Integrated Stable Isotope Labeling by Amino Acids in Cell Culture (SILAC) and Isobaric Tags for Relative and Absolute Quantitation (iTRAQ) Quantitative Proteomic Analysis Identifies Galectin-1 as a Potential Biomarker for Predicting Sorafenib Resistance in Liver Cancer*

Sorafenib has become the standard therapy for patients with advanced hepatocellular carcinoma (HCC). Unfortunately, most patients eventually develop acquired resistance. Therefore, it is important to identify potential biomarkers that could predict the efficacy of sorafenib. To identify target proteins associated with the development of sorafenib resistance, we applied stable isotope labelling with amino acids in cell culture (SILAC)-based quantitative proteomic approach to analyze differences in protein expression levels between parental HuH-7 and sorafenib-acquired resistance HuH-7 (HuH-7R) cells in vitro, combined with an isobaric tags for relative and absolute quantitation (iTRAQ) quantitative analysis of HuH-7 and HuH-7R tumors in vivo. In total, 2,450 quantified proteins were identified in common in SILAC and iTRAQ experiments, with 81 showing increased expression (>2.0-fold) with sorafenib resistance and 75 showing decreased expression (<0.5-fold). In silico analyses of these differentially expressed proteins predicted that 10 proteins were related to cancer with involvements in cell adhesion, migration, and invasion. Knockdown of one of these candidate proteins, galectin-1, decreased cell proliferation and metastasis in HuH-7R cells and restored sensitivity to sorafenib. We verified galectin-1 as a predictive marker of sorafenib resistance and a downstream target of the AKT/mTOR/HIF-1α signaling pathway. In addition, increased galectin-1 expression in HCC patients' serum was associated with poor tumor control and low response rate. We also found that a high serum galectin-1 level was an independent factor associated with poor progression-free survival and overall survival. In conclusion, these results suggest that galectin-1 is a possible biomarker for predicting the response of HCC patients to treatment with sorafenib. As such, it may assist in the stratification of HCC and help direct personalized therapy.

the only effective systemic drug for such patients. Sorafenib is a multikinase inhibitor that targets Raf kinase, vascular endothelial growth factor receptor (VEGFR) and platelet-derived growth factor receptor (PDGFR), showing activity against both tumor cell proliferation and tumor angiogenesis (5). In the pivotal SHARP study and subsequent Asia-Pacific Study, sorafenib improved the median overall survival by 2-3 months in patients with advanced HCC (3,6). Despite this significant improvement in survival, the efficacy of sorafenib against HCC is modest, with an objective tumor response rate as low as 2% to 3% (3). In other words, many HCC patients are inherently resistant to sorafenib. For those who show an initial response or stabilization to sorafenib, disease progression inevitably ensues, indicating development of acquired resistance. Therefore, it is imperative to identify biomarkers that can predict the efficacy of sorafenib and outcomes in advanced HCC patients. Further, targeting drug resistance mechanisms of sorafenib may lead to the development of novel strategies to improve the efficacy of sorafenib in HCC.
Mass spectrometry-based proteomic technology is currently used to study and compare the proteomes of in vitro and in vivo models of cancer as well as patient tumors, and has opened up new avenues for tumor-associated biomarker discovery. A number of studies have employed this tool to examine drug resistance, and have revealed significant differences in the expression of proteins associated with key biological processes, such as cell proliferation, survival, and motility (7). Because they facilitate the simultaneous analysis of whole proteomes, proteomic technologies have led to the identification of various biomarkers associated with resistance to anticancer therapy (8). A number of studies have sought to identify tumor and/or plasma biomarkers that could be used to predict clinical benefit for patients with advanced HCC receiving sorafenib therapy (9). Changes in biomarker concentrations during treatment may predict drug response and provide insights into mechanisms of drug action or patient resistance. There is thus an urgent need to identify predictive biomarkers that could exclude advanced HCC patients who are unlikely to benefit from sorafenib therapy.
