|Home | About | Journals | Submit | Contact Us | Français|
The advent of targeted therapies in hepatocellular carcinoma (HCC) has underscored the importance of pathway characterization to identify novel molecular targets for treatment. Based on its role in cell growth and differentiation, we evaluated mTOR signaling activation in human HCC, as well as the anti-tumoral effect of a dual-level blockade of the mTOR pathway.
The mTOR pathway was assessed using integrated data from mutation analysis (direct sequencing), DNA copy number changes (SNP-array), mRNA levels (qRT-PCR and gene expression microarray), and protein activation (immunostaining) in 351 human samples, including HCC (n=314), and non-tumoral tissue (n=37). Effects of dual blockade of mTOR signaling using a rapamycin analog (everolimus) and an EGFR/VEGFR inhibitor (AEE788) were evaluated in liver cancer cell lines, and in a tumor xenograft model.
Aberrant mTOR signaling (phosphorylated-RPS6) was present in half of the cases, associated with IGF pathway activation, EGF upregulation, and PTEN dysregulation. PTEN and PI3KCA-B mutations were rare events. Chromosomal gains in RICTOR (25% of patients) and positive pRPS6 staining correlated with recurrence. RICTOR-specific siRNA downregulation reduced tumor cell viability in vitro. Blockage of mTOR signaling with everolimus in vitro and in a xenograft model decelerated tumor growth and increased survival. This effect was enhanced in vivo after EGFR blockade.
MTOR signaling has a critical role in the pathogenesis of HCC, with evidence for the role of RICTOR in tumor oncogenesis. MTOR blockade with everolimus is effective in vivo. These findings establish a rationale for targeting mTOR pathway in clinical trials in HCC.
Hepatocellular carcinoma (HCC) is a major international health problem. It has become the third cause of cancer-related death worldwide and the most common cause of death among cirrhotic patients in Western countries1, a trend paralleled by an increase in advanced hepatitis C virus (HCV)-related liver disease2. Currently only 30% of HCC patients are eligible for potentially curative treatments such as liver resection, transplantation, or local ablation, according to an established therapeutic algorithm3. Systemic chemotherapy is not efficacious in HCC4. Recently, regulatory agencies have approved the Raf/VEGFR/PDGFR inhibitor sorafenib, which significantly increased overall survival in patients with advanced HCC in a phase III trial5. This positive effect has spurred intensive research to unravel the genomic aberrations and signaling pathways implicated in hepatocarcinogenesis, and to find new therapeutic targets to tackle this devastating disease6.
The serine/threonine kinase AKT was isolated as an oncogene transduced by the acute transforming retrovirus7. Its essential role in human cancer was established by demonstration of its frequent amplification and overexpression in various cancers. AKT acts as a cytoplasmic regulator of numerous signals related to cell cycling (Cyclin D1), cell survival (Mdm2/p53), cardiovascular homeostasis (eNOS), and cell growth (mTOR), among others8. The tumor suppressor PTEN is a negative regulator of the pathway, and its loss activates AKT.
mTOR complex 1 (mTORC1) is a downstream target of AKT comprised of mTOR, a regulatory associated protein of mTOR (RAPTOR) and mammalian LST8/G-protein β-subunit like protein (mLST8/GβL). MTORC1 acts as a central regulator of cell growth and proliferation by activating S6 kinase, which in turn regulates protein synthesis and allows progression from G1 to S phase of the cell cycle9. In contrast, mTOR complex 2 (mTORC2), formed by mTOR, RICTOR and PRR5/GβL, is primarily responsible for the activation of AKT through its phosphorylation of serine 47310. In previous studies, activation of mTOR pathway in HCC ranged from 15 to 41%, and mTOR inhibitors show antineoplasic activity in experimental models of HCC11–14. However, these studies relied exclusively on immunohistochemistry data11, 13, associations with clinical outcomes were analyzed only in transplanted patients13, and there was no evaluation of multi-level blockade of mTOR pathway in HCC.
We report herein a significant role of mTOR pathway activation in human HCC, by using an integrative genomic approach and a training-validation scheme in a large cohort of patients. Aberrant pathway signaling was associated with activation of either the IGF and/or EGF cascade, rather than with somatic mutations of their components. MTOR blockade with a rapamycin analog (RAD001, everolimus; Novartis, Basel, Switzerland), decreased tumor growth and expanded survival in experimental HCC xenograft models. This effect was enhanced in vivo when administered along with an EGFR/VEGFR kinase inhibitor. Overall, the data supports efforts to target mTOR signaling in liver cancer patients.
