How is the Hedgehog pathway activated?

Journal Article

Jason K. Sicklick,

* To whom correspondence should be addressed at: Duke University Medical Center, Division of Gastroenterology, Snyderman-GSRB I Suite 1073, Box 3256, Durham, NC 27710, USA. Tel: +1 919 684 4173; Fax: +1 919 684 4183; Email:

Search for other works by this author on:

Yin-Xiong Li,

Search for other works by this author on:

Aruna Jayaraman,

Search for other works by this author on:

Rajesh Kannangai,

Search for other works by this author on:

Yi Qi,

Search for other works by this author on:

Perumal Vivekanandan,

Search for other works by this author on:

John W. Ludlow,

Search for other works by this author on:

Kouros Owzar,

Search for other works by this author on:

Wei Chen,

Search for other works by this author on:

Michael S. Torbenson,

Search for other works by this author on:

... Show more

Received:

21 September 2005

Revision received:

26 October 2005

Accepted:

22 November 2005

Published:

08 December 2005

• PDF
• Split View
• • Article contents
  • Figures & tables
  • Video
  • Audio
  • Supplementary Data

• Cite

  Cite

  Jason K. Sicklick, Yin-Xiong Li, Aruna Jayaraman, Rajesh Kannangai, Yi Qi, Perumal Vivekanandan, John W. Ludlow, Kouros Owzar, Wei Chen, Michael S. Torbenson, Anna Mae Diehl, Dysregulation of the Hedgehog pathway in human hepatocarcinogenesis, Carcinogenesis, Volume 27, Issue 4, April 2006, Pages 748–757, //doi.org/10.1093/carcin/bgi292

  Close

• • Email
  • Twitter
  • Facebook
  • More

Close

Navbar Search Filter Microsite Search Term Search

Abstract

Hedgehog [Hh] pathway activation promotes tumors in several endodermally derived tissues, but its role in the pathogenesis of hepatocellular carcinoma [HCC] is unknown. Although normal hepatocytes lack Hh signaling, activation of the Hh pathway in endodermal progenitors is required for liver development. Thus, we hypothesized that hepatocarcinogenesis may involve regulation of Hh signaling. This pathway is activated when Hh ligand binds to its receptor, Patched [PTC]. In an unoccupied state, PTC normally functions as a tumor suppressor that inhibits Smoothened [SMO], a proto-oncoprotein, from activating downstream components and transcription of target genes. Here we show that in HCCs, overexpression of the Smo proto-oncogene, as well as an increase in the stoichiometric ratio of Smo to Ptc mRNA levels, correlated with tumor size, a prognostic indicator in HCC biology. In one tumor we identified a novel Smo mutation in an evolutionarily conserved residue. We also demonstrated that HCC cell lines [HepG2 and Hep3B] expressed Hh pathway components and activated Hh transcriptional targets. In Hep3B cells, cyclopamine, an inhibitor of wild-type SMO, had no effect, but KAAD-cyclopamine, a blocker of oncogenic SMO, inhibited Hh signaling activity by 50%, decreased expression of the hepatocarcinogenic oncogene, c-myc , by 8-fold, and inhibited the growth rate of Hep3B cells by 94%. These data support our hypothesis that Hh signaling is dysregulated in human hepatocarcinogenesis. We demonstrate that overexpression and/or tumorigenic activation of the Smo proto-oncogene mediates c-myc overexpression which plays a critical role in hepatocarcinogenesis and suggests that Smo is a prognostic factor in HCC tumorigenesis.

ρ, Spearman's rank correlation, Afp, alpha-fetoprotein, Cyc, Cyclopamine, Gus, β-glucuronidase, HCC, hepatocellular carcinoma, Hh, Hedgehog, Ihh, Indian hedgehog, KAAD-cyclopamine or KAAD-Cyc, 3-Keto- N -[aminoethyl-aminocaproyl-dihydrocinnamoyl]-cyclopamine , mRNA, messenger RNA, PTC, Patched, RT–PCR, reverse transcription polymerase chain reaction, SMO, Smoothened, Shh, Sonic hedgehog, Tom, Tomatidine