In the present study, we used quantitative proteomics to analyze parental HuH-7 and sorafenib-acquired resistance HuH-7 R HCC cell lines using the stable isotope labeling with amino acid in cell culture (SILAC) approach. We further extended this approach by incorporating HCC xenograft models using isobaric tags for relative and absolute quantitation (iTRAQ) quantitative analysis. This approach allowed the identification of 10 proteins involved in cell motility or invasion processes that were differentially expressed between HuH-7 and HuH-7 R cells. Among these proteins, galectin-1 was identified as a predictive marker for sorafenib resistance and a downstream target of the AKT/mTOR/HIF-1␣ signaling pathway. These results reveal a new role for galectin-1 in sorafenib resistance that could be of therapeutic value in the detection of sorafenib-resistant HCCs. We believe that the results of this study could provide additional insight into the mechanisms underlying the sensitivity and resistance to sorafenib in HCC cells. This, in turn, may help identify possible novel therapeutic targets, as well as biomarkers that aid patient stratification for optimal therapy.

EXPERIMENTAL PROCEDURES
Cell Lines, Tumor Models, and Transfection-The HCC HuH-7 cell line was obtained from the Health Science Research Resources Bank (JCRB0403, Osaka, Japan). The sorafenib-resistant HCC cell line, HuH-7 R , was established by long-term exposure of cells to sorafenib as previously reported (10).
The Institutional Laboratory Animal Care and Use Committee of National Taiwan University approved the animal studies. The tumor xenograft model was prepared by subcutaneously injecting 5 ϫ 10 6 HuH-7 or HuH-7 R cells into 5-week-old BALB/c nude mice. Tumor dimensions were measured with a caliper at 3-day intervals, and tumor volume was calculated as length ϫ width ϫ height (in cm 3 ). For the tail vein inoculation model, 1 ϫ 10 6 HuH-7 or HuH-7 R cells were injected by tail vein and mice were sacrificed after 6 weeks. Paraffinembedded, hematoxylin and eosin (H&E)-stained lung sections were analyzed microscopically for tumor nodules.
Target sequences used for galectin-1 knockdown experiments are listed in supplemental Table S1. Lentiviruses expressing small hairpin (inhibitory) RNA (shRNA) against galectin-1 (shGal-1) or control shRNA (shCtrl) was produced in HEK293FT cells. Medium containing shGal-1 or shCtrl viruses was applied to cultures of HuH-7 and HuH-7 R cells. Cell-proliferation, wound-healing, and invasion assays were performed after transduction of cells with shRNA-expressing viruses.
For iTRAQ, total protein was extracted from xenograft tumors formed from HuH-7 or HuH-7 R tumors (n ϭ 6 each) and enriched using a 3-kDa centrifugal filter (Millipore, Watford, UK). This process was repeated twice using double-distilled H 2 O to desalt and remove the protease inhibitor mixture. A total of 400 g of protein was collected from paired HuH-7 and HuH-7 R tumors for iTRAQ analysis. The protein mixtures were incubated in 0.5 M triethylammonium bicarbonate (TEAB; pH 8.5) and 2% SDS, reduced with 5 mM Tris (2-carboxyethyl) phosphine (TCEP) for 1 h at 60°C, and alkylated with 10 mM s-methyl methanethiosulfonate (MMTS) at room temperature for 10 min. Each 100 g of protein was digested overnight in tryptic solution (1:100) at 37°C. Digested peptides from HuH-7 and HuH-7 R tumors were labeled with 114,115 and 116,117 iTRAQ reagents (SCIEX, Foster City), respectively.
Each fraction was trapped on a reverse phase C18 column (Acclaim PepMap100, 3 m, 100 Å, 75 m ϫ 2 cm; Dionex, Sunnyvale) and separated using coupled reverse phase C18 chromatography (Acclaim PepMap RSLC, 2 m, 100 Å, 75 m ϫ 15 cm; Thermo Fisher Scientific, Waltham) with an acetonitrile gradient in 0.1% formic acid. The injection volume was 2 l, and the flow rate was 250 nL/min. The mobile phases consisted of buffer A (0.1% formic acid) and buffer B (0.1% formic acid in 90% acetonitrile). The gradient condition was 4 -30% buffer B for 90 min, 30 -90% buffer B for 15 min, hold in 90% buffer B for 10 min, and then equilibrate with buffer A for 15 min. Full-scan MS spectra (m/z 300 -1600) were acquired in an Orbitrap mass analyzer at a resolution of 60,000. The lock mass calibration feature was enabled to improve mass accuracy, with lock mass set at 445.12003 (polycyclodimethylsiloxane).