A total of 351 human liver samples obtained either from liver resection or transplantation were analyzed in different sets (Figure 1). All of the samples were obtained from the HCC Genomic Consortium: Mount Sinai School of Medicine, NY (US), Hospital Clinic, Barcelona (Spain) and Istituto Nazionale dei Tumori, Milan (Italy). Upon Institutional Review Board approval and after patient written informed consent was obtained, tissue specimens were collected. First, we used an exploratory set of 77 fresh frozen samples including normal liver (n=10), cirrhosis (n=10), dysplasia (n=17), and different HCC stages (n=40, as previously described15, 16) to assess the mRNA levels of key genes of the mTOR pathway encompassing the entire pathogenic spectrum in HCV-related HCC. A replication set of 78 HCV-related HCC samples (fresh frozen and paraffin-embeded tissue) was used to confirm results in the exploratory set. For the clinical correlations, we selected patients from both sets who underwent liver resection (n=82, mTOR pathway clinical training set); all of them with HCV-related HCC. Finally, to validate the correlations between mTOR pathway and clinical outcome, we constructed a tissue microarray (TMA) with 196 HCC samples containing samples from all etiologies, including HBV and HCV-related liver disease (mTOR pathway clinical validation set, Table 1).
RNA extraction, purification, cDNA synthesis and PCR conditions can be found elsewhere15. Expression of mRNA was measured with Taqman Probes® (Supplementary Table 1) obtained from Taqman Gene Expression Assays® (Applied Biosystems). Ribosomal RNA (18S) was chosen for normalization. Gene expression microarray studies were conducted following procedures provided by the Affymetrix GeneChip Technical Manual.
DNA was extracted from human samples with the ChargeSwitch gDNA Mini Tissue Kit (Invitrogen) and quantified using PicoGreen (Invitrogen). A detailed explanation of the SNP-array procedures can be found elsewhere17. PCR and sequencing were conducted by GENEWIZ (http://www.genewiz.com). PCR primers for PTEN (exons 5 and 7), PI3KCA (exons 9 and 20), and PI3KB (exons 9 and 20) are listed in Supplementary Table 2 (see Supplementary Information for details).
Three well-characterized human liver cancer cell lines: Huh-7 (p53 mutated, PTEN wt), HepG2 (PTEN wt, N-RAS and β-Catenin mutated) and Hep3B were maintained as adherent monolayers in DMEM supplemented with 10% FBS, 5% L-glutamine and 1% penicillin-streptomycin. Cells were incubated at 37°C in 5% CO2. AEE788 is an ATP-competitive dual EGFR/VEGFR tyrosine kinase inhibitor, and RAD001 (everolimus) is a soluble derivative of rapamycin that inhibits mTOR activity. Both drugs were provided by Novartis Pharma AG (Basel, Switzerland).
Athymic female NU/NU mice (6–8 weeks of age) were subcutaneously injected in the right flank with 5 × 106 Huh7 cells. When tumors reached a volume between 100–250 mm3, mice were randomized in 4 groups: placebo (n=9; drug vehicle), AEE788 (n=9; 25 mg/kg), RAD001 (n=7; 5 mg/kg) and a combination of AEE788 (n=9; 25 mg/kg) plus RAD001 (5 mg/kg). Both drugs were administered 3 times/week per gavage until the animal was euthanized. Response was assessed by measuring delay in tumor growth and overall survival. Apoptosis and proliferation in tumors were evaluated using the DeadEnd™ Colorimetric TUNEL System (Promega) and Ki-67 immunostaining with mouse monoclonal anti-human Ki-67 antigen (Clone MIB-1, DakoCytomation), respectively. To assess the blocking activity of AEE788 and RAD001 in vivo, we used immunostaining for p-EGFR (Tyr1068, Cell Signaling) and p-RPS6 (Ser240/244, Cell Signaling), respectively (see Supplementary Information for details).
