Thioredoxin Downregulation Enhances Sorafenib Effects in Hepatocarcinoma Cells

antioxidants Article Thioredoxin Downregulation Enhances Sorafenib Effects in Hepatocarcinoma Cells María JoséLópez-Grueso 1, Raúl González 2, Jordi Muntané2,3,4 , JoséAntonio Bárcena 1,5,* and C. Alicia Padilla 1,5 1Department of Biochemistry and Molecular Biology, University of Córdoba, 14071 Córdoba, Spain; [email protected] (M.J.L.-G.); [email protected] (C.A.P.) 2 Institute of Biomedicine of Seville (IBiS), Hospital University “Virgen del Roc í o”/CSIC/University of Seville, 41013 Sevilla, Spain; [email protected] (R.G.); [email protected] (J.M.) 3Departamento de Cirugía General, Hospital Universitario Virgen del Rocío/Instituto de Biomedicina de Sevilla (IBiS)/CSIC/Universidad de Sevilla, 41013 Sevilla, Spain 4Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), 41013 Sevilla, Spain 5Instituto Maimónides de Investigación Biomédica de Córdoba (IMIBIC), 14004 Córdoba, Spain *Correspondence: bb1bar[email protected] Received: 11 September 2019; Accepted: 21 October 2019; Published: 22 October 2019   Abstract: Sorafenib is the first-line recommended therapy for patients with advanced hepatocarcinoma (HCC) in de-differentiation stage (presenting epithelial–mesenchymal transition, EMT). We studied the role of the thioredoxin system (Trx1/TrxR1) in the sensitivity or resistance of HCC cells to the treatment with Sorafenib. As a model, we used a set of three established HCC cell lines with different degrees of de-differentiation as occurs in metastasis. By quantitative proteomics, we found that the expression levels of Trx1 and TrxR1 followed the same trend as canonical EMT markers in these cell lines. Treatment with Sorafenib induced thiol redox reductive changes in critical elements of oncogenic pathways in all three cell lines but induced drastic proteome reprograming only in HCC cell lines of intermediate stage. Trx1 downregulation counteracted the thiol reductive effect of Sorafenib on Signal Transducer and Activator of Transcription 3 (STAT3) but not on Mitogen-Activated Protein Kinase (MAPK) or Protein Kinase B (Akt) and transformed advanced HCC cells into Sorafenib-sensitive cells. Ten targets of the combined Sorafenib–siRNATrx1 treatment were identified that showed a gradually changing expression trend in parallel to changes in the expression of canonical EMT markers, likely as a result of the activation of Hippo signaling. These findings support the idea that a combination of Sorafenib with thioredoxin inhibitors should be taken into account in the design of therapies against advanced HCC. Keywords: hepatocarcinoma; thioredoxin; sorafenib; redox signaling; EMT 1. Introduction Hepatocarcinoma (HCC) represents 80% of the primary hepatic neoplasms that appear mainly in a context of chronic liver cirrhosis. It is the sixth most frequent neoplasm, the third cause of cancer death, and accounts for 7% of registered malignancies [ 1 ]. Sorafenib is the standard of care for advanced-stage HCC, as demonstrated in two large-scale trials [2] and the Asia-Pacific trial [3]. Epithelial–mesenchymal transition (EMT) is an important process that happens in normal development in which epithelial cells lose many of their properties to become mesenchymal cells by means of drastic changes in architecture and behavior. A similar transition also occurs during tumor progression and malignant transformation, leading to increased cell motility and invasiveness. Antioxidants 2019,8, 501; doi:10.3390/antiox8100501 www.mdpi.com/journal/antioxidants Antioxidants 2019,8, 501 2 of 18 Transforming Growth Factor-ß (TGFB) is known as the main, although not exclusive, inductor of EMT, which takes place through various signaling pathways [ 4 ]. A critical molecular event of EMT is the downregulation of the cell adhesion molecule E-cadherin and its replacement by N-cadherin [ 5 ]. Activation of Akt leads to a significant reduction in E-cadherin expression and to nuclear localization of SNAI1, suggesting a role for the PI3K/Akt signaling pathway in the shift from E-cadherin to N-cadherin expression and in EMT progression in cancer [ 6 , 7 ]. The serine/threonine kinase Akt is thought to be responsible for mediating the acquired resistance to Sorafenib in HCC cells [ 8 ]. Akt can be activated by insulin and survival and growth factors through phosphorylation mediated by mTOR2 and PDK1 at specific threonine residues. The process can be reverted by protein phosphatase 2 (PP2A) when Akt is oxidized, suggesting that Akt is subjected to redox regulation [ 9 ]. Moreover, it was shown that the interaction between Akt and PP2A under conditions of oxidative stress could be impaired by the reduction of a specific disulfide bond in Akt catalyzed by glutaredoxin (Grx). Activation of Akt after stimulation with insulin and various growth and survival factors, leading to the phosphorylation at Thr308 and Ser473 by PDK1 and mTOR2, respectively. Akt in its oxidized form can also be inactivated by protein phosphatase 2 (PP2A) dephosphorylation, which suggests that Akt is a redox-regulated protein [ 9 ]. It has been shown that Grx prevented Akt from forming a specific disulfide bond between Cys-297 and Cys-311 and suppressed its association with PP2A under oxidative stress, [9–11]. STAT3 is a pro-survival transcription factor which is constitutively activated in human cancer cell lines. Constitutive phosphorylation of STAT3 causes important changes in apoptosis, angiogenesis, invasion, migration, and proliferation, resulting in cell malignant transformation [ 12 ]. It has been observed that STAT3 activation is involved in EMT, invasion, and generation of metastasis in HCC [ 13 ]. Tyrosine phosphorylation of STAT3 is