For SILAC analysis, the most intense ions (up to 20) with a minimal signal intensity of 1000 were sequentially isolated for MS/MS fragmentation in order of the intensity of precursor peaks in the linear ion trap using a collision-induced dissociation energy of 30%, Q activation at 0.25, an activation time of 10 ms, and an isolation width of 2.0. Targeted ions with m/z Ϯ 10 ppm were selected for MS/MS and dynamically excluded for 60 s.
For iTRAQ analysis, MS data were acquired using the following parameters: 10 data-dependent CID-HCD dual MS/MS scans per full scan; CID scans acquired in LTQ with two-microscan averaging; full scans and HCD scans acquired in Orbitrap at a resolution of 60,000 and 15,000, respectively; normalized collision energy (NCE) of 30% in CID and 50% in HCD; Ϯ 2.0 m/z isolation window; and dynamic exclusion for 60 s. In CID-HCD dual scan, each selected parent ion was first fragmented by CID and then by HCD.
Protein Identification and Quantification-The precursor mass tolerance was set at 7 ppm, and fragment ion mass tolerance set at 0.5 Da. The dynamic modifications were deamidated (NQ), oxidation (M), and N-terminal acetylation. The static modification was cysteine carbamidomethylation, and a maximum of two miscleavages were allowed. False discovery rate was calculated by enabling the peptide sequence analysis using a decoy database. Identified peptides were validated using a Percolator algorithm with a q-value threshold of 0.01. Mass spectrometry data were processed and quantified using Proteome Discoverer (Version 1.3) software (Thermo Fisher Scientific) workflow from the Mascot search engine (version 2.3.02), and searched against the Swiss-Prot 57.2 version with Homo sapiens (human) protein database containing 20,232 sequences.
For SILAC-based proteomics, the search parameters were set using isotope labeling of lysine (ϩ6.020 Da) and isotope labeling of arginine (ϩ10.008 Da) as the dynamic modifications. For each SILAC pair, Proteome Discoverer determines the area of the extracted ion chromatogram and computes the "heavy/light" ratio. Protein ratios are then calculated as the median of all the quantified unique peptides belonging to a certain protein. The ratios among proteins in the heavy and light versions were used as fold-change.
For iTRAQ-based proteomics, the search parameters were set using methyl methanethiosulfonate as cysteine, iTRAQ 4-plex at ly-sine, and the N-terminal residue as static modifications. Fragment ion mass tolerance and precursor ion tolerance were set to 0.2 Da with a 95% confidence threshold.
Bioinformatics Analysis-Data sets representing proteins with altered expression profile derived from quantitative proteomics (SILAC and iTRAQ) analyses were categorized into functional groups based on the Ingenuity Pathway Analysis Tool (Ingenuity Systems, Redwood City; http://www.ingenuity.com). In IPA, differentially expressed proteins are analyzed in terms of biological responses and canonical pathways. Ranking and significance of the bio-functions and the canonical pathways were tested by the p value. The bio-functions and canonical pathways were ordered by the ratio (numbers of genes from the input data set that map to the pathway divided by the total number of molecules that exist in the canonical pathway). Additionally, differentially expressed proteins are mapped to gene networks available in the Ingenuity database and then ranked by score. The networks created are ranked depending on the number of significantly expressed genes they contain; the most significant associated diseases are also listed. A network is a graphical representation of the molecular relationships among these molecules. Genes or gene products are represented as nodes, and the biological relationship between two nodes is represented as an edge (line). All edges are supported by at least one literature reference and canonical information stored in the Ingenuity Pathways Knowledge Base. The intensity of the node color indicates the expression level of up-regulation (red) or down-regulation (green).
Reverse Transcription-polymerase Chain Reaction (RT-PCR) and Chromatin Immunoprecipitation (ChIP) Assays-The expression of galectin-1 mRNA was quantified by RT-PCR using ␤-actin as an internal standard for normalization. For ChIP assays, cells were grown under normoxia or treated with CoCl 2 and then cross-linked and quenched. Subsequently, cells were lysed and sonicated, yielding 200 -1000 bp DNA fragments. ChIP assays were performed using the SimpleChIP Enzymatic Chromatin IP Kit (Cell Signaling). The specific primers used for RT-PCR and ChIP are shown in supplemental Table  S1.
