Salirasib inhibits the growth of hepatocarcinoma cell lines in vitro and tumor growth in vivo through ras and mTOR inhibition

Hepatocellular carcinoma (HCC) is the fifth most common cancer worldwide, and the third leading cause of cancer related mortality. Its incidence has more than doubled during the last two decades in the western world, where it is the fastest growing cause of cancer related death [1]. Despite the magnitude of the problem, existing therapies are of limited efficacy. No more than 30% of the patients are eligible for curative treatment, and recurrence is a frequent issue affecting up to 70% of the patients after tumor ablation. Moreover, due to underlying cirrhosis, systemic therapy with classical cytotoxic drugs is poorly tolerated and ineffective [2]. Accordingly, new therapeutic approaches for this disease are eagerly awaited.

Several growth factor signaling pathways are dysregulated in hepatocarcinogenesis [3]. In particular, altered intracellular signaling elicited by epidermal growth factor (EGF), insulin-like growth factor (IGF) and Vascular Endothelial Growth Factor have been involved in the pathogenesis of HCC. Hence, inhibitors of their receptors are under intense investigation. While anti-IGF receptor (IGF-1R)-based therapies are currently studied in preclinical and early clinical trials, inhibition of the EGF receptor (EGFR) by either tyrosine kinase inhibitors or monoclonal antibodies has shown limited efficacy in several phase II studies in HCC [4]. In non hepatic epithelial tumor cell lines, inhibition of EGFR or IGF-1R individually promotes activation of the reciprocal receptor [5] and IGF-2 overexpression has been involved in the resistance of HCC to EGFR inhibition in a rat model [6]. Treatment interfering with both receptors could thus represent a better strategy to treat this disease. Alternatively, targeting one or several of their downstream signaling pathways could be an elegant way to block growth factor signaling. Among those, both ras-raf-MEK-ERK and PI3K-Akt-mTOR pathways are activated upon EGFR and IGF-1R stimulation. While ras activation upon EGFR stimulation induces PI3K activation [7], IGF-1R is able to activate the PI3K-Akt-mTOR pathway independently of ras [8].

Ras activation has been shown to be an ubiquitous and early event in human HCC [9], whereas mTOR activation is present in half of the cases [10]. Downstream receptor signaling inactivation has proved its efficacy as demonstrated by the results of the SHARP trial evaluating sorafenib, a multikinase inhibitor targeting the VEGFR and PDGFR kinases as well as raf, in advanced HCC. However, it only leads to a modest increase in median overall survival of 3 months [11], highlighting the need for the development of new and more effective targeted therapies for HCC.

Salirasib (Farnesylthiosalicylic acid, FTS) is a S-farnesyl cysteine analog that affects docking of active GTP-bound ras in the cell membrane by competing with ras for its membrane anchorage sites and consequently inhibits ras-dependent cell growth [12]. In cell lines, this leads to an accelerated degradation of cytosolic ras and a decrease in the total amount of cellular ras [13]. This mode of action affecting all ras isoforms differentiates salirasib from farnesyltransferase inhibitors, which fail to block K-ras and N-ras activity because they undergo geranylgeranylation following treatment with those molecules [7]. In addition, salirasib has also been shown to directly inhibit mTOR complex 1 activity by disruption of the mTOR-raptor complex [14]. It exhibits anti-tumoral effects in several non-liver cancer cell lines [15] and has recently been evaluated in a phase 1 study in patients with solid non hepatic tumors, showing that it was well tolerated [16]. Targeting both ras and mTOR, along with a good tolerance in patients, make salirasib a good candidate for HCC therapy.

Previous work of our team has shown that high dose salirasib blocks hepatocytes proliferation in vivo in rats after partial hepatectomy [17]. This inhibitory effect was at least partially mediated by inhibition of ERK phosphorylation. More recently, we have shown that salirasib administration prevents liver tumor development in a model of diethylnitrosamine-induced hepatocarcinogenesis in rats [18].

The aims of the present study are to evaluate the efficacy of salirasib in human HCC cell lines, and to understand its underlying molecular mechanisms of action in these particular cells thereby providing a rationale for testing it as a novel anti-cancer treatment in HCC clinical trials.

