Preclinical evaluation of antitumor activity of the proteasome inhibitor MLN2238 (ixazomib) in hepatocellular carcinoma cells

Hepatocellular carcinoma (HCC) is one of the common malignancies and is an increasingly important cause of cancer death worldwide. Surgery, chemotherapy, and radiation therapy extend the 5-year survival limit in HCC patients by only 6%. Therefore, there is a need to develop new therapeutic approaches for the treatment of this disease. The orally bioavailable proteasome inhibitor MLN2238 (ixazomib) has been demonstrated to have anticancer activity. In the present study, we investigated the preclinical therapeutic efficacy of MLN2238 in HCC cells through in vitro and in vivo models, and examined its molecular mechanisms of action. MLN2238 inhibited cell viability in human HCC cells HepG2, Hep3B, and SNU475 in a time- and dose-dependent manner. Flow cytometry analysis demonstrated that MLN2238 induced G2/M cell cycle arrest and cellular apoptosis in HCC cells. Cell cycle arrest was associated with increased expression levels of p21 and p27. MLN2238-induced apoptosis was confirmed by caspase-3/7 activation, PARP cleavage and caspase-dependent β-catenin degradation. In addition, MLN2238 activated ER stress genes in HCC cells and increased the expression of the stress-inducible gene nuclear protein-1. Furthermore, MLN2238 treatment induced upregulation of myeloid cell leukemia-1 (Mcl-1) protein, and Mcl-1 knockdown sensitized HCC cells to MLN2238 treatment, suggesting the contribution of Mcl-1 expression to MLN2238 resistance. This result was also confirmed using the novel Mcl-1 small molecule inhibitor A1210477. Association of A1210477 and MLN2238 determined synergistic antitumor effects in HCC cells. Finally, in vivo orally administered MLN2238 suppressed tumor growth of Hep3B cells in xenograft models in nude mice. In conclusion, our results offer hope for a new therapeutic opportunity in the treatment of HCC patients.


Introduction
Hepatocellular carcinoma (HCC) is known to be the second most frequent type of solid tumor 1 . Surgical intervention provides the best response in the early stages of the disease, but this approach is not feasible in all HCC patients. Standard therapy in advanced HCC patients involves the administration of Sorafenib, an oral multi-kinase inhibitor, which, unfortunately, has many side effects and increases life expectancy by only 3 months. This has led to the investigation of new treatment strategies and the identification of new target molecules, such as proteasome. Inhibition of proteasome causes an accumulation of misfolded proteins within the cell, an event that triggers the activation of the apoptotic pathway. Bortezomib (Velcade, PS-341), is a first-generation proteasome inhibitor, which the US Food and Drug Administration (FDA) has approved in multiple myeloma 2 and non-Hodgkin's lymphoma treatment 3 . At the molecular level, bortezomib treatment induces cell death through endoplasmic reticulum (ER) stress induction 4-7 , nuclear factor kappa B inhibition 8 , and caspase-8 activation 9 . However, although preclinical results have shown that bortezomib has antitumor effects in HCC [10][11][12] , a multicenter single-arm phase II trial conducted in cases of unresectable HCC showed that although bortezomib is well tolerated, it lacks significant activity 13 . Moreover, in many cases patients treated with bortezomib rapidly develop drug resistance, the mechanisms of which are poorly understood 14 .
The good clinical outcome observed with bortezomib in liquid tumor has led to the development of nextgeneration proteasome inhibitors to improve efficacy, avoid pharmaco-resistance and minimize cytotoxicity. Among them, MLN2238 (ixazomib) holds great promise: it is a next-generation reversible proteasome inhibitor, whose main value is that it can be administered orally. MLN2238 is the biologically active form of MLN9708 (ixazomib citrate), which in plasma or after exposure to aqueous solutions quickly hydrolyzes to MLN2238, the biologically active boronic acid. MLN2238 inhibits the 20 S proteasome chymotrypsin-like proteolytic (β5) subunit. It has a greater antitumor activity in solid and hematologic tumor models compared to bortezomib 15 . Several studies conducted in multiple myeloma patients have shown that ixazomib has great antitumor effects (NCT00963820; NCT00932698), and therefore the FDA has given its approval for treating this disease, also in association with other drugs, such as lenalidomide and dexamethasone (NCT02389517; NCT02917941) 16,17 . Furthermore, other newer reports have shown that MLN2238 is efficacious in other tumor cell types, such as osteosarcoma 18 , colon adenocarcinoma 19 , melanoma 20 , and neuroblastoma cells 21 . Treatment with MLN2238 results in the stabilization and accumulation of p21 Waf1/ Cip122 , E2F1 and p53 18 , which lead to the activation of caspase-3, -8, -9-dependent cell death pathways, with upregulation of Mcl-1 and NOXA 23,24 .
To date there are no studies on MLN2238 administration in HCC. In this study, we used HCC cells to explore the antitumor effects of MLN2238 in vitro as well as in vivo.