Quantification of Galectin-1 in Patient Serum-A total of 91 HCC patients who received sorafenib-based treatment as the first-line therapy for advanced HCC from 2007 to 2012 and who consented to having their peripheral blood collected for analysis before the treatment started were enrolled in this study. The study was approved by the Institute Research Ethical Committee of National Taiwan University Hospital.
Serum levels of galectin-1 were determined with a galectin-1 sandwich enzyme-linked immunosorbent assay (ELISA). In brief, 96-well microplates (PerkinElmer, Shelton) were precoated with galectin-1 capture antibody (AF1152; R&D Systems, Minneapolis) at 4°C overnight. After washed, the plate was treated with blocking buffer (Block-PRO Blocking buffer; Visual Protein, Taipei, Taiwan) at 37°C for 1 h. Plates were then washed, and serum samples (100 l) were added and further incubated at 37°C for 2 h. Thereafter, biotinylated galectin-1 detection antibody (BAF1152; R&D Systems) was added and incubated at 37°C for 2 h. The wells were then rinsed and 100 l of a solution containing streptavidin-horseradish peroxidase (1:200) was added. After 1 h incubation, plates were washed and an NeA-Blue (tetramethylbenzidine substrate; Clinical Science Product Inc., Massachusetts) solution was added to the wells; the reaction was stopped by adding 1 mol/L H 2 SO 4 . The absorbance of each sample was determined at 450 nm. A standard curve prepared from 5 to 120 ng of galectin-1 was generated for each ELISA.
Statistical Analysis-Statistical analyses were conducted using SAS software. An independent t test was utilized to compare serum galectin-1 levels between healthy volunteers and patients. The associations between high or low galectin-1 levels and disease control or other baseline characteristics as nominal variables were analyzed using the Chi-square test or Fisher's exact test. Progression-free survival and overall survival were estimated using the Kaplan-Meier method and compared using with a log-rank test. In multivariate analyses, the Cox proportional hazards regression model was used to adjust for other potential clinicopathologic parameters described elsewhere (13). All tests were two-sided, and a p value Յ 0.05 was considered statistically significant.

Functional Analyses of HuH-7 and HuH-7 R Cells-Resistant
HuH-7 R cell lines were established previously (10). As shown in supplemental Fig. S1, the IC 50 value for sorafenib against these cells (8.75 M) is shifted to a higher value compared with that against HuH-7 cells (4.13 M). HuH-7 cells grew in monolayer clusters, whereas HuH-7 R cells adopted a spindle shape and lost cell-cell contact, suggesting that resistant cells display a more mesenchymal phenotype (Fig. 1A). To further confirm these observations, we performed woundhealing and invasion assays, which revealed that migration rate (Fig. 1B) and invasiveness (Fig. 1C) were dramatically increased in HuH-7 R cells compared with HuH-7 cells. These data suggest that HuH-7 R cells possess a more aggressive phenotype than HuH-7 cells. sorafenib resistant HuH-7 R cells compared with parental HuH-7 cells, we utilized two different quantitative proteomic analyses: SILAC (for in vitro labeling) and iTRAQ (for in vivo labeling). A schematic diagram of the experimental design for exploring sorafenib-acquired resistance in HuH-7 cells is shown in Fig. 1D. SILAC-based proteomic analysis yielded a total of 4,616 quantified proteins in both forward and reverse experiments, which could avoid biases in cell labeling. Of these proteins, 699 were found to have statistically significant changes in expression in the HuH-7 R cells (supplemental Fig.  S2). To further determine the in vivo response to sorafenib resistance, a total of 2,836 proteins were successfully identified and quantified using iTRAQ-based proteomic analysis. Outliers were identified based on a p value Ͼ 0.05 and 114/ 116 and 115/117 ratio Ͼ2.0 or Ͻ 0.5. This resulted in 567 proteins being considered statistically reliable hits (supplemental Fig. S2). Among those data sets, a total of 2,450 proteins common to both SILAC and iTRAQ experiments were reliably (false discovery rate [FDR] Ͻ 1%) identified and quantified. Ultimately, quantitative data from both data sets were normalized against the 5% trimmed means to minimize the effect of extreme outliers and to center the protein log 2 ratio distribution on zero (14).