Introduction

Dysregulation of the Hedgehog [Hh] pathway has been implicated in the genesis of cancers that are derived from multiple tissue types [ 1 ]. Along with studies of other developmentally regulated signaling pathways, such as Wnt, these findings have added to a growing body of evidence for the stem cell theory of cancer, which holds that tumors, like normal tissues, are generated by a small number of self-renewing stem cells [ 1 ]. From embryogenesis to adulthood, skin and gastrointestinal progenitors are regulated by Hh signaling [ 2 – 4 ]. This pathway is activated when Sonic hedgehog [SHH] or Indian hedgehog [IHH] ligand bind to their receptor, Patched [PTC]. When unoccupied by ligand, PTC is a tumor suppressor that binds and represses Smoothened [SMO] [ 5 ], preventing the SMO proto-oncoprotein from activating downstream transcription factors, such as GLI1. Conversely, when ligand binds to PTC, SMO is released and GLI1 is activated, resulting in the transcription of target genes including Ptc and Gli1 [ 5 ].

During health and disease Hh signaling is now known to play critical roles in the gastrointestinal tract. During embryogenesis, defective or absent Hh signaling has been implicated in the development of tracheoesphageal fistula [ 6 , 7 ], annular pancreas [ 7 , 8 ], gut malrotation [ 7 ] and imperforate anus [ 7 ]. However, the role of the Hh pathway is not limited to prenatal development. Recent work has demonstrated that Hh signaling is also critical for normal post-natal fundic gland differentiation in the stomach [ 9 ] and patterning of the crypt–villus axis in the colon [ 4 ]. Hh signaling is also altered in several gastrointestinal diseases of adults including chronic pancreatitis [ 10 ], as well as Barrett's oesophagus, gastritis, Crohn's disease and ulcerative colitits [ 11 ]. Thus, Hh signaling remains a critical pathway in the gut throughout life.

The effects of Hh signaling are more widespread that just within the gastrointestinal tract. This pathway is crucial for the morphogenesis of several other organs, including the skin and nervous system [ 12 ]. Given this fact, it is not surprising that overactivation of the Hh pathway underlies the nevoid basal cell carcinoma syndrome [Gorlin's syndrome], which is characterized by numerous basal cell carcinomas, medulloblastomas and rhabdomyosarcomas [ 13 ]. Gorlin's syndrome results from homozygous mutations in the tumor-suppressor gene, Ptc . Point mutations in the proto-oncogene, Smo , have also been implicated in the formation of sporadic basal cell carcinomas [ 14 ]. In addition to cancers that result from inherited or acquired mutations in Hh signaling components, other tumors, including esophageal, gastric, pancreatic and biliary cancers, have been associated with excessive expression of Shh and Ihh ligands that promote Hh pathway activation [ 15 , 16 ].

Although Hh activity clearly modulates tissue homeostasis and regeneration in many foregut-derived adult tissues, this pathway is not considered to retain function in the adult liver because mature hepatocytes lack Hh pathway activity [ 15 ], despite the liver's requirement for Hh signaling during embryogenesis [ 17 ]. The pivotal role of Hh in liver development is proven by evidence that induction of Shh promotes hepatogenesis, whereas pancreatic differentiation ensues in the absence of Shh [ 17 ]. The latter finding suggests that the liver and pancreas are derived from a common, Hh-responsive endodermal progenitor. If Hh regulates progenitors in post-natal livers, the Hh pathway may play an overlooked role in the formation of liver cancers. This possibility is further suggested by our recent finding that hepatic stellate cells, which reside in the mesenchyme of adult livers, produce Hh ligands [ 18 ]. Herein, we evaluate our hypothesis that Hh signaling regulates hepatocarcinogenesis by assessing Hh pathway expression and function in cultured hepatocellular carcinoma [HCC] cell lines, comparing Hh pathway expression in non-neoplastic and malignant human livers, and correlating the expression of Hh pathway components with human HCC biology.

Materials and methods

Animal care

Adult, male Ptc-lacZ reporter mice were obtained from Dr P.A. Beachy [Johns Hopkins University, Baltimore, MD]. Animal experiments fulfilled NIH, Johns Hopkins University and Duke University requirements for humane animal care.