dependent on the thiol redox state modulated by H 2 O 2 and Prx2 levels and the thioredoxin system activity [ 14 ]. Trx1–STAT3 disulfide exchange intermediates were detected, suggesting that Trx1 may be a direct mediator of STAT3 disulfide reduction. Overexpression of Trx1 has been detected in human cervical neoplastic squamous epithelial cells, lung, colon, and HCC tumors [15–18]. Targeting Trx1 for cancer therapy and as prognostic marker of HCC has been considered because of the observed relationship between Trx1 expression and tumor aggressiveness, although the mechanisms underlying this association are still not well known [ 19 ]. Overexpression of Trx1 is one of the mechanisms for drug resistance in HCC treatment. A possible strategy that is gathering strength is to combine Trx1 gene therapy with chemotherapy to increase the effectiveness of treatments in HCC [ 20 ]. Thioredoxin-interacting protein (TXNIP) is a member of the alpha-arrestin protein family that binds to the active site and counteracts the action of Trx1 functioning as tumor suppressor [ 21 ]. Downregulation of TXNIP exacerbates cancer progression, and its expression is actually reduced in various human cancer cells [ 22 ]. TXNIP deficiency enhances the induction of the Zinc finger proteins SNAI1 and SNAI2 by TGF-ß and promotes TGF-ß-induced EMT [23]. A comparative genomic characterization of 19 cell lines derived from HCC allowed clustering them into two groups (I and II) according to their gene expression patterns [24]. In this study, we selected three hepatocarcinoma cell lines, i.e., HepG2 belonging to group I and SNU423 and SNU475 belonging to group II, to carry out a comparative study of their basal proteome and of how it responds to Sorafenib treatment and/or thioredoxin downregulation. We also analyzed redox modifications and phosphorylation of Akt, MAPK, and STAT3. The results obtained support the use of a combination of Sorafenib with Trx1 inactivation in HCC therapy. 2. Materials and Methods 2.1. Materials HepG2 (HB-8065 ™ , ATCC/LGC Standards, SLU, Barcelona, Spain), SNU423 (CRL-2238, ATCC/LGC Standards) and SNU475 (CRL-2236, ATCC/LGC Standards) cell lines were obtained from American Type Culture Collection (ATCC; LGC Standards, S.L.U., Barcelona, Spain). Anti-Trx1 was obtained in-house from rabbit immunization with Trx1. Antibodies against STAT3 (#4904S), Antioxidants 2019,8, 501 3 of 18 MAPK (#9102), Thr 202 /Tyr 204 pMAPK (#4370P), Ser 473 pAkt (#4060S) were obtained from Cell Signaling Technology. Antibodies against Tyr 705 pSTAT3 (#sc-7993), TXNIP (#sc-271237), Akt1 (#sc-5298), and ß-actin (#sc-47778) were from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA). Anti-TrxR1 (#ab124954) was provided by Abcam, Inc. Secondary antibodies were from Sigma. ECL reagent was from GE Healthcare (Chicago, Illinois, USA); siRNA for Trx1 was from Dharmacon, Inc. (#L-006340-00; Lafayette, Colorado, USA). 2.2. Cell Growth Conditions Cell cultures were negative for mycoplasma contamination. Cells were maintained in Eagle’s minimum essential medium (EMEM), pH 7.4, supplemented with 10% FBS (#S181G-500, Biowest), sodium pyruvate (1mM) (#L0642-100, Biowest), and a penicillin–streptomycin–amphotericin solution (100U/mL–100 µ g/mL–0.25 µ g/mL) (#L0010-100, Biowest) at 37 ◦ C in a humidified incubator with 5% CO 2 . The cells were seeded at a density of 10 5 cells/cm 2 in 2D culture. When applied, Sorafenib was added to the cell culture at a concentration of 10 µ M, as set up in a previous study [ 25 ] and the cells were incubated for 24 h before cell lysis to measure cell proliferation and protein activities and expression and to perform proteomic analysis. Cells lysis was carried out using 50 mM HEPES pH 7.5, 5 mM EDTA, 150 mM NaCl, 1% NP-40, a proteases inhibitor cocktail (#P8340, Sigma-Aldrich), 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM NaF, and 1mM Na 3 VO 4 . The lysates were incubated on ice and vortexed for 15 s in four intervals of 5 min each. The samples were centrifuged at 13,000 rpm at 4◦C. The supernatants were collected for protein quantification and analysis. 2.3. Determination of Proliferation and Caspase-3 Activity Measurement Cell proliferation was analyzed using a colorimetric ELISA (Roche Applied Science, Penzberg, Germany). In total, 2 × 10 4 cells/cm 2 were cultured in 96-well plates at 37 ◦ C and 5% CO 2 . At the end of the experiment, the cells were incubated with a 10 µ M BrdU labelling solution for 6h at 37 ◦ C following the protocol recommended by the manufacturer. Caspase-3-associated activity was determined using the peptide-based substrate Ac-DEVD-AFC (100 µM), as described elsewere [10]. 2.4. SDS-PAGE and Western Blotting The protein expression levels of Trx1, TrxR1, TXNIP, STAT3, pSTAT3, Akt1, pAkt, MAPK, and pMAPK were determined by SDS-PAGE coupled to Western blotting analysis, following the same procedure and using the same antibody dilutions as described before [10]. 2.5. Silencing of Trx1 Human Trx1 was knocked down in HepG2, SNU423, and SNU475 cells using a pool of four specific siRNA in 6-well plates (20,000 cells/cm 2 ), according to the manufacturer’s recommendations (Dharmacon, GE Healthcare Life Sciences) and as described before for HepG2 cells [ 10 ]. Non-targeting negative controls were run. Silencing of Trx1 was always checked by Western blotting and activity assay to confirm that its levels were reduced by ≈80%. 