Ras and mTOR are regarded as relevant therapeutic targets in HCC [9, 10]. In this study, we report for the first time the effect of salirasib, a novel prenylcysteine analogue inhibiting cell growth in three human HCC cell lines through interference with ras and mTOR. Even more importantly, salirasib was able to inhibit both EGF- and IGF-induced proliferation in human HCC cell lines, potentially reducing the possibility for escape mechanisms related to activation of one growth factor pathway in response to the inhibition of the other one. Although IC₅₀ were similar after three days of treatment in the three tested cell lines, time-course experiments suggests that Hep3B cells are the most sensitive to salirasib among the three tested cell lines, while Huh7 cells are more resistant. Importantly, our results also demonstrate that on the long-term salirasib treatment is effective at doses far below the estimated IC₅₀.

The growth inhibitory effect is mainly mediated by inhibition of cell proliferation, which is observed in the three tested cell lines to a similar extent. This reduction of proliferation is associated with a profound modulation of the expression of cell cycle mediators. Cyclin A expression was strongly decreased in HepG2 and Huh7, and to a lesser extent in Hep3B. In the latter however, the cell cycle machinery disruption became clearly evident on the level of cyclin D1, the expression of which was almost completely abrogated upon treatment. In the two more sensitive cell lines, HepG2 and Hep3B, expression of the cell cycle inhibitors p21 and p27 was increased, reaching the highest magnitude in the most sensitive Hep3B cells. These observations partially mirror the impact of activated K-ras on the cell cycle, which is known to upregulate cyclin A and cyclin D, and to downregulate p27 [20]. On the other hand, mTOR inhibitors are known to induce a G1/S cell cycle arrest through an increase in p27 and a decrease in cyclin D [24] and cyclin A [25]. Thus, the impact of salirasib on cell proliferation might be due to a combination of both previously described effects of this compound, i.e. ras inhibition and mTOR inhibition [12, 14].

On the other hand, apoptosis also contributes to the growth inhibitory effect of salirasib, and the relative resistance of Huh7 compared to the two other cell lines might be due to the absence of apoptosis induction upon treatment in these cells. However, the contribution of apoptosis seems to be less prominent than the anti-proliferative action of salirasib, at least under our experimental conditions. Indeed, caspase activation is more pronounced in HepG2 cells than in the more sensitive Hep3B cells. In addition, in these latter cells, no apoptosis induction could be observed at 50 μM or 100 μM salirasib, although these doses already induce a dramatic decrease in cell counts over time.

Nevertheless, high dose salirasib elicited caspase-3/7 activation in two cell lines that might at least partially be mediated by the mitochondrial apoptotic pathway. Apoptosis could have been caused in our cells by down-regulation of survivin, as salirasib has been shown to reduce survivin expression in glioblastoma cells [26], which was sufficient to elicit apoptosis. In addition, survivin down-regulation by antisense oligonucleotides has been shown to inhibit cell growth and to induce apoptosis in several cell lines, including HepG2 [27, 28]. However, it was also repressed in the apoptosis-resistant Huh7 cells, suggesting that additional events are required to trigger cell death. Our results also suggest that salirasib might sensitize the cells to death receptor-induced apoptosis through up-regulation of the TRAIL receptors DR4 and DR5 in HepG2 and Hep3B cells, along with increased Fas expression in HepG2 cells and TNFα induction in Hep3B cells. Fas and TRAIL receptor upregulation alone might, however, not be sufficient to induce a major impact in vitro for their ligands, FasL and TRAIL, are mainly expressed on immune cells [29], which are not present in monocultures. Nevertheless, up-regulation of death receptors on tumor cells by treatments like salirasib and interaction with their respective ligands on immune cells could be of major importance in vivo, further potentiating the anti-tumor effect of salirasib.