Ptc-lacZ staining and reporter assay

We studied mice in which one allele of Ptc is replaced in-frame with the β- galactosidase gene by homologous recombination in order to evaluate Hh signaling in the liver. As Ptc is a transcriptional target of the GLI proteins, expression of β-galactosidase indicates activation of the Hh pathway [ 19 , 20 ]. Staining and quantification of reporter expression were performed as described previously using the β-galactosidase Detection Kit [Promega, Madison, WI] [ 21 ].

Culture of cell lines

HepG2, Hep3B and C3H10T½ cell lines were purchased from American Type Culture Collection [Manassas, VA] and cultured according to their instructions. The HCT116 cell line was purchased from the Duke University Cancer Center Tissue Culture Facility [Durham, NC] and cultured according to supplier instructions.

Isolation of hepatocytes

Donated human livers, not suitable for orthotopic liver transplantation, were obtained from federally designated organ procurement organizations. Informed consent was obtained from next of kin for use of the livers for research purposes. The portal vein and/or the hepatic artery were cannulated and the organ perfused with EGTA-containing buffer for 15 min followed by digestion with 125 mg/l Liberase [Roche, Nutley, NJ], a highly purified form of collagenase, for 30 min at 34°C. Following enzymatic digestion of the liver, Glisson's capsule was serrated and the cells were mechanically separated from the vascular tree. The resulting cell suspension was then passed through 1000, 500 and 150 μm filters. The collected cells were then separated using Percoll-density centrifugation and human hepatocytes were isolated with a purity of 90–95% [ 22 , 23 ]. Isolation of primary murine hepatocytes was performed as described previously [ 24 ]. Freshly isolated human and murine hepatocytes were subsequently used for RNA analysis or Ptc-lacZ reporter analysis, respectively.

Pharmacological regulation of Hh signaling

The Hep3B line was treated with regulators of the Hh signaling pathway in a dose- and time-dependent fashion. Cultures were treated with mouse IgG 1 isotype control antibody reconstituted in sterile phosphate-buffered saline [PBS] with 1% bovine serum albumin [BSA] as per manufacturer instructions [R&D Systems, Minneapolis, MN] or 5E1 Hh neutralizing antibody in PBS [University of Iowa Developmental Studies Hybridoma Bank, Iowa City, IO] at concentrations of 0.1–10 μg/ml [ 15 , 25 ]. The lines were also treated with the pharmacological Hh inhibitors, cyclopamine [Cyc, Calbiochem, San Diego, CA], KAAD-cyclopamine [KAAD-Cyc, Toronto Research Chemicals, Canada] or their catalytically inactive analog, tomatidine [Tom, 0.03–3.0 μM; Calbiochem] dissolved in sterile DMSO [dimethyl sulfoxide] as per manufacturer instructions [ 15 , 26 – 28 ]. For all experiments, Cyc- and KAAD-Cyc-treated groups were compared to Tom-treated controls.

Cell counting assay

Cell viability was measured with the Cell Counting Kit-8 [Dojindo Molecular Technologies, Gaithersburg, MD] in replicate experiments [ N = 3–4 per group]. The Hep3B line was passaged, plated at a density of 5000 cells per well, cultured for 24 h, and then treated with reagent medium or appropriate control medium for up to 96 h. Cells were then incubated with tetrazolium reagent for 1 h. In viable cells, the tetrazolium salt is metabolized by mitochondrial dehydrogenase to a colorimetric dye and cell number is proportional to the absorbance intensity at 450 nm [ 29 , 30 ]. As detailed in the information that the manufacturer provides with the reagent, the use of tetrazolium dyes has been validated against thymidine incorporation for a variety of cell types [ 31 – 33 ], as well as employed for studies of cell viability in liver cancer cell lines [ 34 ].