2.6. Redox Mobility Shift Assay Detection of thiol redox changes in Akt, MAPK, and STAT3 proteins was performed by differentially labeling reduced cysteines with 5 mM N-ethylmaleimide (NEM) and reversibly oxidized cysteines by reduction with 5 mM tris (2-carboxyethyl) phosphine (TCEP), followed by labeling with 10 mM 4-acetamido-4’-maleimidylstilbene-2,2’-disulfonic acid (AMS) (Thermo Scientific Pierce), which increases the molecular weight of the native protein by 536.44 Da. The protocol was the same as described before [10]. Antioxidants 2019,8, 501 4 of 18 2.7. LC–MS/MS, Label-Free MS Protein Quantification, and Systems Analysis Proteomic analyses were performed at the Proteomics Facility, University of C ó rdoba (SCAI). The procedure started with a culture of 2 × 10 4 cells/cm 2 , and the protocol was the same as described previously [ 26 ]. Peptides were scanned and fragmented with the Thermo Orbitrap Fusion (Q-OT-qIT, Thermo Scientific) equipped with a nano-UHLC Ultimate 3000 (Dionex-Thermo Scientifics). The analysis of MS raw data was performed using the MaxQuant (v1.5.7.0) software [ 27 ] and the label-free quantification configuration described before [ 26 ]. Three RAW data files per sample from three separate experiments were analyzed. Human UniProtKB/Swiss-Prot protein database (February 2018 version) was searched for protein identification as described before [ 26 ]. The method used for the imputation of missing values was the mean imputation. Finally, only the conditions with 3 values per identification were considered and analyzed through ANOVA and Tukey’s post-hoc tests. A protein was considered differentially expressed if identified and quantified with at least 2 unique peptides and had a fold change of at least 1.50 and pvalue ≤0.05. The differentially expressed proteins together with the fold change values were analyzed with the online IPA software package (Qiagen, version 48207413) and the open software DAVID [28]. For IPA, each protein was mapped to its corresponding gene object in the Ingenuity Pathways Knowledge. The “core analysis” function was carried out considering direct and indirect relationships experimentally observed in all mammalian tissues and species, as well as all node types, data sources, and mutations. The list of significantly enriched canonical pathways, biological functions, and upstream factors is presented together with the activation or inhibition z-score values in a scale of colors. 2.8. Statistical Analysis Results are expressed as mean ± SD of data from ≥ 3 independent experiments. One-way ANOVA with the least significant difference Tukey’s test as post-hoc multiple comparison analysis with a single pooled variance was used for comparisons; output pvalue ranges were handled in GP style: 0.0332 (*), 0.0021 (**), 0.0002 (***), <0.0001 (****). The threshold for statistically significant differences was set at p adjust ≤0.05 value. 3. Results 3.1. Tracing the Proteomic Signature of EMT in Human Hepatocarcinoma Cell Lines A “label-free” quantitative proteomics analysis detected 1170 proteins with significant differences between HepG2, SNU423, and SNU475 cell lines (Supplementary File S1). This is the first comparative proteomic analysis carried out with these cell lines, and the results obtained for canonical markers of EMT showed increasing and decreasing gradients, in agreement with the classification of these human HCC cells as epithelial or mesenchymal [ 29 , 30 ]. These results strongly correlate with previous microarray and Western blotting analyses of human HCC cell types [ 29 , 30 ] and constitute a definitive proof of concept for our experimental approach. E-cadherin was not detected, likely because our proteomic protocol was not optimized for membrane proteins. The members of the Trx system Trx1 and TrxR1 were also present in increasing levels from HepG2 to SNU475 cells (Figure 1), which agrees with the role described for Trx1 as a pro-metastasis factor [31]. Antioxidants 2019,8, 501 5 of 18 Antioxidants 2019, 8, x FOR PEER REVIEW 5 of 18 Figure 1. Epithelial–mesenchymal transition (EMT) markers and thioredoxin system in three hepatocarcinoma (HCC) cells lines. Data for vimentin, alpha-fetoprotein, CD44 antigen, Trx1, and TrxR1 were retrieved from the quantitative proteomic analysis of HepG2, SNU423, and SNU475 cells (Supplementary File S1). The scale in the vertical axis is the relative intensity from the LC–MS/MS quantitative analysis and varies between proteins; the maximum value for each protein ranges from 3.25e + 007 for alpha-fetoprotein in HepG2 cells to 8.97e + 008 for Trx1 in SNU475 cells. (N = 3, individual values are shown). A system analysis of these 1170 differential proteins yielded significant enrichments in several canonical pathways (Figure 2A; the identities of the proteins are shown in Supplementary File S1). Integrin and actin cytoskeleton signaling, remodeling of epithelial adherent junction, and PI3K/Akt signaling, which have been described as being involved in EMT, were activated. On the reverse side, there was an overall inactivation of amino acid metabolism, fatty acid beta-oxidation, neurotransmitter catabolism, glutathione metabolism, and oxido-reduction processes. SNU423 and SNU275 cells showed similar activation and inactivation trends in many processes, although these processes were affected to a lesser extent in SNU423 cells, in parallel to their degree of mesenchymal properties. System analysis of 100 upregulated and 156 downregulated proteins common to the first two cell lines by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) biological processes using DAVID software confirmed and stressed these similarities (Figure S1). A search for upstream regulators brought to light differences and commonalities (Figure 2B). To name a few, MYCN and MITF transcription factors appeared to be activated in SNU423 but inactivated in SNU475 cells, whereas HIF1A and the KDM5A histone H3 demethylase behaved in the opposite way. Altogether, these results identified a number of proteins and regulatory