For inhibitory studies, cells were pretreated for 1 hour with AEE788 (10 μM), RAD001 (20 nM) or a combination of both. Then, cells were incubated with rh-EGF (40 ng/mL, Invitrogen). Blotting of membranes was performed using the following primary antibodies (all from Cell Signaling Technology, Danvers, MA): EGFR, phospho-EGFR (Tyr1173), Akt, phospho-Akt (Ser473), mTOR, phospho-mTOR (Ser2448), ERK1–2, phospho-ERK1–2 (Thr202/Tyr204), RPS6, phospho-RPS6 (Ser240/244), RICTOR and β-Tubulin (Santa Cruz). Secondary antibodies used were anti-mouse and anti-rabbit (GE Healthcare).
Formalin-fixed, paraffin-embedded sections were used to assess phosphorylated proteins in human tissue. Samples were incubated with anti-pEGFR (Tyr1068, Cell Signaling), anti-pIGF-IR (Tyr1316, provided by Dr. Rubini, University of Ferrara), anti-pAkt (Ser473, IHC specific, Cell Signalling), anti-pmTOR (Ser2448, IHC specific, Cell Signalling), and anti-pRPS6 (Ser240/244, Cell Signaling). Immunoreactivity was independently graded by three liver pathologists (ST, IF and MS) and finally determined based on their agreement. The variables measured were as follows: (1) Intensity of staining (0=absent, 1=weak, 2=moderate, 3=strong); (2) distribution of staining (1=very focal, 2=focal, 3=difuse) and (3) localization of the staining (membranous, cytoplasmic, or nuclear). Samples were defined as positive for p-EGFR, p-IGF-IR, p-Akt and p-mTOR when intensity of staining was +2 or higher, regardless of distribution. For p-RPS6, samples were defined positive when intensity and distribution were +2 or higher. TMA blocks were constructed using the Advanced Tissue Arrayer ATA-100 (Chemicon International, see Supplementary Information for details).
Cells were plated into 12-well (10,000 cells/well) or 24-well (5,000 cells/well) plate for 3H-Thymidine Incorporation and MTT Assays, respectively. Cells were incubated with increasing concentrations of AEE788 [0.5–10 μM], RAD001 [1–20 nM], and RNAi (RICTOR and control, Stealth Technology®, Invitrogen). For cell viability studies, after 24, 48 and 72 hours, cells were incubated with tetrazolium reagent for 1 hour. Culture medium supernatant was then removed and N-propyl alcohol was added. Following thorough solubilization, the absorbance (OD) of each well was measured using a microculture plate reader at 570 nm. For proliferation studies, cells were incubated with the drugs for 24 hours and were then labeled for 3 hours with 1 μCi/mL 3H-Thymidine, fixed in 1N hydrochloride acid and lysed in 0.25N NaOH. Thymidine incorporation was measured in a scintillation counter. The same experiments were also conducted under serum-free conditions (24 hours of starvation).
Apoptosis was measured by propidium iodide (PI) staining and fluorescent-activated cell sorting (FACS) in order to analyze the percentage of cells with subdiploid DNA content (sub-G1) characteristic of fragmentation. Huh-7 cells were plated into each well of a 6-well plate (50,000 cells/well). Cells were treated with AEE788 [10 μM], RAD001 [20 nM] and a combination of both. Forty-eight hours later, cells received a second dose of drug and were lysed 48 later. Harvested Huh-7 cells were pelleted by centrifugation and fixed with 100% ethanol. Samples were then stored at −20°C for 16 hours until FACS analysis was performed. To measure DNA content, cells were pelleted and resuspended in PBS containing 50 μg/mL propidium iodine and 100 μg/mL RNAse. Data was acquired on FACSCalibur flow cytometer (Becton Dickinson Instrument, San Jose, CA) and analyzed using CellQuest 3.1 software (Becton Dickinson). Apoptosis was further confirmed by measuring PARP cleavage using immunoblotting with PARP antibody (Cell Signaling).