MLN2238 reduces cell viability and proliferation HCC cells
The impact of MLN2238 on HCC cell viability was assessed by MTS assay. HCC cell lines HepG2, Hep3B and SNU475 were exposed to treatment for 24, 48, and 72 h with increasing concentrations of MLN2238. As shown in Fig. 1a Furthermore, a BrdU incorporation assay was performed to determine cell proliferation after MLN2238 treatment. Fig. 1d shows that after 48 h MLN2238 significantly inhibited HCC cell proliferation according to the drug concentration.
In addition, inhibition of proliferation was evidenced by observations using optical microscopy. The images in Fig. 1e clearly show that the number of cells decreased at the various concentrations, and in addition, in HepG2 and SNU475 cells, a typical morphological feature of apoptotic cells was also observed, i.e, cells became rounded and detached from the substrate (Fig. 1e).

MLN2238 induction of G2/M cell cycle arrest and p27 and p21 upregulation in HCC cells
To clarify MLN2238-induced inhibitory effects on HCC cell proliferation, the cell cycle phase distributions of HepG2, Hep3B, and SNU475 cells were examined by flow cytometry analysis. Fig. 2a shows representative cell cycle phase distributions of the controls and cells treated at 250 and 500 nM MLN2238 concentrations. Quantitative analysis (Fig. 2a, right side) showed that MLN2238 treatment for 24 h induced a higher cell population in the G2/M phase, and a lower population in the G0/G1 phase than in the control.
Next, to explore the molecular mechanism triggering MLN2238-induced cycle arrest in cells, the expression of proteins involved in controlling G2/M phase cell cycle progression was examined. Treatment with 250 and 500 nM MLN2238 increased expression of the inhibitors of cyclin-dependent kinase p21 and p27 (Fig. 2b).

MLN2238 induces apoptotic cell death in HCC cells
We further investigated whether MLN2238 elicited apoptotic cell death in HepG2, Hep3B, and SNU475 cells, using flow cytometry analysis. The percentages of early and late apoptotic and necrotic cells increased in comparison with the control according to the doses administered, suggesting that MLN2238 increased cell death in HCC cells (Fig. 3a, b).
Apoptosis induction was also investigated by measuring the activation of caspase-3/7 and analyzing Poly(ADPribose) polymerase (PARP) cleavage. Fig. 3 shows that at 24 h of treatment an increases in caspase-3/7 activity (Fig. 3c) in all cell lines was observed, whereas PARP cleavage ( Fig. 3d) was observed only in HepG2 and SNU475 cells. However, in Hep3B cells PARP cleavage was observed after treatment for 48 h (Fig. 3d).
Proteasome inhibitors have been reported to activate apoptotic response through a specific caspase-dependent cleavage of β-catenin 25 . Therefore, we analyzed the expression of β-catenin following MLN2238 treatment. Fig. 3e shows that MLN2238 treatment triggered the cleavage of β-catenin in HepG2 and SNU475 cells.  Furthermore, the involvement of caspases in PARP and βcatenin cleavage after treatment with MLN2238 was confirmed by using z-VAD-fmk, a cell-permeable and irreversible pan-caspase inhibitor. PARP and β-catenin cleavage was inhibited by co-treatment with z-VAD-fmk and MLN2238 (Fig. 3e).
The above findings suggest that MLN2238 induced the apoptosis pathway.