elements related to dedifferentiation of HCC cells that could be an interesting focus for further research. Figure 1. Epithelial–mesenchymal transition (EMT) markers and thioredoxin system in three hepatocarcinoma (HCC) cells lines. Data for vimentin, alpha-fetoprotein, CD44 antigen, Trx1, and TrxR1 were retrieved from the quantitative proteomic analysis of HepG2, SNU423, and SNU475 cells (Supplementary File S1). The scale in the vertical axis is the relative intensity from the LC–MS/MS quantitative analysis and varies between proteins; the maximum value for each protein ranges from 3.25e +007 for alpha-fetoprotein in HepG2 cells to 8.97e +008 for Trx1 in SNU475 cells. (N=3, individual values are shown). A system analysis of these 1170 differential proteins yielded significant enrichments in several canonical pathways (Figure 2A; the identities of the proteins are shown in Supplementary File S1). Integrin and actin cytoskeleton signaling, remodeling of epithelial adherent junction, and PI3K/Akt signaling, which have been described as being involved in EMT, were activated. On the reverse side, there was an overall inactivation of amino acid metabolism, fatty acid beta-oxidation, neurotransmitter catabolism, glutathione metabolism, and oxido-reduction processes. SNU423 and SNU275 cells showed similar activation and inactivation trends in many processes, although these processes were affected to a lesser extent in SNU423 cells, in parallel to their degree of mesenchymal properties. System analysis of 100 upregulated and 156 downregulated proteins common to the first two cell lines by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) biological processes using DAVID software confirmed and stressed these similarities (Figure S1). A search for upstream regulators brought to light differences and commonalities (Figure 2B). To name a few, MYCN and MITF transcription factors appeared to be activated in SNU423 but inactivated in SNU475 cells, whereas HIF1A and the KDM5A histone H3 demethylase behaved in the opposite way. Antioxidants 2019,8, 501 6 of 18 Antioxidants 2019, 8, x FOR PEER REVIEW 6 of 18 Figure 2. Systems Biology analysis of differential proteomic data from three HCC cell lines. Quantitative data on protein levels in SNU423 and SNU475 cells relative to the levels in HepG2, the more differentiated cell line, were analyzed. Up- and down-regulated proteins were analyzed using the IPA software in terms of (A) canonical pathways based on the content of the Ingenuity Knowledge Base, and (B) Upstream Regulator Analysis to identify the cascade of upstream transcriptional regulators (see M&M section 2.7 for more details). The degree of activation is represented in a color scale as indicated. Red brackets have been added to highlight significant data commented on in the main text. 3.2. The Trx/TrxR/TXNIP System is Highly Sensitive to Sorafenib Treatment in HCC Cell Lines The basal levels of Trx1 and TrxR1 in control cells were markedly higher in mesenchymal SNU475 cells (Figure 1), but Sorafenib induced a significant decrease in their levels in all HCC cell lines (Figure 3A,B). TXNIP behaved in a reciprocal manner relative to Trx1 and TrxR1. It was upregulated both by Sorafenib and by siRNATrx1, either independently or jointly, in all three cell lines (Figure 3C). These results indicate that the downregulation of the Trx system is part of the antitumoral action mechanism of Sorafenib and support our hypothesis that co-treatment with Trx1 inactivation adjuvants would potentiate Sorafenib-based antitumoral therapy. Figure 2. Systems Biology analysis of differential proteomic data from three HCC cell lines. Quantitative data on protein levels in SNU423 and SNU475 cells relative to the levels in HepG2, the more differentiated cell line, were analyzed. Up- and down-regulated proteins were analyzed using the IPA software in terms of ( A ) canonical pathways based on the content of the Ingenuity Knowledge Base, and ( B ) Upstream Regulator Analysis to identify the cascade of upstream transcriptional regulators (see M&M Section 2.7 for more details). The degree of activation is represented in a color scale as indicated. Red brackets have been added to highlight significant data commented on in the main text. Altogether, these results identified a number of proteins and regulatory elements related to de-differentiation of HCC cells that could be an interesting focus for further research. 3.2. The Trx/TrxR/TXNIP System is Highly Sensitive to Sorafenib Treatment in HCC Cell Lines The basal levels of Trx1 and TrxR1 in control cells were markedly higher in mesenchymal SNU475 cells (Figure 1), but Sorafenib induced a significant decrease in their levels in all HCC cell lines (Figure 3A,B). TXNIP behaved in a reciprocal manner relative to Trx1 and TrxR1. It was upregulated both by Sorafenib and by siRNATrx1, either independently or jointly, in all three cell lines (Figure 3C). These results indicate that the downregulation of the Trx system is part of the antitumoral action mechanism of Sorafenib and support our hypothesis that co-treatment with Trx1 inactivation adjuvants would potentiate Sorafenib-based antitumoral therapy. Antioxidants 2019,8, 501 7 of 18 Antioxidants 2019, 8, x FOR PEER REVIEW 7 of 18 Figure 3. Response of the thioredoxin system to Sorafenib and siRNATrx1 treatments. Changes in the levels of three components of the thioredoxin system, (A) Trx1, (B) TrxR1 and (C) TXNIP were determined in HepG2, SNU423 and SNU475 cells by Western blotting with their specific antibodies on treatment with Sorafenib or siRNATrx1 independently or combined under the conditions described in M&M. Densitometric data normalized for ß-actin are shown; samples from 3 different experiments (N = 3) for each cell line were run in the same gel; protein levels between cells cannot be compared since each set of 3 replicas for each cell line was developed on a different WB membrane; for a comparison of Trx1, TrxR1 between cell lines, refer to quantitative proteomic data in Figure 1; a composition of representative blots is shown below each graph. p values 0.0332 (*), 0.0021 (**), 0.0002(***), <0.0001 (****), see Materials & Methods for statistical details. 