Huh7 transfection efficiency was evaluated using the Block-it™ Fluorescent Oligo (Invitrogen). We performed a dual luciferase reporter assay using a human c-fos promoter subcloned into the pGL3 luciferase vector (Promega, Fitchburg, WI) upstream of the luciferase gene (a gift from Dr. Ron Prywes, Columbia University, NY)18. Briefly, Huh-7 cells were transfected with 1 μg of the c-fos reporter and 5 ng of pRL-TK plasmid driving the Renilla luciferase gene using 2.5 μL of Lipofectamine 2000 (Invitrogen, Carlsbad, California). Cells were incubated with EGF ([40 ng/mL], Invitrogen) for 30 minutes, lysed, and analyzed for firefly and renilla luciferase activity using the Dual Luciferase Kit® (Promega). Luciferase activity was measured with a luminometer and was normalized against the activity of the Renilla luciferase gene. RNAi experiments targeting RICTOR were conducted using Stealth® Technology (Invitrogen) following the manufacture’s instructions. Downregulation efficiency was evaluated at the protein level using western blot for RICTOR (Cell Signaling).
All the qRT-PCR calculations were analyzed by using the ddCt Method following the same methology as previously described15. Comparisons between groups were made using either the t-test or the non-parametric Mann-Whitney test for continuous variables, and the Fisher exact test for comparison of proportions. Correlations were calculated either with the Pearson’s coefficient or the non-parametric Spearman’s coefficient.
SNP arrays were processed using the GenePattern software package19 with modules based on dChipSNP algorithms. Gene set enrichment analysis (GSEA) was used to assess which gene sets were significantly associated with differential gene expression among tumors with positive staining for pRPS6 (Ser240/244) and gains in RICTOR20. RPS6 and RICTOR differential expressed genes were obtained using the Significance Analysis of Microarrays Package21.
For clinical correlations, the probability curves of recurrence and early recurrence were calculated according to Kaplan-Meier and compared by Mantel-Cox test. Protein staining status and median inferred copy numbers for the genes analyzed for each sample were used as covariates for univariate association with recurrence (deaths were censored in this analysis22). Clinico-pathological variables assessed in the univariate analysis were tumor size, multinodularity/satellites, vascular invasion, differentiation degree, BCLC stage and AFP levels. Molecular variables analyzed were: staining status of p-RPS6, p-Akt, p-IGF-IR, p-EGFR, p-mTOR, gains in RICTOR, mRNA levels of EGF and IGF2. Significant variables (p<0.05) were included in a step-wise Cox regression analysis of recurrence. Early recurrence was defined as within two years of surgical resection23. All calculations were done by the SPSS package (SPSS 15.0, Chicago, Illinois, see Supplementary Information for details).
We conducted an expression study using qRT-PCR in two different human cohorts, exploratory (n=77) and replication (n=78) sets (Supplementary Fig. 1 and Fig. 2). Dysregulation of key growth regulatory genes including EGF, IGFBP3 and PTEN was evident in overt HCC. EGF was up-regulated, particularly in advanced HCC cases (p=0.001), and the tumor suppressor IGFBP3 was down-regulated in early and advanced HCC (p<0.001). Also, a subgroup of 9 HCC patients had very high (>500 fold) upregulation of IGF2, what justifies the asymmetric distribution of this variable. In both sets, PTEN was down-regulated in advanced HCC (p=0.01). RAPTOR and mTOR were coordinately up-regulated in advanced tumors (1.7 and 2.3 mean fold-increase, respectively, p<0.01; Pearson’s coefficient=0.57, p<0.001). These data was consistent with whole-genome microarray transcriptomic analysis that was conducted in parallel (data not shown).
We used SNP-array technologies to assess copy number alterations in nine genes of the mTOR pathway (PI3KCA, AKT1, PTEN, mTOR, S6K, IGF2, IGFBP3, RAPTOR and RICTOR) in 99 HCC fresh-frozen samples and their cirrhotic counterparts. Overall, there were no high-level amplifications (cut-off above 3.4 copies) or deletions (data not shown), and only RICTOR showed significant DNA gains. Sequencing analysis showed a very low mutation rate of PTEN (1%, 1/102), PI3KB (1%, 1/102) and PI3KCA (0%).
To assess the activation status of mTOR pathway, we studied different members of the mTOR cascade at the protein level. Rates of tumoral staining for p-Akt, IGF-IR and p-RPS6 were 31.2% (29/92), 20.3% (16/79) and 47.7% (41/86), respectively; all were significantly higher than surrounding cirrhotic tissue (p<0.01, Supplementary Fig. 2). Activation of EGF signaling was present in 48.5% of cases (ns). In contrast to the null positive staining in cirrhotic tissue, 19.2% (15/78, p<0.01) of the tumor samples also displayed prominent staining for p-RPS6 in endothelial cells.