MLN2238 induces ER stress
Several studies have established that proteasome inhibitors, including MLN2238, trigger the unfolded protein response and ER stress 4-7,26-28 . Therefore, we examined the effects of MLN2238 treatment on genes known to be regulated during ER stress response. As shown in Fig. 4, treatment with MLN2238 activated ER stress genes in HCC cells after 24 h. There was an upregulation of the ER stress-regulated genes, including GRP78, ATF4, CHOP and TRB3 (Fig. 4a), and XBP1 mRNA splicing was also induced (Fig. 4b).
In human HCC tissues, we recently observed an overexpression of nuclear protein-1 (NUPR1), a stressinducible protein involved in regulating cell survival, apoptosis, and invasiveness, as well as chemoresistance in HCC 29 . Therefore, expression levels of NUPR1 were also investigated after MLN2238 treatment. As shown in Fig. 4a, MLN2238 treatment increased NUPR1 mRNA expression in Hep3B and SNU475 cells after 24 h of treatment. These results confirm that MLN2238 treatment caused the activation of ER stress response in HCC cells.

MLN2238 treatment induces Mcl-1 upregulation
Numerous studies have demonstrated that proteasome inhibition can trigger the accumulation of myeloid cell leukemia-1 (Mcl-1), which is an anti-apoptotic protein belonging to the Bcl-2 family. To examine whether treatment with MLN2238 promotes the accumulation of Mcl-1 in HCC, HepG2, Hep3B, and SNU475 cells were exposed to 125, 250, and 500 nM of MLN2238 treatment for 24 h and with 500 nM for 24 and 48 h. Western blot analysis demonstrated that MLN2238 induced Mcl-1 increase in a dose-dependent manner in all HCC cell lines at 24 h (Fig. 5a), whereas at 48 h, the levels of Mcl-1 decreased in HepG2 and SNU475 cells, and remained higher than the control in Hep3B cells (Fig. 5b).
In addition, we evaluated whether treatment with MLN2238 induced the accumulation of other antiapoptotic proteins. However, western blot analysis of Bcl-2 protein expression demonstrated that MLN2238 induced Bcl-2 downregulation according to the treatment dose and duration (Fig. 5a, b).
Therefore, to explore the relationship between the expression of Mcl-1 and cell susceptibility to MLN2238 treatment, HCC cells were transfected with Mcl-1-specific small interference RNA (siRNA) to inhibit Mcl-1 gene expression. Transfection efficiency after 72 h of transfection with siMcl-1 was confirmed by a clear reduction in Mcl-1 protein expression levels (Fig. 5c, left side). Furthermore, MTS assays showed that Mcl-1 knockdown (KD) significantly sensitized Hep3B and SNU475 cells to MLN2238 treatment compared with cells transfected with control siRNA (siNC), whereas although a reduction of cell viability was observed in HepG2 cells after Mcl-1 KD, differences did not reach statistical significance (Fig. 5c, right side). In addition, in Hep3B cells silencing of Mcl-1 promoted MLN2238-induced PARP cleavage and βcatenin reduction already at 24 h (Fig. 5d). Overall, these results suggest that on MNL2238 treatment, Mcl-1induced expression protects HCC cells from MLN2238 antitumor effects.

Co-treatment with MLN2238 and Mcl-1 inhibitor A1210477 synergistically reduces cell viability in HCC cells
To explore the function played by Mcl-1 in MLN2238 resistance in HCC cells, we investigated whether the pharmacological inhibition of Mcl-1 with a potent and specific inhibitor, A1210477, would exert a similar effect to that obtained using the siRNA approach. We thus first determined the dose of A1210477 at which cell viability was minimally affected. At a dose of 10 μM A1210477 exerted only minimal effects on cell viability, with 90%, 85%, and 93% of cell survival in HepG2, Hep3B, and SNU475 cells, respectively (data not shown). Consequently, A1210477 alone (10 μM) or in combination with MLN2238 was used for cell treatment. Combinations of the two agents increased treatment efficacy compared with a single agent, as assessed by MTS assay at 24 h of treatment (Fig. 6). To define the type of interaction between the two agents, the combination index (CI) for each association was determined. As reported in Table 2, according to the CI, a combined treatment with different concentrations of MLN2238 and 10 μM A1210477 synergistically inhibited cell viability.