Identification and Quantification of Differentially Expressed Proteins in HuH-7 and HuH-7 R Cells and Cell-Derived
Biological Function, Pathway, and Network Analysis-An analysis of the abundance of proteins in SILAC and iTRAQ data sets showed that 156 proteins were differentially expressed between HuH-7 and HuH-7 R cells: expression of 81 proteins was increased in HuH-7 R cells (Ͼ2.0-fold), and expression of 75 proteins was decreased (Ͻ0.5-fold) ( Fig. 2A and Table I, II). For a few proteins with only one quantified peptide, MS and MS/MS spectra were manually inspected to avoid error erroneous quantification (supplemental Fig. S3). To identify altered biological functions that might play a role in sorafenib resistance, we further analyzed the 156 quantified proteins using the functional analysis of up-regulated proteins, which were mainly related to cellular movement (n ϭ 9), cellular growth and proliferation (n ϭ 19), cellular development (n ϭ 19) and cellular assembly and organization (n ϭ 11) ( Fig.  2B and Supplemental Table S2); whereas the down-regulated proteins were predominantly involved in amino acid metabolism (n ϭ 7), small molecule biochemistry (n ϭ 7) and nucleic acid metabolism (n ϭ 8) ( Fig. 2B and supplemental Table S2). IPA was further adopted for grouping proteins into networks and canonical pathways to determine the altered cellular activities during sorafenib resistance. The top one network associated with up-regulated proteins was found to be mainly involved in cellular movement, cell-to-cell signaling and interaction and tissue development. On the contrary, the top networks of down-regulated proteins involved in drug metabolism, endocrine system development and function (Table III). Additionally, the most significant biological network, which received an IPA score 47, included several differentially expressed proteins that correlated with the PI3K/AKT and mTOR signaling pathways (Fig. 2C). Among those proteins were simultaneously associated with different biological functions and disease, such as metastasis, formation of cellular protrusions, liver cancer, and proliferation of tumor cells (Fig.  2C and Table IV). In summary, we found 10 significantly differentially expressed proteins identified in proteomic dataannexin A1 (ANXA1), annexin A2 (ANXA2), coiled-coil domaincontaining 88A; gridin (CCDC88A), connective tissue growth factor (CTGF), EPH receptor A2 (EPHA2), ezrin (EZR), galectin-1 (LGALS1), IQ motif-containing GTPase-activating protein 1 (IQGAP1), Ral GTPase-activating protein, alpha subunit 2 (RALGAPA2), and vimentin (VIM), which mainly participated in cellular movement. These finding led us to focus on proteins that could play a relevant role in cell motility and metastasis.
Selected In Vitro-and In Vivo-Overexpressed Proteins Associated with Epithelial-Mesenchymal Transition (EMT)-A set of six out of the 10 candidate proteins associated with EMT including vimentin, CTGF, IQGAP1, galectin-1, ezrin, and annexin A2, were selected. MS spectra of representative peptides are shown in Fig. 3 and these proteins were further validated by Western blotting analysis. The SILAC-based quantitative MS spectrum was consistent with the iTRAQbased quantitative MS spectrum. Western blotting results were consistent with those of proteome analysis (supplemental Fig. S4). To further identify proteins dysregulated in HuH-7 R cells that might be used as HCC serum biomarkers for predicting sorafenib resistance, we analyzed quantified proteins using the SignalP program. A total of 22 proteins were putative secreted proteins; two of these candidatesgalectin-1 and CTGF-were highly expressed in HuH-7 R cells. Interestingly, galectin-1, which was significantly up-regulated in HuH-7 R cells and is known to play a crucial role in the regulation of cell migration, was identified in HuH-7 R cell conditioned medium, confirming that it was secreted (supplemental Fig. S5). In contrast, CTGF was not detected in conditioned medium (data not shown).