Activation of pRPS6 was significantly associated with EGF signaling: p-EGFR (63.1%, 24/38 vs 35.1% 13/37; p=0.02) and high EGF mRNA levels (>5 fold; p=0.01). Similarly, pRPS6 activation was also consistently associated with positive p-IGF-IR (68.7% 11/16 vs 39.6% 25/63; p=0.03, Table 2). All the above suggests a more prominent ligand-dependant mechanism of activation, rather than a mutation-based phenomenon. It has to be emphasized that mTOR signaling activation was identified in different HCC molecular subclasses recently reported based on unsupervised clustering of gene expression microarray data17. However, there was a significant enrichment of mTOR activation in the proliferation subclass, characterized by AKT-mTOR and IGF signaling activation17.
Activation of pRPS6 was associated with moderate/poorly differentiated tumors (p=0.04) BCLC B/C (p=0.003), and higher levels of AFP (p=0.012), whereas gains in RICTOR (p=0.03) and p-Akt positive staining were more prevalent in larger tumors (p=0.007). Also, gains in RICTOR were significantly associated with p-mTOR staining (p=0.01). There was a clear shift in p-mTOR localization in cirrhotic tissue and HCC. Staining in cirrhosis was predominantly membranous, while it was typically located in the cytoplasm in HCCs (p<0.01).
For outcome prediction we used two independent cohorts of HCC patients treated by surgical resection (Table 1), one including 82 HCV-derived HCCs, and a validation set of 196 HCC patients from all etiologies, where 67.3% of tumors showed positive pRPS6 staining (132/196). Overall, most of the patients had well-preserved liver function (98% Child-Pugh class A), early HCC (BCLC class 0-A ~90%) and small size tumors (median size 3.5 cm). Clinical variables such as tumor size, BCLC class, macrovascular invasion, and multinodularity/satellites were significantly associated with recurrence (p<0.05). In the independent set of 196 samples, p-RPS6 was an independent predictor of recurrence (HR=1.8, IC 95%: 1.1–2.8, p=0.01) along with BCLC staging (HR=1.9, IC 95%:1.2–3.1, p=0.004) and the presence of tumor multinodularity/satellites (HR=2.2, IC 95%: 1.4–3.5, p=0.001). The median time to recurrence in p-RPS6 positive and negative patients were of 25 and 50 months, respectively (p=0.004, Fig. 3). These results suggest a potential prognostic relevance of mTOR activation in HCC patients.
To compile a specific gene signature associated with mTORC1 pathway activation, we profiled 91 HCC samples using the human U133 plus 2.0 array (Affymetrix). After supervised analysis using the Significance Analysis of Microarrays Package, we found 193 up-regulated and 127 down-regulated genes distinguishing patients according to p-RPS6 staining status (FDR<0.5%). Among them, up-regulation of genes related to NF-Kappa β (ZDHHC13), MAPK (MAPK13) pathways, AMPK subunits, and angiogenesis (VEGFB) were most prominent (Supplementary Table 3). As expected, GSEA showed that a gene set formed by 121 genes involved in capping, splicing, editing and modification of mRNA (i.e. mRNA Processing Reactome) was enriched in phospho-RPS6 positive samples (q-value=0.06).
SNP-array analysis showed increased copy numbers in RICTOR in one fourth of cases (25/100), which were significantly associated with mRNA up-regulation (p=0.03, Fig. 4A–B). Gains in RICTOR (p=0.002, Fig 4C–D) and combined gains in RICTOR and activated RPS6 (9.8%, 8/81) were significantly associated with recurrence (p<0.001) in the training set. Also, gains in RICTOR were an independent predictor of recurrence (HR=2.3, IC 95%: 1.2–4.4, p=0.01) along with BCLC staging (HR=4.8, IC 95%: 2.7–8.7, p<0.001). Supervised analysis of gene expression data show that EGR2, a candidate tumor suppressor gene that interacts with PTEN24, was significantly downregulated in samples with gains in RICTOR. Also, several subunits of AMPK were significantly enriched in these samples. In vitro, RNAi mediated downregulation of RICTOR in Huh7 cells, which harbor gains in RICTOR, induced a 17% reduction in cell viability measured with the MTT assay, when compared with cells transfected with control siRNA (Fig 4E–F, p<0.05). Conversely, cell viability in HepG2, a cell line without gains in RICTOR, remained unchanged.