MLN2238 treatment inhibits in vivo tumor growth
A xenograft tumor model was utilized to demonstrate the antitumor efficacy of MLN2238 on HCC in vivo. Treatment with MLN2238 led to a significant reduction in tumor volume when compared with the volume measured in the control group (vehicle alone) (Fig. 7a).
Furthermore, drug-associated cytotoxicity was monitored by analyzing changes in the body weight of animals. Treatment of mice with MLN2238 did not significantly  modify body weight when compared to control group mice, which suggests that drug cytotoxicity was at a tolerable level (Fig. 7b).
To confirm the data observed in vitro, Mcl-1 and p21 expression in tumor tissues in mice receiving MLN2238 was investigated by western blot analysis (Fig. 7c). Mice treated with MLN2238 expressed higher levels of Mcl-1 and p21 than those of mice without the treatment. Therefore, these findings confirmed data previously observed in the in vitro experiments. In addition, histochemical analysis indicated that treatment with MLN2238 inhibited in tumor tissues the expression of Ki-67, a cell proliferation marker (Fig. 7d, e), confirming data obtained in vitro using a cell proliferation assay (bromodeoxyuridine (BrdU) assays).

Discussion
Proteasome inhibitors represent a new class of promising therapeutic drugs in cancer 2,3 . However, although several preclinical studies have shown that proteasome inhibition by bortezomib could represent a viable therapeutic strategy for treatment of HCC 10-12 , clinical results obtained in patients with unresectable HCC have been disappointing 13 .
By contrast, the good clinical results observed with bortezomib in other tumor types have led to the development of next-generation proteasome inhibitors, such as MLN2238 (ixazomib) 15,22 , which has demonstrated better pharmacokinetics, pharmacodynamics, and antitumor activity compared with bortezomib 15,30 . Recently, ixazomib was shown to inhibit the in vitro growth of  Here, we demonstrated the potent antitumor effects of MNL2238 in HCC in both in vitro as well as in vivo models. To our knowledge, this is the first study assessing the use of the next-generation proteasome inhibitor MLN2238 in HCC.
Consistent with studies by other authors using different proteasome inhibitors, we showed that the cytotoxic effects of MLN2238-induced ER stress, apoptosis and cell cycle arrest at the G2/M phase, with upregulated p21 and p27 proteins [18][19][20]22,28 .
However, although the IC 50 values of MLN2238 in Hep3B cells were lower than in the other two cell lines, the apoptotic response (in terms of PARP cleavage) to MLN2238 was delayed in time. These differences might be owing to different characteristics of differentiation, genetic defects, different ethnic origins, and to the biological behavior of these cell lines 25,31 , which are reflected in the tumor heterogeneity characterizing primary HCC. As an example: HepG2 cells carry a wild-type p53 gene, Hep3B cells are p53-null and SNU475 cells carry the mutated p53 gene. Interestingly, a recent study, by Dabiri et al. 32 , showed that p53-null colorectal cancer cells were transiently resistant, in the first 24 h, to the proteasome inhibitor bortezomib, whereas becoming responsive after 48 or 72 h of treatment, when the oncosuppressor isoform of p73, TAp73, translocated and accumulated in the nuclei of treated cells. On the other hand, KD of p73 in p53-null cells significantly prolonged the time of resistance to bortezomib. These results suggest that TAp73 was responsible for the delayed pro-apoptotic effects of bortezomib in cells lacking functional p53 32 . Therefore, although we did not analyze the role of p73 during MLN2238 treatment in Hep3B cells, it could be hypothesized that a mechanism similar to those described by Dabiri et al., could explain the transient MLN2238 resistance observed in these cells. This interesting aspect needs further investigation and could be clarified by future studies.
We observed that the expression of Mcl-1, an antiapoptotic protein, was upregulated upon treatment with  Viability after treating HCC cells with different doses of MLN2238 + 10 µM A1210477 for 24 h was evaluated by MTS assay. *P < 0.05, 10 µM A1210477 + MLN2238 vs. 10 µM A1210477 alone; # P < 0.005, 10 µM A1210477 + MLN2238 vs. 10 µM A1210477 alone MLN2238 in Hep3B cells after 24 h, and it also remained high at 48 h post-treatment, whereas expression declined at 48 h in the other two cell lines. It has been described that Mcl-1 accumulates as an undesired effect after exposure to proteasome inhibitors, and in addition, its expression decreases inhibitor efficacy 33,34 . Moreover, Mcl-1 expression continues in cells that have already undergone apoptotic response and is downregulated only at late time points after treatment 33 . This observation prompted us to assess the functional role of Mcl-1 in the apoptotic resistance to MLN2238 in Hep3B cells, especially considering that Mcl-1 has a pivotal role in HCC. Mcl-1 expression is higher in human HCC tissue than in the adjacent non-tumor liver 35,36 , and it also protects HCC cells against the drug-induced apoptosis of chemotherapeutic drugs 35,37 .
Here, we demonstrated that Mcl-1 KD increased HCC cell sensitivity to MLN2238 and shortened the apoptotic response timing of HCC cells. In addition, we presented a novel strategy to enhance the efficacy of MLN2238 in HCC via modulation of Mcl-1 activity by pharmacological inhibition, using the small molecule Mcl-1 inhibitor A1210477. Our results demonstrate that combined treatment with MLN2238 and A1210477 represents a novel strategy for inducing synergistic inhibitory effects on cell viability in HCC cells. These data are consistent with results obtained by others using different proteasome inhibitors in various tumor types 33,38 , including MLN2238 39 .
Therefore, although our study demonstrated that MLN2238 is cytotoxic to HCC cells, potential mechanisms of chemoresistance should be taken into consideration, owing to Mcl-1 upregulation. However, as we have demonstrated here, this could be overcome by a combination strategy, using a pharmacological approach to inhibit Mcl-1 with small molecule inhibitors.
In conclusion, our findings suggest that proteasome inhibitors are promising drugs to tackle HCC and justify further investigation and clinical validation.