Growth inhibition effects of salirasib are p53-independent as salirasib affect in a similar fashion HepG2 (p53 wild type) and Hep3B (p53 null) cells. This is further substantiated by the decrease in p53 expression observed after 2 days of treatment in HepG2 cells. This aspect could be clinically relevant, because most human HCC harbor defective p53 function [1]. A treatment strongly depending upon p53 activation could thus be less effective in these tumors. Our results contrast with a previous report of increased p53 function in colon cancer cells in response to salirasib [21]. However, p53 downregulation is compatible with ras inhibition, because K-ras activation is known to induce p53 up-regulation [20]. This lack of p53 upregulation in our study could be related to the absence of ERK inhibition upon treatment. Indeed, in HepG2 cells, ERK is a major activator of Mdm2, which is responsible for p53 degradation [30].

Total Ras protein expression was reduced in the three tested cell lines after 2 days of treatment, while Ras mRNA levels remained stable. In addition, salirasib reduced the expression of active GTP-bound Ras in HepG2 cells stimulated with EGF. These observations indicate an increase in ras protein degradation, which is consistent with the postulated mechanism of action of salirasib, involving the dislodgement of ras from the cell membrane followed by a cytosolic degradation [13]. Surprisingly, salirasib was unable to inhibit neither ERK nor Akt phosphorylation. On the contrary, it even tended to increase their phosphorylation levels, which could be due to a strong inhibition of p70 and to the consequent relief of a negative feedback loop affecting ERK and Akt [31].

Importantly, p70 phosphorylation was abrogated upon treatment in all cell lines when stimulated with EGF, which occurred without concomitant inhibition of ERK or Akt, both of which are known to activate mTOR. Moreover, salirasib also efficiently reduced p70 phosphorylation in all cell lines upon IGF2 stimulation, a situation where stimulation of the Akt-mTOR axis is independent of ras activation [8]. Indeed, no ras activation above baseline levels was observed in HepG2 cells stimulated with IGF2, and IGF2 did not induce ERK phosphorylation in any of the tested cell lines. Altogether, these data suggest that salirasib induced inhibition of mTOR in HCC cells occurs, at least in part, independently of ras, and thus point to a direct inhibitory effect on the mTOR complex 1, confirming earlier observations [14].

Nonetheless, it should not be concluded that the growth inhibitory effect that is observed in HCC cell lines solely relies on mTOR inhibition, as other unexplored ras mediators could be affected. Although, both ras and mTOR inhibition taken separately could explain the decrease in cyclin A and the increase in p27 levels, it is worth to note that these changes parallel the downregulation of ras in HepG2 and Hep3B cells.

Finally, we show that salirasib inhibits tumour growth in vivo in a subcutaneous xenograft model at a well tolerated dose. As salirasib is metabolized in the liver by cytochrome P450 2C subfamily (Concordia Pharmaceutical Inc., personal communication), there might be some concern about its potential efficacy in this organ. With regard to maintaining its efficiency in the liver as a target organ, we have shown that low-dose of salirasib prevented tumour occurrence in a model of diethylnitrosamine-induced hepatocarcinogenesis [18], while others have shown an impact of low-dose salirasib on liver fibrosis both in the preventive and the curative settings [32, 33]. Both observations confirm that salirasib remains active in the liver.

Our results indicate that salirasib elicits a dose- and time-dependent growth inhibitory effect in human HCC cell lines, related to inhibition of both EGF- and IGF-induced cell proliferation, and to a lesser extent to induction of apoptosis. This effect is linked with ras and mTOR inhibition, while ERK and Akt remained activated. Furthermore, we show that salirasib also exhibits anti-tumor activity in vivo in a mouse subcutaneous xenograft model. Our group has also previously described that salirasib prevents the development of preneoplastic liver foci in an animal model of diethylnitrosamine-induced hepatocarcinogenesis [18]. These results in human HCC cell lines, along with our previous observation of tumor prevention in vivo provide a rationale for testing salirasib in human HCC. Furthermore, investigation of combination therapies of salirasib and inhibitors of the raf/MEK/ERK pathway, the PI3K/Akt pathway, as well as combination with apoptosis-inducing treatments such as conventional chemotherapy or TRAIL-agonists are warranted in order to try to further improve the antitumor effect of salirasib.

The authors are grateful to Martine Petit for expert technical assistance. This work was supported by Saint Luc Foundation and University FSR grants to PS. NC is the recipient of Saint Luc Foundation and FSR fellowships. IL is a FRS-FNRS research associate.