3.3. The Effect of Sorafenib on the Global Proteome Depends on the Cell De-Differentiation Stage Sorafenib alone was very effective in SNU423 cells, inducing significant changes in 465 proteins, whereas in HepG2 and SNU475 cells, the number of proteins affected was restricted to only 23 and 83, respectively (Figure 4A). The effect of siRNATrx1 alone was rather low in all three cell types, as should be expected since the effects of this antioxidant protein at the transcriptional regulatory level are limited. The cellular environment determined which proteins were affected by Trx1 downregulation, since there were no common siRNATrx1 targets among the three cell lines, although most of the proteins affected did also change when siRNATrx1 treatment was combined with Sorafenib (Supplementary File S1). Interestingly, a very strong synergistic effect was observed when Sorafenib was applied to Trx1-downregulated SNU475 cells (Figure 4A). This synergistic effect was also observed in HepG2 cells, although to a lesser extent, but was absent in SNU423 cells. Figure 3. Response of the thioredoxin system to Sorafenib and siRNATrx1 treatments. Changes in the levels of three components of the thioredoxin system, ( A ) Trx1, ( B ) TrxR1 and ( C ) TXNIP were determined in HepG2, SNU423 and SNU475 cells by Western blotting with their specific antibodies on treatment with Sorafenib or siRNATrx1 independently or combined under the conditions described in M&M. Densitometric data normalized for ß-actin are shown; samples from 3 different experiments ( N=3 ) for each cell line were run in the same gel; protein levels between cells cannot be compared since each set of 3 replicas for each cell line was developed on a different WB membrane; for a comparison of Trx1, TrxR1 between cell lines, refer to quantitative proteomic data in Figure 1; a composition of representative blots is shown below each graph. pvalues 0.0332 (*), 0.0021 (**), 0.0002 (***), <0.0001 (****), see Materials & Methods for statistical details. 3.3. The Effect of Sorafenib on the Global Proteome Depends on the Cell De-Differentiation Stage Sorafenib alone was very effective in SNU423 cells, inducing significant changes in 465 proteins, whereas in HepG2 and SNU475 cells, the number of proteins affected was restricted to only 23 and 83, respectively (Figure 4A). The effect of siRNATrx1 alone was rather low in all three cell types, as should be expected since the effects of this antioxidant protein at the transcriptional regulatory level are limited. The cellular environment determined which proteins were affected by Trx1 downregulation, since there were no common siRNATrx1 targets among the three cell lines, although most of the proteins affected did also change when siRNATrx1 treatment was combined with Sorafenib (Supplementary File S1). Interestingly, a very strong synergistic effect was observed when Sorafenib was applied to Trx1-downregulated SNU475 cells (Figure 4A). This synergistic effect was also observed in HepG2 cells, although to a lesser extent, but was absent in SNU423 cells. Antioxidants 2019,8, 501 8 of 18 Antioxidants 2019, 8, x FOR PEER REVIEW 8 of 18 Figure 4. Summary of the quantitative proteomic analysis of HCC cell lines subjected to Sorafenib and/or siRNATrx1 treatments. (A) The total number of significantly (adjusted p ≤ 0.05) different proteins (fold change ≥1.5 and ≤0.6) quantified upon each treatment relative to their controls; the numbers above the brackets and thin lines are the “synergy factor” reached by siRNATrx1 and Sorafenib calculated by the ratio between the number of proteins affected by the combined treatment and the number of proteins affected by individual Sorafenib and siRNATrx1 treatments. (B) Venn diagrams of upregulated and downregulated proteins by each treatment, by cell line; the color code indicated in the legend applies to (A) and (B). (C) Venn diagrams of proteins affected in each cell line by treatment. Altogether, these proteomic data indicate that Sorafenib is highly active in moderately differentiated tumor cells but it has a limited effect in Trx1-rich and poorly differentiated cells, unless the antioxidant thioredoxin-related signaling is weakened. Its action is minor in epithelial HepG2 cells even in combination with Trx1 downregulation. A comparison of the Sorafenib and Trx1 silencing targets of each cell line (Figure 4C) highlighted the fact that there were no common targets between the se cell lines, suggesting that the effects of each treatment are dependent on the de-differentiation stage of HCC cell lines. However, there were 12 common targets of the combined treatment in the three cell lines (Figure 4C and Figure 5). These common proteins are related to lipids biosynthesis, protein nuclear transport, and actin organization. The increasing degree of downregulation of most of these common proteins runs parallel to their degree of de-differentiation. Figure 4. Summary of the quantitative proteomic analysis of HCC cell lines subjected to Sorafenib and/or siRNATrx1 treatments. ( A ) The total number of significantly (adjusted p ≤ 0.05) different proteins (fold change ≥ 1.5 and ≤ 0.6) quantified upon each treatment relative to their controls; the numbers above the brackets and