Hh-responsive luciferase reporter assay

The Hh-responsive luciferase reporter assay was performed on replicate cultures [ N = 7–9] of the Hep3B cell line, an Hh-responsive, positive control cell line [C3H10T½] [ 35 ], and an Hh-unresponsive, negative control cell line [HCT116] [ 15 ], as described previously [ 35 ]. In each experiment, luciferase activity was evaluated in 3–4 wells. Each experiment was also replicated 2–3 times on separate days. Therefore, the data shown are the mean ± SD of 7–9 replicate assays for each group. Briefly, the lines were grown to near confluence and then transfected with 9× Gli-binding site-luciferase plasmid and pRL-TK [Promega]. Additionally, some cultures were also transfected with vector for constitutively active, wild-type Smo using Lipofectamine 2000 [Invitrogen, Carlsbad, CA] according to the manufacturers' recommendations. After a 3.5 h transfection, cells were washed twice with DMEM:Ham's F-12 [1:1] medium and then cultured overnight in DMEM:Ham's F-12 [1:1] medium. Cells were harvested after 16–48 h and lysed in reporter lysis buffer [Promega]. Reporter activity was determined by using the Dual-Luciferase Reporter Assay System [Promega]. Activity of the Firefly luciferase reporter was normalized to the activity of a Renilla luciferase internal control for transfection efficiency.

Human liver tissues

Human studies were performed after obtaining approval from the Johns Hopkins University Institutional Review Board. Fourteen HCCs and adjacent non-neoplastic livers were harvested at the time of liver resection/explantation and were snap frozen in liquid nitrogen. For all tissues, the histological diagnoses were confirmed under light microscopy by an experienced liver pathologist.

Two-step real-time RT–PCR

Two-step real-time RT–PCR was performed to compare the expression of Hh pathway components in primary human hepatocytes, HCC lines, as well as HCC and non-neoplastic liver tissues from 14 patients. Total RNA was extracted from cells with RNeasy kits followed by RNase-free DNase I treatment [Qiagen, Valencia, CA]. Reverse transcription to cDNA templates was performed using Ready-To-Go You-Prime First-Strand Beads [Amersham, Piscataway, NJ] with pd[N]6 First-strand cDNA primers [Amersham]. The primers were employed as described previously [ 15 , 36 – 43 ] or were designed using Genbank sequences [Supplementary Table I ]. For quantitative RT–PCR, 0.5–1 μl of the first-strand reaction was amplified using iQ-SYBR Green Supermix [Bio-Rad, Hercules, CA], an iCycler iQ Real-Time Detection System [Bio-Rad], and the specific oligonucleotide primers for target sequences as well as the β-glucuronidase [ Gus ] housekeeping gene in triplicate [ 44 , 45 ]. Target gene levels in treated cells or tumor tissues are presented as a ratio to levels detected in the corresponding control cells or patients' non-neoplastic livers, respectively, according to the ΔΔCt method [ 46 ]. These fold changes were determined using point and interval estimates. Products were separated by electrophoresis on a 2.0% agarose gel buffered with 0.5× TBE.

Table I.

Demographics, underlying diseases and tumor related factors for the cohort

Number [ N = 14] Percentage

Gender
Female	9	64.3
Male	5	35.7
Race
White	10	71.4
Other	4	28.6
Age [years]
Mean	61.3 ± 9.5
Median	61
Range	40–77
Underlying disease
No	4	28.6
Yes	10	71.4
Alcohol	1	7.1
Viral	7	50
Hepatitis B virus	2	14.3
Hepatitis C virus	6	42.9
Hepatitis B and C viruses	1	7.1
Cryptogenic	3	21.4
Serum AFP [ng/ml]
Mean	6.4 ± 4.3
Range	0–16
Tumor size [cm]
Mean	4.06 ± 2.48
Median	4
Range	1–11

Number [ N = 14] Percentage

Gender
Female	9	64.3
Male	5	35.7
Race
White	10	71.4
Other	4	28.6
Age [years]
Mean	61.3 ± 9.5
Median	61
Range	40–77
Underlying disease
No	4	28.6
Yes	10	71.4
Alcohol	1	7.1
Viral	7	50
Hepatitis B virus	2	14.3
Hepatitis C virus	6	42.9
Hepatitis B and C viruses	1	7.1
Cryptogenic	3	21.4
Serum AFP [ng/ml]
Mean	6.4 ± 4.3
Range	0–16
Tumor size [cm]
Mean	4.06 ± 2.48
Median	4
Range	1–11

Table I.