Galectin-1 Knockdown Inhibits HuH-7 R Cell Proliferation, Migration, and Invasion, and Restores Sorafenib Sensitivity-We next sought to investigate the role of galectin-1 in conferring sorafenib resistance and increasing migration. Because HuH-7 cells expressed negligible levels of galectin-1 compared with HuH-7 R cells, we employed lentiviral-mediated delivery of galectin-1 shRNAs to inhibit the expression of galectin-1 in HuH-7 R cells (Fig. 4A). Transduction of HuH-7 R cells with shGal-1 dramatically decreased galectin-1 expression (Fig. 4B). Subsequent MTT assays showed that knockdown galectin-1 significantly suppressed proliferation in HuH-7 R cells (Fig. 4C). Wound-healing and invasion assays performed in galectin-1-knockdown HuH-7 R cells revealed that suppression of galectin-1 expression significantly blocked migration ability (Fig. 4D) and invasion activity (Fig.  4E) compared with HuH-7 R cells. Importantly, we found that repression of galectin-1 restored sorafenib sensitivity in HuH-7 R cells (Fig. 4F), reducing the IC 50 of sorafenib to a

Galectin-1 as a Predictive Marker of Sorafenib Resistance
value close to that for HuH-7 cells. Taken together, these results show that knockdown of galectin-1 not only attenuates cell proliferation and metastasis in HuH-7 R cells, it also restores sorafenib sensitivity.

High Expression of Galectin-1 in HuH-7 R Cells Promotes Tumorigensis and Pulmonary Metastasis
In Vivo-To further assess the tumorigenic and metastatic potential of HuH-7 R cells, which express galectin-1 at elevated levels, we employed mouse xenograft tumor models created by subcutaneous or tail vein injection of HuH-7 or HuH-7 R cells. As shown in Fig. 5A, HuH-7 R cells exhibited enhanced tumorigenic ability compared with HuH-7 cells. Immunohistochem-  FKBP8, FSH, HEXIM1, Hsp90, HSPB1, IGF2R

Galectin-1 as a Predictive Marker of Sorafenib Resistance
istry revealed intense staining for galectin-1 and the proliferation marker Ki-67 in tumors formed by HuH-7 R cells, showing that proliferation rates were increased in these galectin-1-overexpressing tumors (Fig. 5B). Moreover, elevated galectin-1 expression in HuH-7 R cells might correlate with the enhanced development of pulmonary metastatic nodules ( Fig.  5C and 5D). Taken together, these results suggest that HuH-7 R cells have greater tumorigenic and metastatic potential than HuH-7 cells in vivo. Galectin-1 Expression is Regulated by PI3K/AKT, mTOR, and HIF-1␣ Pathways-Bioinformatics analyses indicated that up-regulation of the mTOR (mammalian target of rapamycin) signaling pathway could be involved in facilitating the sorafenib resistance of HuH-7 R cells (Table IV). To test this, we examined the involvement of the mTOR-signaling pathway in galectin-1 expression in HuH-7 R cells. Time-course experiments showed that treatment of HuH-7 R cells with rapamycin (an inhibitor of mTOR) almost completely blocked phosphorylation of eukaryotic translation initiation factor 4E binding protein 1 (4EBP1), ribosomal protein S6 kinase, 70 kDa (P70S6K) and ribosomal protein S6 (S6), and markedly atten-uated expression of galectin-1 at the protein level (Fig. 6A). Furthermore, we found that inhibition of AKT phosphorylation with the phosphoinositide 3-kinase (PI3K) inhibitor LY294002 significantly reduced galectin-1 expression in HuH-7 R cells (Fig. 6B). Moreover, we also detected the mRNA level of galectin-1 declined after LY294002 and rapamycin treatment, respectively (Fig. 6D, upper panel). These data suggest that both the AKT and mTOR pathways are involved in galectin-1 up-regulation.