thin lines are the “synergy factor” reached by siRNATrx1 and Sorafenib calculated by the ratio between the number of proteins affected by the combined treatment and the number of proteins affected by individual Sorafenib and siRNATrx1 treatments. ( B) Venn diagrams of upregulated and downregulated proteins by each treatment, by cell line; the color code indicated in the legend applies to (A) and (B). (C) Venn diagrams of proteins affected in each cell line by treatment. Altogether, these proteomic data indicate that Sorafenib is highly active in moderately differentiated tumor cells but it has a limited effect in Trx1-rich and poorly differentiated cells, unless the antioxidant thioredoxin-related signaling is weakened. Its action is minor in epithelial HepG2 cells even in combination with Trx1 downregulation. A comparison of the Sorafenib and Trx1 silencing targets of each cell line (Figure 4C) highlighted the fact that there were no common targets between the se cell lines, suggesting that the effects of each treatment are dependent on the de-differentiation stage of HCC cell lines. However, there were 12 common targets of the combined treatment in the three cell lines (Figures 4C and 5). These common proteins are related to lipids biosynthesis, protein nuclear transport, and actin organization. The increasing degree of downregulation of most of these common proteins runs parallel to their degree of de-differentiation. Antioxidants 2019,8, 501 9 of 18 Antioxidants 2019, 8, x FOR PEER REVIEW 9 of 18 Figure 5. Common target proteins of the Sorafenib + siRNATrx1 treatment. Twelve proteins that changed significantly in HepG2, SNU423, and SNU475 cells are shown. A brief description of their function was extracted from UniProt; fold changes relative to their respective controls treated with solvent + Non-target siRNA are indicated in a color scale. Quantitative proteomic analysis as described in Materials and Methods (N = 3). Tenfold-change values correspond to proteins that were present in the controls but were not detected in the treated samples. System analysis of the proteomic data provided some clues about the differential effect of Sorafenib in each cell type (Figure 6). The only pathways altered by Sorafenib in SNU475 cells were those of IGF-1 and renin–angiotensin signaling, which appeared inactivated with the concomitant differential inactivation of several upstream regulators, i.e., NFE2L2, PDGF-B, a potent mitogen for cells of mesenchymal origin, XBP1, involved in the unfolded protein response (UPR), and Ige (Figure 6). None of these pathways and upstream regulators were affected in SNU423 cells. Figure 5. Common target proteins of the Sorafenib +siRNATrx1 treatment. Twelve proteins that changed significantly in HepG2, SNU423, and SNU475 cells are shown. A brief description of their function was extracted from UniProt; fold changes relative to their respective controls treated with solvent +Non-target siRNA are indicated in a color scale. Quantitative proteomic analysis as described in Materials and Methods (N=3). Tenfold-change values correspond to proteins that were present in the controls but were not detected in the treated samples. System analysis of the proteomic data provided some clues about the differential effect of Sorafenib in each cell type (Figure 6). The only pathways altered by Sorafenib in SNU475 cells were those of IGF-1 and renin–angiotensin signaling, which appeared inactivated with the concomitant differential inactivation of several upstream regulators, i.e., NFE2L2, PDGF-B, a potent mitogen for cells of mesenchymal origin, XBP1, involved in the unfolded protein response (UPR), and Ige (Figure 6). None of these pathways and upstream regulators were affected in SNU423 cells. Antioxidants 2019,8, 501 16 of 18 siRNATrx1 combined treatment (“ST”); columns P–U show the abundance fold change (“FC”) relative to the control and their respective statistical significance (p-values, a value of “1” means not significant FC). In the 4 th sheet, (“Cell lines”), columns A–D show protein ID, protein name, gene name, and number of unique peptides; columns E–M show the normalized abundance for each of three replicates of untreated HepG2 cells (“HepG2”), SNU423 cells (“S3”), and SNU475 cells (“S5”); columns N–Q show the protein abundance fold change of SNU423 and SNU475 cells (“FCS3_HepG2” and “FCS5_HepG2”) relative to HepG2 together with the statistical significance index (p value). A value of “1” means not significant FC. Author Contributions: Conceptualization, J.M., J.A.B., and C.A.P.; Formal analysis, M.J.L.-G. and R.G.; Funding acquisition, J.M., J.A.B., and C.A.P.; Investigation, M.J.L.-G. and R.G.; Methodology, M.J.L.-G. and R.G.; Software, M.J.L.-G.; Writing—original draft, J.A.B., and C.A.P.; Writing—review & editing, J.M. Funding: This research has been financed by grants from the Spanish Ministry of Economy and Competitiveness (BFU2016-80006-P), the Institute of Health Carlos III (ISCIII) (PI13/00021 and PI16/00090), and the Andalusian Government (Consejer í a de Econom í a, Innovaci ó n, Ciencia y Empleo, BIO-0216 and CTS-6264; Consejer í a de Igualdad, Salud y Pol í ticas Sociales, PI-00025-2013, and PI-0198-2016). We thank the Biomedical Research Network Center for Liver and Digestive Diseases (CIBERehd) founded by the ISCIII and co-financed by European Development Regional Fund “A way to achieve Europe” ERDF for their financial support. Acknowledgments: Technical support by the staffof the Proteomics facility, Central Service for Research Support (SCAI) at the University of Cordoba, and the computer resources, technical expertise and assistance provided by the PAB (Andalusian Bioinformatics Platform) center located at the University of Málaga are acknowledged. Conflicts of Interest: The authors declare no conflict of interest. References 1. Galle, P.R.; Forner, A.; Llovet, J.M.; Mazzaferro, V.; Piscaglia, F.; Raoul, J.L.; Schirmacher, P.; Vilgrain, V. EASL Clinical Practice Guidelines: Management of hepatocellular carcinoma. J. Hepatol. 