Demographics, underlying diseases and tumor related factors for the cohort

Number [ N = 14] Percentage

Gender
Female	9	64.3
Male	5	35.7
Race
White	10	71.4
Other	4	28.6
Age [years]
Mean	61.3 ± 9.5
Median	61
Range	40–77
Underlying disease
No	4	28.6
Yes	10	71.4
Alcohol	1	7.1
Viral	7	50
Hepatitis B virus	2	14.3
Hepatitis C virus	6	42.9
Hepatitis B and C viruses	1	7.1
Cryptogenic	3	21.4
Serum AFP [ng/ml]
Mean	6.4 ± 4.3
Range	0–16
Tumor size [cm]
Mean	4.06 ± 2.48
Median	4
Range	1–11

Number [ N = 14] Percentage

Gender
Female	9	64.3
Male	5	35.7
Race
White	10	71.4
Other	4	28.6
Age [years]
Mean	61.3 ± 9.5
Median	61
Range	40–77
Underlying disease
No	4	28.6
Yes	10	71.4
Alcohol	1	7.1
Viral	7	50
Hepatitis B virus	2	14.3
Hepatitis C virus	6	42.9
Hepatitis B and C viruses	1	7.1
Cryptogenic	3	21.4
Serum AFP [ng/ml]
Mean	6.4 ± 4.3
Range	0–16
Tumor size [cm]
Mean	4.06 ± 2.48
Median	4
Range	1–11

Genetic analysis of Smo

DNA was isolated from the HCC lines and human liver samples using the Get pureDNA Kit-Cell, Tissue [Dojindo Molecular Technologies]. Following PCR amplification as described previously [ 14 , 47 ], amplicons covering exons 9 and 10 of the Smo gene were directly sequenced by the Duke University DNA Sequencing Facility and screened for point mutations using Sequencher™ software [Gene Code, Ann Arbor, MI] and Chromas 2.3 shareware [Technelysium, Australia].

Statistical analysis

Descriptive measures were calculated as the mean ± SD, median, or percent of the appropriate denominator. All statistical calculations and simulations were carried out using R version 2.1 [ 48 ]. The stochastic discrepancies between the distributions of two continuous variables were assessed using the Wilcoxon–Mann–Whitney test [ 49 ]. In two sample problems, the employment of non-parametric tests over their parametric counterparts is generally more provident as the underlying distributions are a priori not known. In the situation of small sample sizes, the utility of non-parametric tests may be limited due to low power to detect stochastic discrepancies. Therefore, we quantitatively assessed these discrepancies using Welch's version of the t -test [ 50 ]. We note that the control of the Type I error is not guaranteed as the underlying distributions are not necessarily normal and that the observations within each sample are not mutually independent by virtue of the normalization method. The pairwise associations between target gene expression levels and continuous clinical outcomes [e.g. tumor size and serum alpha-fetoprotein [AFP] level] were estimated using Spearman's rank correlation [ρ] [ 49 ]. Given the sample size and the presence of ties in the data, the null distribution was approximated using 50 000 permutation replicates rather than using asymptotics. To explore gene expression relationships, exploratory cluster analyses using Spearman's correlation coefficient, as the distance measure, were employed. P -values were not adjusted for multiple testing.

Results

Normal hepatocytes lack Hh pathway activity

Given that Hh pathway activation is obligatory for liver bud formation, it is conceivable that cells in adult livers might have residual Hh pathway activity. To address this issue, we studied Ptc-lacZ mice where Hh-responsive elements in Ptc , a known downstream gene target of the Hh pathway, drive β- galactosidase expression to report Hh activity. We examined three healthy Ptc-lacZ mice [10–14 weeks old] to determine if mature hepatocytes exhibited Hh activity. LacZ- expressing hepatocytes were not detected at ×20 [ Figure 1A ] or ×100 magnifications [ Figure 1B ], although there were numerous β-galactosidase-positive cells in the wall of the gallbladder [×40 magnification] [ Figure 1C ], consistent with the role of Hh signaling in gallbladder cancer [ 15 ]. Our finding that mature hepatocytes lacked Hh activity was consistent with results from other groups [ 7 , 15 ], and further verified by our subsequent studies of primary hepatocytes isolated from the livers of two additional Ptc-lacZ mice. Protein extracted from the freshly isolated hepatocyte fraction did not exhibit β-galactosidase activity [data not shown].