Materials and methods
Cell lines, cell culture, and reagents The human HCC cell lines HepG2, Hep3B, and SNU475 were acquired from the American Type Culture Collection (ATCC) (HB-8065, HB-8064, and CRL-2236, respectively) and were maintained as previously described 40 . The cell lines were authenticated using short tandem repeat profiling (BMR Genomics, Padua, Italy). MLN2238, purchased from Selleck Chemicals (Houston, TX, USA), and A1210477, from Active Biochem (Bonn, Germany) were dissolved in dimethyl sulfoxide (DMSO).

Cell viability and proliferation assays
A total of 3 × 10 3 cells were added to 96-well plates and after 24 h were exposed to varying doses of MLN2238 alone, or combined with a subtoxic dose (10 µM) of Mcl-1 inhibitor A1210477. MTS assay was used to evaluate cell viability (see previous description) 41 . Data were recorded as a percentage of absorbance, comparing treated cells with controls (vehicle alone), and values expressed as the mean ± standard deviation (S.D.) of three experiments, each performed in triplicate.
To evaluate eventual effects of MLN2238 on cell proliferation, the incorporation of BrdU into DNA was measured by a colorimetric assay (Roche Diagnostics GmbH, Mannheim, Germany), as previously described 40 . Data are reported as the percentage of BrdU incorporation in treated cells compared with control cells. Values are reported as the mean ± S.D. of three experiments performed in triplicate.

Cell cycle and apoptosis analyses by flow cytometry
HepG2 (6 × 10 5 ), Hep3B (6 × 10 5 ), and SNU475 (4 × 10 5 ) cells were grown on 60-mm tissue culture dishes. At 24 h, the cells were exposed to MLN2238, and after 24 h flow cytometry was used for analyses (Becton Dickinson, Mountain View, CA, USA), performed as previously described 40 . The annexin V-FITC apoptosis kit (Dojindo Laboratories, Munich, Germany) was used to detect apoptosis, as previously described 42 . Results are presented as a percentage. The values indicated are the mean ± S.D. of two distinct experiments performed in triplicate.