2018 ,69, 182–236. [CrossRef] [PubMed] 2. Llovet, J.M.; Ricci, S.; Mazzaferro, V.; Hilgard, P.; Gane, E.; Blanc, J.F.; De Oliveira, A.C.; Santoro, A.; Raoul, J.L.; Forner, A.; et al. Sorafenib in Advanced Hepatocellular Carcinoma. N. Engl. J. Med. 2008 ,359, 378–390. [CrossRef] [PubMed] 3. Cheng, A.L.; Kang, Y.K.; Chen, Z.; Tsao, C.J.; Qin, S.; Kim, J.S.; Luo, R.; Feng, J.; Ye, S.; Yang, T.S.; et al. Efficacy and safety of sorafenib in patients in the Asia-Pacific region with advanced hepatocellular carcinoma: A phase III randomised, double-blind, placebo-controlled trial. Lancet Oncol. 2009,10, 25–34. [CrossRef] 4. Lamouille, S.; Xu, J.; Derynck, R. Molecular mechanisms of epithelial-mesenchymal transition. Nat. Biotechnol. 2014,15, 178–196. [CrossRef] [PubMed] 5. Thiery, J.; Sleeman, J. Complex networks orchestrate epithelial–mesenchymal transitions. Nat. Rev. Mol. Cell Biol. 2006,7, 131–142. [CrossRef] [PubMed] 6. Barber, A.G.; Castillo-Martin, M.; Bonal, D.M.; Jia, A.J.; Rybicki, B.A.; Christiano, A.M.; Cordon-Cardo, C. PI3K/AKT pathway regulates E-cadherin and Desmoglein 2 in aggressive prostate cancer. Cancer Med. 2015 , 8, 1258–1271. [CrossRef] 7. Larue, L.; Bellacosa, A. Epithelial-mesenchymal transition in development and cancer: Role of phosphatidylinositol 3ºkinase/AKT pathways. Oncogene 2005,24, 7443–7454. [CrossRef] 8. Dong, J.; Zhai, B.; Sun, W.; Hu, F.; Cheng, H.; Xu, J. Activation of phosphatidylinositol 3-kinase/AKT/snail signaling pathway contributes to epithelial-mesenchymal transition-induced multi-drug resistance to sorafenib in hepatocellular carcinoma cells. PLoS ONE 2017,12, e0185088. [CrossRef] 9. Murata, H.; Ihara, Y.; Nakamura, H.; Yodoi, J.; Sumikawa, K.; Kondo, T. Glutaredoxin exerts an antiapoptotic effect by regulating the redox state of Akt. J. Biol. Chem. 2003,278, 50226–50233. [CrossRef] 10. Gonz á lez, R.; L ó pez-Grueso, M.J.; Muntan é , J.; B á rcena, J.A.; Padilla, C.A. Redox regulation of metabolic and signaling pathways by thioredoxin and glutaredoxin in NOS-3 overexpressing hepatoblastoma cells. Redox Biol. 2015,6, 122–134. 11. Huang, X.; Begley, M.; Morgenstern, K.A.; Gu, Y.; Rose, P.; Zhao, H.; Zhu, X. Crystal Structure of an InactiveAkt2 Kinase Domain. Structure 2003,11, 21–30. [CrossRef] 12. Linher-Melville, K.; Singh, G. The complex roles of STAT3 and STAT5 in maintaining redox balance: Lessons from STAT-mediated xCT expression in cancer cells. Mol. Cell. Endocrinol. 2017 ,451, 40–52. [CrossRef] [PubMed] Antioxidants 2019,8, 501 17 of 18 13. Zhang, C.H.; Guo, F.L.; Xu, G.L.; Jia, W.D.; Ge, Y.S. STAT3 activation mediates epithelial-to-mesenchymal transition in human hepatocellular carcinoma cells. Hepatogastroenterology 2014,61, 1082–1089. [PubMed] 14. Sobotta, M.C.; Liou, W.; Stöcker, S.; Talwar, D.; Oehler, M.; Ruppert, T.; Scharf, A.N.; Dick, T.P. Peroxiredoxin-2 and STAT3 form a redox relay for H2O2signaling. Nat. Chem. Biol. 2015,11, 64–71. [CrossRef] 15. Gasdaska, P.Y.; Oblong, J.E.; Cotgreave, I.A.; Powis, G. The predicted amino acid sequence of human thioredoxin is identical to that of the autocrine growth factor human adult T-cell derived factor (ADF): Thioredoxin mRNA is elevated in some human tumors. Biochim. Biophys. Acta 1994 ,1218, 292–296. [CrossRef] 16. Berggren, M.; Gallegos, A.; Gasdaska, J.R.; Gasdaska, P.Y.; Warneke, J.; Powis, G. Thioredoxin and thioredoxin reductase gene expression in human tumors and cell lines, and the effects of serum stimulation and hypoxia. Anticancer Res. 1996,16, 3459–3466. 17. Fujii, S.; Nanbu, Y.; Nonogaki, H.; Konishi, I.; Mori, T.; Masutani, H.; Yodoi, J. Coexpression of adult T-cell leukemia-derived factor, a human thioredoxin homologue, and human papillomavirus DNA in neoplastic cervical squamous epithelium. Cancer 1991,68, 1583–1591. [CrossRef] 18. Nakamura, H.; Masutani, H.; Tagaya, Y.; Yamauchi, A.; Inamoto, T.; Nanbu, Y.; Fujii, S.; Ozawa, K.; Yodoi, J. Expression and growth-promoting effect of adult T-cell leukemia-derived factor. A human thioredoxin homologue in hepatocellular carcinoma. Cancer 1992,69, 2091–2097. [CrossRef] 19. Reichl, P.; Mikulits, W. Accuracy of novel diagnostic biomarkers for hepatocellular carcinoma: An update for clinicians (Review). Oncol. Rep. 2016,36, 613–625. [CrossRef] 20. Li, W.; Shi, J.; Zhang, C.; Li, M.; Gan, L.; Xu, H.; Yang, X. Co-delivery of thioredoxin 1 shRNA and doxorubicin by folate-targeted gemini surfactant-based cationic liposomes to sensitize hepatocellular carcinoma cells. J. Mater. Chem. B 2014,2, 4901–4910. [CrossRef] 21. Nishiyama, A.; Matsui, M.; Iwata, S.; Hirota, K.; Masutani, H.; Nakamura, H.; Takagi, Y.; Sono, H.; Gon, Y.; Yodoi, J. Identification of thioredoxin-binding protein-2/vitamin D(3) up-regulated protein 1 as a negative regulator of thioredoxin function and expression. J. Biol. Chem. 1999 ,274, 21645–21650. [CrossRef] [PubMed] 22. Zhou, J.; Chng, W.J. Roles of thioredoxin binding protein (TXNIP) in oxidative stress, apoptosis and cancer. Mitochondrion 2013,13, 163–169. [CrossRef] [PubMed] 23. Masaki, S.; Masutani, H.; Yoshihara, E.; Yodoi, J. Deficiency of Thioredoxin Binding Protein-2 (TBP-2) Enhances TGF-βSignaling and Promotes Epithelial to Mesenchymal Transition. PLoS ONE 2012,7, e39900. [CrossRef] [PubMed] 24. Zimonjic, D.B.; Keck, C.L.; Thorgeirsson, S.S.; Popescu, N.C. Novel recurrent genetic imbalances in human hepatocellular carcinoma cell lines identified by comparative genomic hybridization. Hepatology 1999 ,29, 1208–1214. [CrossRef] [PubMed] 25. Rodr í guez-Hern á ndez, M.A.; Gonz á lez, R.; de la Rosa, Á .J.; Gallego, P.; Ord ó ñez, R.; Navarro-Villar á n, E.; Contreras, L.; Rodr í guez-Arribas, M.; Gonz á lez-Gallego, J.; Á lamo-Mart í nez, J.M.; et al. Molecular characterization of autophagic and apoptotic signaling induced by sorafenib in liver cancer cells. J. Cell Physiol. 