Fig. 1.

Normal adult hepatocytes lack Hh pathway activity. Liver sections of transgenic Ptc-lacZ mice in which β- galactosidase reports cellular Hh activity [blue] at [ A ] ×20 and [ B ] ×100 magnifications. [ C ] In the same sections, the gallbladder wall was a positive control [×40 magnification].

Malignant human HCC lines express Hh pathway components

In order to determine if malignant hepatocytes express components of the Hh signaling pathway, we studied two well-characterized in vitro models of liver cancer, the HepG2 and Hep3B cell lines [ 51 ]. Using two-step RT–PCR we found that both lines expressed the Hh ligands, Shh and Ihh , the tumor-suppressor gene Ptc , the proto-oncogene Smo , as well as the downstream transcription factor, Gli1 [ Figure 2A ]. Quantitative real-time RT–PCR was done to compare gene expression in the two cancer cell lines and Percoll-isolated primary human hepatocytes [Hep]. In each assay, expression levels were normalized to that of the housekeeping gene, β- glucuronidase [ Gus ], in the same RNA samples. As expected, both malignant and non-malignant hepatocytes expressed Albumin [ Figure 2B ]. However, when compared to Albumin gene expression in Hep, the HepG2 cells expressed 2.1-fold more Albumin [ P < 0.0004] and the Hep3B cells expressed 40% less Albumin [ P < 0.0007]. Consistent with the routine use of the immature hepatocyte marker, AFP, as a serologic marker for HCC, both cancer cell lines strongly expressed this gene, while expression was barely detected in Hep. The HepG2 and Hep3B cancer lines expressed 208 064-fold [ P < 0.0001] and 602-fold [ P < 0.0001] more Afp than Hep, respectively [ Figure 2C ]. Expression of Hh ligands and Hh pathway signaling components was detected in both non-malignant and malignant hepatocytes [ Figure 2D ]. However, compared to Hep, the two HCC lines had 3- to 50-fold higher expression of Ihh [HepG2, P < 0.069; Hep3B, P < 0.055], Ptc [HepG2, P < 0.0011; Hep3B, P < 0.026] and Smo [HepG2, P < 0.012; Hep3B, P < 0.05]. Interestingly, the relative expression levels of Ptc , a tumor-suppressor gene, and Smo , a proto-oncogene, differed between the two HCC lines. HepG2 cells expressed relatively more Ptc than Smo , whereas Hep3B expressed higher levels of Smo relative to Ptc . These findings suggested that the activation of GLI1, a downstream target of SMO, may inherently differ between the two cancer cell lines.

Fig. 2.

Two HCC lines, HepG2 and Hep3B, express components of the Hh pathway. [ A ] Agarose gel electrophoresis of two-step RT–PCR products showing expression of Hh ligands and pathway components. Comparison of [ B ] Albumin , [ C ] Afp and [ D ] Hh pathway component expression in primary human hepatocytes [Hep] and the two HCC lines. Results are normalized to those in Hep [ †P < 0.05, *P < 0.01 and ‡P < 0.001].