siRNA transfection
For siRNA transfection, 3 × 10 5 Hep3B and HepG2 cells, and 1.8 × 10 5 cells SNU475 cells were distributed into 6-well plates, containing culture medium without antibiotics. At 24 h, this was substituted with fresh culture medium and 50 nmol/l of Mcl-1 siRNA (siMcl-1) was used to transfect cells. Control cell transfection was performed with a Negative Control siRNA (siNC). Transfections were carried out with Lipofectamine RNAiMax (Invitrogen, Carlsbad, CA, USA), following the manufacturers' instructions. siMcl-1 (SI02781205) and siNC (1027281) were acquired from QIAGEN (Germantown, MD, USA). After 24 h of transfection, cells were detached and seeded in 96-well plates or in 60-mm plates in preparation for MTS assay and for protein extraction, respectively.

Western blotting
RIPA buffer (Cell Signaling Technologies Inc., Beverly, MA, USA) was used to obtain cell lysates, and western blotting was performed as reported previously 41 . Primary antibodies against β-actin were obtained from SIGMA (Milan, Italy), Mcl-1 was from Santa Cruz (Dallas, TX, USA) and PARP, p21Waf1/Cip1, p27 from Cell Signaling Technologies (Beverly, MA, USA), while β-catenin was from BD Transduction Laboratories (San Jose, CA, USA.). The relative expressions were calculated as the ratio of drug-treated samples vs. control (DMSO) and corrected using the quantified level of β-actin expression.
RNA extraction, real-time, and semiquantitative PCR Total RNA was extracted using the TRIzol method and 1 µg of RNA was used for reverse transcription to generate cDNA using the QuantiNova Reverse Transcription Kit (QIAGEN), following the instructions of the manufacturer.
Quantitative Real-time PCR was analyzed using the QuantiNova SYBR Green PCR Kit (QIAGEN) and QuantiTect primers specific for Mcl-1 (QT00094122), NUPR1 (QT00088382), CHOP (QT00082278), GRP78 (QT00096404), TRB3 (QT00088543), and β-actin (QT00095431). Data are expressed as the relative mRNA expression level of the different genes in treated cells compared with control cells. Values given are the mean ± S.D. of three different experiments performed in triplicate.

In vivo studies
For the in vivo studies, MLN2238 was dissolved in 5% 2hydroxypropyl-β-cyclodextrin at 1 mg/ml concentration. Female athymic nude mice (Fox1 nu/nu) 4 weeks of age were acquired (Envigo, Udine, Italy). Animals were left to acclimatize for 1 week. An inoculation of 10 × 10 6 Hep3B cells in 200 µl of culture medium was made in the flank of the animal. When tumors reached the size of 150 mm 3 , the mice were randomly assigned to two groups of six animals each, receiving: 5% solution of 2-hydroxypropylβ-cyclodextrin, two times/week by oral gavage in the control group (C) and 11 mg/kg of MLN2238, two times/ week by oral gavage, in the treated MLN2238 group (T). Volumes of tumors and body weight were documented (see previous report) 41 .
After harvesting, each tumor was divided in half. One part was fixed in formalin and used for immunohistochemistry analysis, whereas the other was frozen in liquid nitrogen and stored at − 80°C for western blotting. All the procedures were performed abiding by institutional guidelines, which comply with national and international laws and policies. Authorization for the study was obtained from the Italian Ministry of Health (no. 1187/2015-PR).

Immunohistochemistry analysis
Immunohistochemical analysis was performed as reported previously 39 . To assess the proliferative index, nuclear antigen Ki-67 staining assay was carried out using a rabbit monoclonal antibody for Ki-67 (1:100) (Biocare Medical, CRM325). Expression level of Ki-67 was quantified as previously described 40 .

Statistical analysis
Student's t-test (two-tailed) was used for the statistical analysis. The statistical significance criterion was p < 0.05. To determine the synergistic effect, data analysis was performed using CalcuSyn Version 2.0 software (Biosoft, Cambridge, UK). CI <1 indicated synergy,~1 an additive effect and >1 antagonism. IC 50 was calculated by linear regression with the GraphPad Prism 6 software (Graph-Pad Software Inc., La Jolla, CA, USA).