Everolimus (RAD001) [20 nM] decreased cell viability in Huh7, Hep G2 and Hep 3B at 72-hours up to 36% (p<0.001, Fig. 5). Increasing concentrations of an EGFR inhibitor (AEE788) induced a time- and dose-dependent reduction in cell viability of the 3 cell lines. After 72 hours, high concentrations of EGFR inhibitor [10 μM] reduced cell viability up to 85% (p<0.006). Combination therapy did not enhance the effect on cell viability compared with single EGFR inhibitor (Fig. 5C). Everolimus significantly decreased proliferation up to 20% in Huh-7 (Fig. 5D), while the inhibition by the EGFR inhibitor [10 μM] was more than 90% in the 3 cell lines (p<0.01).
We further examined the mechanism of action of the kinase inhibitors in vitro by FACS analysis. The mTOR inhibitor did not induce apoptosis, whereas the EGFR inhibitor [10 μM] alone and in combination with everolimus significantly increased the percentage of cells in sub-G1 phase up to 38% and 40%, respectively (Fig. 5E). Apoptosis was confirmed by measuring PARP cleavage (Fig. 5F).
To elucidate the efficacy of the kinase inhibitors in blocking downstream targets, we measured the effect of both drugs in the phosphorylation status of different proteins of the Akt/mTOR pathway (EGFR, Akt, mTOR, RPS6) as well as ERK1/2 (Fig. 6A). As predicted, EGFR-inhibitor decreased the phosphorylation of EGFR, Akt and ERK1/2 in Huh7 while everolimus significantly reduced the phosphorylation of RPS6. Combination therapy simultaneously blocked both signals. Similar results were obtained in HepG2 and Hep3B lines (data not shown).
We employed a c-fos luciferase reporter as a surrogate of EGF signaling activation, and found a significant decrease in luciferase activity up to 65% in Huh-7 cells treated with EGFR inhibitor [10 μM] alone and in combination with everolimus after 30 minutes of stimulation with rh-EGF. In accordance with the protein studies, everolimus [20 nM] did not modify the signal from the c-fos reporter (Fig. 6B).
Oral administration of an mTOR inhibitor (everolimus;5 mg/kg), EGFR-inhibitor (AEE788; 25 mg/kg), or placebo (drug vehicle) were well tolerated by tumor-bearing mice without significant weight loss. Everolimus and the EGFR inhibitor induced a significant delay in tumor growth in comparison with control mice (mean volume after 15 days of treatment: 1039 and 1395 mm3, respectively vs 2396 mm3, controls; p<0.05). Combination therapy further reduced the tumor volume to 784 mm3 (p<0.05, Fig. 6C). Complete tumor response was evident in 2 animals in the EGFR inhibitor and combination arms. Overall survival was enhanced in combination therapy (median 27 days), in comparison with each drugs alone (median everolimus: 19 days, EGFR-inhibitor: 21 days), or with control mice (median 16 days), (p<0.05, Fig. 6D).
We then examined the impact of blocking mTOR and EGFR signaling on apoptosis and proliferation using TUNEL and Ki-67 staining in vivo, respectively (Supplementary Fig. 3). There was a significant reduction in the proliferation index from 80.8±9.5% in control mice to 52.6±25% in the everolimus arm (p<0.05) and to 57±21% in the combination arm (p<0.05). Regarding apoptosis, there was a significant increase in the apoptotic index from 8±6 apoptotic bodies per 10 high power fields in control mice to 18±6 (p<0.05) and 16±7 (p<0.05) in EGFR-inhibitor alone and in combination, respectively.
To evaluate selective inhibition of downstream targets, we assessed the expression p-EGFR and p-RPS6 in tumor sections using immunhistochemistry. Immunohistochemical reactivity of p-EGFR was decreased in mice treated with EGFR inhibitors or combination therapy, whereas p-RPS6 staining was diminished in mice treated with everolimus and those treated with the combination therapy (Supplementary Fig. 3).