Hep3B cells have Hh signaling activity

To further evaluate the relationship between Hh pathway expression and function, we assessed transcriptional activity of Gli , a downstream target of Hh signaling, in the Hep3B line, which had high expression of Smo relative to Ptc . Results were compared to an Hh-responsive, positive control cell line [C3H10T½] that was co-transfected with plasmids for a Gli -luciferase reporter and constitutively active Smo [ Figure 3A ]. As expected, C3H10T½ cells had endogenous Gli reporter activity, consistent with basal Hh pathway activity. Transfection of Smo further increased Gli -luciferase activity in these cells [ P < 0.0024]. Although not statistically significant, basal Gli activity in Hep3B cells was slightly higher than that of the positive control cell line [C3H10T½]. Smo transfection of Hep3B cells also significantly enhanced their Hh reporter activity [ P < 0.0002]. This 4.4-fold increase in reporter activity was also slightly higher than the 3.5-fold increase which Smo induced in the positive control cell line. The specificity of our assay was confirmed using an Hh-unresponsive, colon cancer cell line [HCT116] [ 15 ] as a negative control. These cells demonstrated a lack of luciferase reporter activity upregulation in the presence of Smo overexpression as compared to the C3H10T½ cells [ P < 0.0002] or Hep3B cells [ P < 0.0001] [ Figure 3A ].

Fig. 3.

Hep3B cells are regulated by Hh signaling. [ A ] To assess basal Hh pathway activity, Hep3B cells and Hh-responsive, positive controls [C3H10T½ cells] were co-transfected with a Gli-BS- Firefly luciferase reporter and a control reporter for Renilla luciferase [Vector]. To assess inducible Hh pathway activity, other cells in each group were also transfected with constitutively active Smo [Smo]. An Hh-unresponsive, negative control cell line [HCT116] was also studied. In all experiments, Firefly luciferase activity was normalized to control Renilla luciferase activity in the same cells [ *P < 0.0024, ‡P < 0.0002]. [ B ] Hep3B viability after 96 h Hh ligand neutralization with monoclonal antibody [5E1] or Hh blockade with Cyc. Results were normalized to appropriately treated controls [e.g. mouse IgG 1 isotype control antibody or tomatidine [Tom]]. [ C ] Hep3B viability following 96 h Hh blockade with KAAD-Cyc. Results were normalized to Tom-treated controls [ †P < 0.013]. [ D ] Hep3B growth rate during the period from 48 to 96 h in culture in Tom-treated and KAAD-Cyc-treated groups. Results were normalized to Tom-treated controls. [ E ] Quantitative RT–PCR analysis of Hep3B mRNA expression of c-myc and Smo following a 5 day treatment with 1000 nM Tom or KAAD-Cyc. Results were normalized to the Tom-treated controls [ †P < 0.046, ‡P < 0.0008]. [ F ] To determine if KAAD-Cyc blocked endogenous Hh pathway activity in Hep3B cells, this HCC line was co-transfected with a Gli-BS- Firefly luciferase reporter and a control reporter for Renilla luciferase. Cells were then treated with 1000 nM Tom or KAAD-Cyc. Firefly luciferase activity was normalized to control Renilla luciferase activity in the same cells [ †P < 0.029, *P < 0.005].

Hep3B cell viability is reduced by an inhibitor of the Hh pathway

As mentioned earlier, Hh signaling promotes the viability and growth of various foregut tumors. We evaluated the influence of Hh pathway activity on Hep3B viability by culturing the line with neutralizing antibody to Hh [5E1] or the pharmacological SMO blocker, Cyc, for up to 72 h in a dose-dependent fashion. Neither treatment reduced the viability of Hep3B cells as compared to controls treated with either isotype control antibody or Tom, an inactive Cyc analog [ Figure 3B ]. However, treatment with KAAD-Cyc, an agent that can inhibit oncogenically mutated SMO [ 26 ], inhibited Hep3B viability in a dose-related fashion, with significant decreases in viability noted at the 1000 nM dose [ P < 0.013, Figure 3C ]. This dose of KAAD-Cyc reduced the Hep3B growth rate from 48 to 96 h by 94% [ Figure 3D ]. These findings suggested that Hh activity promoted the viability of the Hep3B cell line.

Hh pathway inhibition regulates gene expression and pathway activity in Hep3B cells

Other groups have shown that induction of the c-myc proto-oncogene is critical for human hepatocarcinogenesis [ 52 , 53 ] and that its expression is regulated by Hh signaling [ 54 ]. Therefore, it is important to determine if inhibiting Hh activity affects c-myc in Hep3B cells. We found that a 5 day treatment with KAAD-Cyc decreased Hep3B mRNA expression of c-myc by 7.7-fold as compared to Tom-treated controls [ Figure 3E ; P < 0.046]. Similarly, KAAD-Cyc inhibition of SMO reduced Smo expression by 4.2-fold [ P < 0.0008], consistent with reports that SMO may regulate Smo expression [ 55 ]. These changes in gene expression are relatively selective because KAAD-Cyc had no effect upon the expression of cyclin B1 , D1 , D2 or E1 m RNA [data not shown]. This contrasts with what others have observed when Hh signaling is blocked with Cyc in medulloblastoma, another Hh-responsive cancer [ 54 ].