A previous study showed that galectin-1 is a direct target of the transcription factor, hypoxia inducible factor 1 alpha (HIF-1␣) (11). To explore further the linkage between HIF-1␣ and galectin-1 in HuH-7 R cells, we exposed the cells to the wellknown hypoxia-mimetic agent, CoCl 2 . CoCl 2 significantly enhanced galectin-1 protein expression in a time-dependent manner (Fig. 6C), and also increased galectin-1 mRNA levels (Fig. 6D, upper panel). To further confirm that these effects are mediated by transcriptional activation of the galectin-1 gene, we examined binding of HIF-1␣ to the endogenous galectin-1 promoter in HuH-7 R cells, with or without CoCl 2 treatment, using ChIP assays. In the chromatin fraction pulled down by E, Transwell migration assays of shGal-1-knockdown HuH-7 R cells. Cells in the central field of each insert were visualized by light microscopy and quantified. F, shGal-1-knockdown cells were exposed to sorafenib at the indicated concentrations for 72 h, and cell viability was analyzed by MTT assay. The concentration-response curve for sorafenib in the shGal-1-knockdown group was shifted toward a lower concentration compare with that for shCtrl HuH-7 R cells. Data are presented as means Ϯ S.D., and are representative of at least three independent biological replicates. shGal-1, shRNA against galectin-1; shCtrl, control shRNA. an anti-HIF-1␣ antibody, galectin-1 promoter PCR fragments were more abundant in CoCl 2 -treated cells than in control cells (Fig. 6D, lower panel). Taken together, these results show that the expression of galectin-1 is mediated by the PI3K/AKT/mTOR/HIF-1␣ pathway (Fig. 6E).
Prognostic Value of Galectin-1 in Advanced HCC Patients-To determine whether galectin-1 expression is predictive of sorafenib resistance, we examined baseline galectin-1 levels before sorafenib treatment in 91 advanced HCC patients using ELISA. The basic characteristics of the 91 advanced HCC patients were showed in Supplemental Table S3. As shown in Fig. 7A, the mean Ϯ S.D. level of serum galectin-1 from 17 healthy volunteers was 89.9 Ϯ 30.2 ng/ml (range: 49.8 -148.5 ng/ml). Using the maximum value of serum galectin-1 for healthy volunteers as the cutoff point, we found that patients with high pretreatment galectin-1 levels (i.e. Ͼ148.5 ng/ml) had significantly lower disease control rates (48%) than patients with low pretreatment galectin-1 levels (72%, p ϭ 0.023; supplemental Table S4). Response rates in patients with high galectin-1 levels also trended lower compared with patients with low galectin-1 levels (2% versus 10%), although this difference did not reach statistical significance. Compared with patients with low galectin-1 levels, patients with high pretreatment galectin-1 levels also had significantly shorter median progression-free survival (2.2 versus 4.2 months, p ϭ 0.026; Fig. 7B) and overall survival (6.1 versus 10.7 months, p ϭ 0.050; Fig. 7C). After adjusting for other potential prognostic factors, multivariate analyses showed that high pretreatment galectin-1 levels remained an independent predictor of shorter progression-free survival (HR ϭ 1.888, p ϭ 0.008) and overall survival (HR ϭ 2.179, p ϭ 0.002) (supplemental Table S5). Notably, an examination of 29 HCC patients who developed progressive disease after sorafenib treatment showed a dramatic increase in serum galectin-1 concentration (Fig. 7D). Our data thus indicate that high galectin-1 serum level is associated with poor treatment efficacy of sorafenib, and shorter survivals in advanced HCC patients treated with sorafenib. DISCUSSION Sorafenib is a kinase-targeted drug for treatment of advanced HCC, but its use is hampered by the development of drug resistance. Therefore, understanding the molecular changes that underlie the biological consequences of acquired drug resistance is of critical importance. In this study, we performed dual SILAC and iTRAQ quantitative proteomics, allowing a broad, systematic examination of changes in the proteome that are associated with the acquisition of sorafenib resistance. The 156 differentially expressed proteins revealed a distinct signaling and EMT protein signature associated with sorafenib resistance in HuH-7 R cells. Among these proteins, 10 were linked to cellular movement, growth/proliferation, and cancer. Notably, our data showed that galectin-1 was linked to the AKT/mTOR/HIF-1␣ pathway, supporting galectin-1 as a predictive biomarker for sorafenib resistance.