Recent studies have identified PI3K/Akt/mTOR pathway as a major oncogenic cascade for targeting molecular therapies in cancer25. MTOR signaling has been implicated in the initiation and progression of multiple tumors, such as leiomyosarcomas and gliomas26. We demonstrate herein that mTOR pathway is activated in a subset of patients with early HCC. Activation of mTOR cascade resulted from ligand-dependant signals from EGF and IGF signaling, rather than from a mutation-dependent mechanism, since no high-level amplifications and only marginal mutation rates in the most prevalent hot spots in PTEN, PI3KCA and PI3KB were identified. In fact, in a subset of patients, high levels of EGF could be primarily responsible for RPS6 activation. Coincidentally, down-regulation of the tumor suppressor PTEN was observed in a significant proportion of patients, mostly in advanced stages of the disease. All these results highlight the relevance of mTOR signaling in HCC, a pathway that has been insufficiently explored in human liver cancer.
These data are the first to characterize the status of RICTOR status in human HCC. RICTOR is part of MTORC2, however its functions and molecular structure are not entirely known27. We found a significant association between gains in RICTOR (25% of HCC) and its transcript expression. Intriguingly, gains in RICTOR were significantly associated with p-mTOR, which could be a relevant mechanism of MTORC2 activation in human cancer. Genes significantly dysregulated in samples with gains in RICTOR (e.g. EGR2 and AMPK subunits) suggests that mTORC2 activity could influence signaling through the PTEN-mTORC1 axis. In vitro, downregulation of RICTOR by siRNA reduced cell viability of Huh7 cells, which harbor gains in RICTOR.
We speculate that in HCC, other mechanisms besides mTORC2 activation may be responsible for Akt phosphorylation at residue 473, since no correlation between p-Akt staining and gains in RICTOR was observed. Membranous localization of p-mTOR was visibly lost in liver cancer; whether this event is due to divergent mTOR complex 1 and 2 cellular localization requires further investigation. Finally, a significant association between gains in RICTOR and early recurrence was identified in the training set, a finding that requires further validation. Strikingly, and in accordance with this data, a recent report indicates that MTORC2 is hyperactivated in gliomas and functions in promoting tumor cell proliferation and invasive potential28. Overall, these results suggest a possible role of MTORC2 in human hepatocarcinogenesis and warrant further evaluation of RICTOR as a potential new target in HCC therapy.
Apart from the potential role of RICTOR as an oncogene in liver carcinogenesis, we describe other mechanisms underlying mTOR pathway activation. We have confirmed dysregulation of key genes of the mTOR pathway (e.g. IGF2, IGFBP3 and PTEN) in HCC29, 30, and herein reported dysregulation of other genes that includes mTOR and RAPTOR. Alterations in copy number or somatic mutations of PTEN, PI3KCA and PI3KB were not identified as major mechanisms of mTOR pathway dysregulation, although these genes were only assessed for mutations in certain exons. The infrequent mutation rate of PTEN (<10%) in earlier studies31, supports our data, and suggests that other mechanisms apart from missense mutations may be responsible for PTEN downregulation (e.g. promother methylation32). There is some controversy regarding PI3KCA mutations, with frequency rates that range from 0%33 to 35%34 in HCC. We did not identify any mutation in 2 exons of the PI3KCA gene.
The associations found at the protein level between p-EGFR/EGF and p-RPS6 suggest that RPS6 activation in HCC is partly driven through EGF signaling. Interestingly, our cohort showed higher rates of p-RPS6 activation than p-Akt, and similar to previous reports13, there was not a significant correlation between both proteins, which reinforces the concept that alternative pathways besides Akt are responsible for the activation of MTORC1 (e.g. LKB1/AMPK signaling35 or direct signals from IGF-IR activation). Interestingly, marker genes significantly up-regulated in RPS6 activated HCC were related to NF-Kappaβ signaling (ZDHHC13), angiogenesis (VEGFB) and MAPK signaling (MAPK13), highlighting the complex signaling cross-talk that is developing even at early clinical stages.
