyclopamine treatment (data not shown).
The Hh signalling pathway plays a key role in regulating cell proliferation and survival in a range of tissues and has been implicated in aberrant cell survival in a number of human malignancies.1 Expression of SHH and IHH in the colon of the mouse embryo is confined to the epithelium, whereas expression of Hh target genes (including the Ptch receptor) is located in the subepithelial mesenchyme and the smooth muscle layers.2, 4 Studies have shown that Hh signalling from the epithelium is required for the correct specification and patterning of the underlying mesenchyme.5, 6, 7 In the adult human colon, expression of Shh mRNA has been demonstrated in a few epithelial cells at the base of normal intestinal crypts,8 suggesting a role for SHH in the maintenance of the stem cell compartment. Further evidence of a role for Hh signalling in intestinal epithelial cell proliferation comes from targeted disruption of Ihh in the mouse. IHH is required for normal proliferation of the intestinal epithelium,2 and the phenotype in Ihh mutants was similar to that observed in Tcf4 mutant mice.3 Targeted disruption of Tcf4 is neonatally lethal, and the animals show no proliferative compartments in the prospective crypt regions. As a consequence, the neonatal epithelium is composed entirely of differentiated, nondividing cells.3
In this study, we demonstrate expression of SHH and IHH in cell lines derived from both nonmalignant adenomatous polyps and adenocarcinomas of the colorectal epithelium. These cells express not only the ligands but, in contrast to the embryonic mouse colon, also the receptor (Ptch) and downstream components (Smo and Gli1) necessary to transduce the Hh signal. This finding suggests that these tumour cells are capable of autocrine Hh signalling. Although activating mutations of SHH signalling have not been detected in colorectal tumours,26, 27 our data suggest that the pathway could be erroneously activated by expression of the receptor and signalling components in the epithelium. Interestingly, the lack of Hh ligand expression coupled with the presence of PTCH, SMO and GLI1 in the colon fibroblast cell line JD/FIBS concurs with the in vivo expression patterns observed in the mouse embryo.2, 4
The adenoma-derived cell lines exhibited higher levels of IHH expression than the carcinoma-derived cell lines, which is contrasted by lower expression levels of PTCH, SMO and GLI1. Although these differences are not reflected in the response of the cells to cyclopamine treatment, they may indicate a change in the role of Hh signalling during tumour progression to invasive carcinoma.
Cyclopamine, a known inhibitor of the Hh pathway,12, 13, 14, 28 was used to test whether Hh signalling influences colorectal tumour cell survival. Treatment of both adenoma- and carcinoma-derived colorectal epithelial cells with cyclopamine reduced cell yield and induced apoptotic cell death.
The CaCo2 cell line showed particular sensitivity to cyclopamine-induced apoptosis. Although the precise nature of this sensitivity is unknown, it may indicate a heavier reliance on Hh signalling for survival. This cell line may represent a subset of colorectal tumours that would respond well to the therapeutic use of cyclopamine. Previous studies have shown that cyclopamine induces apoptosis in small-cell lung cancer cell lines16 and rapid death of cells from freshly resected medulloblastomas.28
Addition of exogenous SHH-N protein reduced cyclopamine-induced apoptosis, suggesting that the cells could be rescued from apoptosis by further stimulation of the Hh pathway. The fact that cyclopamine can induce apoptosis that is partially blocked by SHH suggests that aberrant expression of the Hh signalling components in colorectal tumour–derived cells is important for cell survival. Further evidence for the specificity of cyclopamine action was demonstrated by its ability to reduce the expression of PTCH and to decrease the activity of a Gli-dependent luciferase reporter.
Whilst this report was under revision, Berman et al.28 published a study of Hh signalling in tumours throughout the gastrointestinal tract. These authors also found expression of Shh and Ihh mRNA in a panel of colorectal carcinoma–derived cell lines in concurrence with our findings. They also found expression of Gli in 4/11 lines but did not detect expression of Ptch. We have consistently been able to detect Ptch, Smo and Gli1 in the colorectal cell lines used in this study in 5 separate RT-PCRs and by Western blotting. The reason for this difference is not clear, but expression of Ptch may correlate with a subset of tumours that are particularly sensitive (e.g., CaCo2 cells) to treatment with cyclopamine. These observations may be important when selecting patients to determine the therapeutic potential of Hh signalling inhibitors.
In conclusion, these data provide evidence of expression not only of the Hh ligands but also of the Hh signalling pathway in human colorectal tumour cells. The induction of apoptosis by cyclopamine, a known inhibitor of Hh signalling, suggests a role for autocrine Hh signalling in colorectal tumour cell survival. These data suggest that this pathway may be an important novel target for colon cancer therapy using agents such as cyclopamine.