2018,234, 692–708. [CrossRef] [PubMed] 26. L ó pez-Grueso, M.J.; Gonz á lez-Ojeda, R.; Requejo-Aguilar, R.; McDonagh, B.; Fuentes-Almagro, C.A.; Muntan é , J.; B á rcena, J.A.; Padilla, C.A. Thioredoxin and glutaredoxin regulate metabolism through different T multiplex thiol switches. Redox Biol. 2019,21, 101049. 27. Cox, J.; Mann, M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 2008,26, 1367–1372. [CrossRef] 28. Huang, D.; Sherman, B.; Lempicki, R. Systematic and integrative analysis of large gene lists using DAVID Bioinformatics Resources. Nat. Protoc. 2009,4, 44–57. [CrossRef] 29. Lee, J.; Thorgeirsson, S. Functional and genomic implications of global gene expression profiles in cell lines from human hepatocellular cancer. Hepatology 2002,35, 1134–1143. [CrossRef] 30. Fuchs, B.C.; Fujii, T.; Dorfman, J.D.; Goodwin, J.M.; Zhu, A.X.; Lanuti, M.; Tanabe, K.K. Epithelial-to-Mesenchymal Transition and Integrin-Linked Kinase Mediate Sensitivity to Epidermal Growth Factor Receptor Inhibition in Human Hepatoma Cells. Cancer Res. 2008,68, 2391–2399. [CrossRef] 31. Guo, Z.; Cao, M.; You, A.; Gao, J.; Zhou, H.; Li, H.; Cui, Y.; Fang, F.; Zhang, W.; Song, T.; et al. Metformin inhibits the prometastatic effect of sorafenib in hepatocellular carcinoma by upregulating the expression of TIP30. Cancer Sci. 2016,107, 507–513. [CrossRef] [PubMed] 32. Misra, J.R.; Irvine, K.D. The Hippo Signaling Network and Its Biological Functions. Ann. Rev. Genet. 2018 , 52, 65–87. [CrossRef] [PubMed] Antioxidants 2019,8, 501 18 of 18 33. Lachaier, E.; Louandre, C.; Godin, C.; Saidak, Z.; Baert, M.; Diouf, M.; Chauffert, B.; Galmiche, A. Sorafenib Induces Ferroptosis in Human Cancer Cell Lines Originating from Different Solid Tumors. Anticancer Res. 2014,34, 6417–6422. [PubMed] 34. Bruix, J.; Reig, M.; Sherman, M. Evidence-Based Diagnosis, Staging, and Treatment of Patients with Hepatocellular Carcinoma. Gastroenterology 2016,150, 835–853. [CrossRef] 35. Stravitz, R.T.; Heuman, D.M.; Chand, N.; Sterling, R.K.; Shiffman, M.L.; Luketic, V.A.; Sanyal, A.J.; Habib, A.; Mihas, A.A.; Giles, H.C.S.; et al. Surveillance for hepatocellular carcinoma in patients with cirrhosis improves outcome. Am. J. Med. 2008,121, 119–126. [CrossRef] 36. Mikulits, W. Epithelial to mesenchymal transition in hepatocellular carcinoma. Future Oncol. 2009 , 5, 1169–1179. 37. Sciacovelli, M.; Frezza, C. Metabolic reprogramming and ephitelial-to-mesenchymal transition in cancer. FEBS J. 2017,284, 3132–3144. [CrossRef] 38. Wang, H.; Chen, Y.; Wu, G. SDHB deficiency promotes TGF β -mediated invasion and metastasis of colorectal cancer through transcriptional repression complex SNAIL1-SMAD3/4. Transl. Oncol. 2016 ,9, 512–520. [CrossRef] 39. Aspuria, P.J.P.; Lunt, S.Y.; Väremo, L.; Vergnes, L.; Gozo, M.; Beach, J.A.; Salumbides, B.; Reue, K.; Wiedemeyer, W.R.; Nielsen, J.; et al. Succinate dehydrogenase inhibition leads to epithelial-mesenchymal transition and reprogrammed carbon metabolism. Cancer Metab. 2014,2, 21. [CrossRef] 40. Den Hollander, P.; Rawls, K.; Tsimelzon, A.; Shepherd, J.; Mazumdar, A.; Hill, J.; Fuqua, S.A.; Chang, J.C.; Osborne, C.K.; Hilsenbeck, S.G.; et al. Phosphatase PTP4A3 Promotes Triple-Negative Breast Cancer Growth and Predicts Poor Patient Survival. Cancer Res. 2016,76, 1942–1953. [CrossRef] 41. Lu, L.; Li, Y.; Kim, S.M.; Bossuyt, W.; Liu, P.; Qiu, Q.; Wang, Y.; Halder, G.; Finegold, M.J.; Lee, J.S.; et al. Hippo signaling is a potent in vivo growth and tumorsuppressor pathway in the mammalian liver. Proc. Natl. Acad. Sci. USA 2010,107, 1437–1442. [CrossRef] [PubMed] 42. Khanal, P.; Jia, Z.; Yang, X. Cysteine residues are essential for dimerization of Hippo pathway components YAP2L and TAZ. Sci. Rep. 2018,8, 3485. [CrossRef] [PubMed] 43. Gandhirajan, R.K.; Jain, M.; Walla, B.; Johnsen, M.; Bartram, M.P.; Anh, M.H.; Rinschen, M.M.; Benzing, T.; Schermer, B. Cysteine S-Glutathionylation Promotes Stability and Activation of the Hippo Downstream Effector Transcriptional Co-activator with PDZ-binding Motif (TAZ). J. Biol. Chem. 2016 ,291, 11596–11607. [CrossRef] [PubMed] 44. Carballo, M.; Conde, M.; El Bekay, R.; Mart í n-Nieto, J.; Camacho, M.J.; Monteseir í n, J.; Conde, J.; Bedoya, F.J.; Sobrino, F. Oxidative stress triggers STAT3 tyrosine phosphorylation and nuclear translocation in human lymphocytes. J. Biol. Chem. 1999,274, 17580–17586. [CrossRef] [PubMed] 45. Li, L.; Cheung, S.H.; Evans, E.L.; Shaw, P.E. Modulation of Gene Expression and Tumor Cell Growth by Redox Modification of STAT3. Cancer Res. 2010,70, 8222–8232. [CrossRef] 46. Xie, Y.; Kole, S.; Precht, P.; Pazin, M.J.; Bernier, M. S-Glutathionylation impairs signal transducer and activator of trancription 3 activation and signaling. Endocrinology 2009,150, 1122–1131. [CrossRef] 47. Butturini, E.; Darra, E.; Chiavegato, G.; Cellini, B.; Cozzolino, F.; Monti, M.; Pucci, P.; Dell’Orco, D.; Mariotto, S. S-Glutathionylation at Cys328 and Cys542 Impairs STAT3 Phosphorylation. ACS Chem. Biol. 2014,9, 1885–1893. [CrossRef] 48. Wan, J.; Liu, T.; Mei, L.; Li, J.; Gong, K.; Yu, C.; Li, W. Synergistic antitumour activity of sorafenib in combination with tetrandrine is mediated by reactive oxygen species (ROS)/Akt signaling. Br. J. Cancer 2013 , 109, 342–350. [CrossRef] © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).