To establish whether blocking SMO influenced Hh-regulated transcriptional activity, we treated replicate Hep3B cultures for 1–2 days with 1000 nM Tom or KAAD-Cyc, and then analyzed the Hh reporter activity of the cells. KAAD-Cyc treatment reduced Hh-responsivity by 50% at one day [ P < 0.029] and 38% [ P < 0.005] at two days when compared to the Tom-treated controls [ Figure 3F ]. These reductions in reporter activity were particularly notable because the inherent inducibility of reporters allows for a greater dynamic range for activation than for repression [ 56 ]. Therefore, these findings confirmed that Hep3B cells have Hh signaling activity and demonstrated that Hh activity was reduced by treatment with KAAD-Cyc.

Oncogenic SMO inhibitor effects are independent of M1 and M2 Smo mutations

In order to determine if oncogenic Smo gene mutations could underlie the differential sensitivity of Hep3B cells to KAAD-Cyc and Cyc, we amplified DNA from Hep3B and HepG2 cell lines and performed direct sequencing analysis for previously described point mutations in the Smo gene. Point mutations in exon 9 [M2] and exon 10 [M1] hot spots are known to cause sporadic basal cell carcinomas [ 14 ]. Our sequencing analysis did not indicate a mutation at these loci in either cell line [data not shown]. Thus, Hep3B resistance to Cyc and sensitivity to KAAD-Cyc could not be explained by an activating Smo mutation at these sites, but suggested the potential for point mutations at other positions in the gene that have not been described as being oncogenic.

Expression of Smo correlates with tumor size in human HCC

Given that two human HCC cell lines overexpress Hh components and that one of the lines [Hep3B] exhibited constitutive Hh signaling activity, we evaluated the Hh pathway in 14 patients with HCC who underwent resection or liver transplantation. The mean tumor size in these individuals was 4.06 ± 2.48 cm [range 1–11]. Patient demographics are noted in Table I .

Total RNA was extracted from the paired non-neoplastic livers and HCCs. Using two-step real-time RT–PCR we compared Hh pathway expression in each patient's HCC with that in the respective non-neoplastic liver tissue at the resection margin. Cluster analysis using Spearman's rank correlation as the distance measure demonstrated that tumors that expressed more Shh than their adjacent non-neoplastic livers also tended to overexpress Ihh [ρ = 0.68, P < 0.01]. Half of the 14 tumors also had an increase in Gli1 expression, ranging from 1.5- to 131-fold higher than the non-neoplastic tissue. HCCs that had relative overexpression of Smo tended to have higher Gli1 expression [ρ = 0.47, P < 0.091]. This suggested that Smo overexpression in some of the tumors was associated with increased Hh activity.

Gene expression patterns were then analyzed for their relationship to patient and tumor characteristics. Overall, the 14 tumors averaged a 2.5-fold increase in Smo proto-oncogene expression. In 6 out of 14 HCCs [42.9%], expression of Smo was upregulated more than 3-fold. No tumors had significantly decreased Smo expression [ Figure 4A ]. Moreover, expression of the Smo proto-oncogene positively correlated with HCC tumor size [ρ = 0.54, P < 0.051].

Fig. 4.

Hh component expression correlates with HCC size in humans. Quantitative RT–PCR analysis of Smo and Ptc in 14 HCCs and matched, non-neoplastic liver tissues. [A ] Smo expression correlated with tumor size [ρ = 0.54, P < 0.051]. Smo mRNA levels in HCCs are normalized to Smo expression in adjacent, non-neoplastic tissues. [ B ] The ratio of Smo to Ptc expression was greater in large [≥5 cm diameter] than small [