The primacy of mTOR activation in early HCC persisted in both cohorts analyzed, but was not associated with specific etiological factors. We found similar rates of p-RPS6 staining as previously described, but p-RPS6 in our cohort was a predictor of poor prognosis13. Recent studies conducted by our group have identified, by unsupervised clustering of microarray data, a singular subgroup of HCC patients (proliferation group) where activation of IGF and mTOR (p-RPS6) were enriched17. This group has been previously associated with poorest outcome36. In our series, patients with mTOR activation showed higher levels of AFP, less differentiated tumors, and a higher incidence of recurrence. These data highlight the relevance of this pathway as an attractive target for the development of new anti-tumoral agents.
To address the issue of multi-targeted blockade of mTOR signaling in HCC and neutralization of potential oncogenic loops, we evaluated the effect of a dual-level blockade of mTOR by an mTOR inhibitor (everolimus, RAD001) and an EGFR/VEGR inhibitor (AEE788) in experimental models. The EGFR inhibitor induced high rates of apoptosis in vitro, a phenomenon not observed with the rapamycin analog. Although everolimus has already proved antitumoral effects in a rat model of HCC14, our in vivo studies in mouse xenografts revealed an additive anti-tumoral effect with survival implications after dual-level blockade of mTOR pathway. Regarding anti-target activity, RAD001 and AEE788 effectively blocked EGFR and RPS6 phosphorylation, respectively, in vitro and in vivo. Unlike everolimus, the EGFR inhibitor significantly increased apoptosis based on FACS analysis, PARP cleavage and TUNEL staining. Everolimus exerted its antineoplasic activity by inducing G1 arrest and hence inhibiting proliferation. Therefore, this combination of pro-apoptotic and anti-proliferative mechanisms may explain the additive effect obtained in the experimental model. Recent reports demonstrate that rapalogs inhibit mTORC2 activity in vitro37, so we cannot exclude additional antitumoral activity related to this mechanism.
The recent breakthrough in the management of HCC by using the muti-kinase inhibitor sorafenib has established the concept of extending survival by using molecular targeted therapies in this otherwise chemo-resistant disease. Sorafenib provides three months survival extension in patients with a base-line median survival of 8 months5. These data have stimulated the search for additional combination therapies, recognizing that HCC is an heterogeneous cancer with many genetic aberrations. In the present study we establish that mTOR pathway activation plays a role in HCC and that RICTOR may be a mediator of human hepatocarcinogenesis, although further evidence will be required to characterized RICTOR as a new oncogene in human cancer. Up-stream ligand-dependant mechanisms might also be responsible for mTOR pathway activation, indicating that this cascade offers relevant targets for cancer drug discovery. The combination of rapamycin analogs and EGFR inhibitors provides evidence for a proof-of-concept effect in experimental models of HCC. Consequently, these data provides the rationale to test combination therapies in early clinical trials in human liver cancer that includes RAD00138. These studies should follow the guidelines reported on design and endpoints in clinical trials in HCC22.
Grant Support: Augusto Villanueva is supported by a grant from Fundación Pedro Barrié de la Maza, Asociación Española para el Estudio del Hígado, National Cancer Center and EASL-Sheila Sherlock Fellowship. Philippa Newell is a recipient of an American Liver Foundation Fellowship. Beatriz Minguez is supported by a grant of Instituto de Salud Carlos III (FIS-CM04/00044). Yujin Hoshida is a recipient of a Charlie A. King Trust fellowship. Vincenzo Mazzaferro was supported by the Italian Association for Cancer Research and the Italian National Ministry of Health. Jordi Bruix is supported by a grant from Instituto Carlos III (ISCIII/FIS PI 05-0150). Josep Llovet was supported by ICREA and grants from the U.S. National Institute of Diabetes and Digestive and Kidney Diseases (1R01DK076986-01), The Samuel Waxman Cancer Research Foundation and the Spanish National Health Institute (SAF-2007-61898). Scott Friedman was supported by grants from NIH (1RO1DK37340-23)
The authors of this manuscript want to honor the memory of our colleague and friend, Eric R Lemmer, member of the Mount Sinai Liver Cancer Program and the Division of Liver Diseases. We also want to thank Dr Ron Prywes (Columbia Unversity, NY) for providing the c-fos luciferase reporter. Novartis Pharma provided the drugs (AEE788 and RAD001) used in this study.
Financial disclosures: There are no financial disclosures relevant to this manuscript. Novartis supplied AEE788 and everolimus in the context of an MTA.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.