As previous reports indicated, when 400 mg of sorafenib was given twice daily, the concentration of sorafenib in human plasma was between 5 and 7 mg/L, which is 7.8 -10.9 M in humans (15). In order to investigate the molecular mechanism of the acquired resistance to sorafenib, we developed HuH-7 R cells, which in the clinically relevant dose about 10 M (the highest clinical achievable concentration). We showed that long-term exposure to sorafenib of HuH-7 cells changed their morphology into spindle shaped cells. These features are typical seen in cells undergoing EMT (16). Moreover, EMT is observed in HuH-7 R cells for loss of E-cadherin and gain of vimentin by Western blotting (Supplemental Fig. S6). The sorafenib resistant cells showed an activation of the EMT process with enhanced invasive and metastatic potentials. We also performed wound-healing and invasion assays, which revealed that migration rate and invasiveness were significantly up-regulated in HuH-7 R cells compared with HuH-7 cells. Recent reports have indicated that the emergence of drug resistance may link EMT as a contributing mechanism, such as cisplatin resistance in ovarian cancer (17) and gefitinib resistance in lung cancer (18). Therefore, this indicated that the selected cells should mimic the tolerance of sorafenib and behavior as the HCC in drug resistance patients.
Among the 10 differentially expressed proteins were associated with cell motility or invasion (19 -28), nine were significantly increased in the highly metastatic HuH-7 R cells compared with the poorly metastatic HuH-7 cells, whereas one was notably decreased. Consistent with the possible metastasis-related functions of vimentin and ezrin, considerable evidence have shown that both proteins are responsible for FIG. 7. Galectin-1 is highly expressed in HCC serum samples and HCC patients treated with sorafenib. A, Serum levels of galectin-1 in healthy volunteers (n ϭ 17; mean ϭ 89.9 ng/ml) and patients with advanced HCC (n ϭ 91; mean ϭ 179.6 ng/ml). Patients with advanced HCC had significantly higher serum galectin-1 levels than healthy volunteers (p Ͻ 0.001). The horizontal lines indicate means Ϯ S.D. B and C, Kaplan-Meier analysis of progress-free survival B, and overall survival C, of patients with advanced HCC, grouped according to high and low pretreatment galectin-1 levels. p values are based on log-rank tests. D, Serum galectin-1 levels in patients before sorafenib treatment and upon disease progression during sorafenib treatment (n ϭ 29). Serum galectin-1 levels significantly increased with disease progression (p Ͻ 0.001). PFS, progress-free survival; OS, overall survival. maintaining cell shape, stabilizing cytoskeletal interactions and cell motility (20,25). Furthermore, annexin A1 is a key regulator of pathological angiogenesis and physiological angiogenic balance (29). Attenuated expression of RALGAPA2 leads to tumor invasion and metastasis of bladder cancer (21). Gridin regulates reorganization of the actin cytoskeleton and modulation of AKT activity, which ultimately result in cancer invasion and angiogenesis (30). Annexin A2, IQGAP1, and EPHA2 are closely associated with drug resistance. Annexin A2 involved in cell adhesion, cell motility, and expressed at higher levels in metastatic cancer and is associated with a drug-resistant phenotype. IQGAP1, which regulates cellular activities associated with cell-cell adhesion and cell migration, is overexpressed in trastuzumab-resistant breast epithelial cells; reducing IQGAP1 both increases the inhibitory effects of trastuzumab and restores trastuzumab sensitivity (31). EPHA2 belongs to the ephrin receptor subfamily of the protein-tyrosine kinase family. Cancer cells that overexpress EPHA2 exhibit increased motility and invasive properties, consistent with a prometastatic phenotype. Consistent with this, silencing EPHA2 inhibits proliferation and invasion, and increases sensitivity to paclitaxel (32). CTGF and galectin-1 are secreted proteins that are important in tumor growth, angiogenesis, and metastasis. CTGF modulates the invasion of certain human cancer cells through binding to integrins (19). Dysregulation of galectin-1 in cancer has also been correlated with the aggressiveness of tumors (33). Taken together, these observations suggest that metastasis is one of the most important causes of poor prognosis in patients with HCC. We hypothesize that the above proteins are involved in adverse responses to sorafenib, although additional study will be needed to verify their specific roles in sorafenib resistance.
The goal of our study was to investigate the potential use of proteins that are differentially released from HCC cells as predictive or prognostic biomarkers for HCC patients treated with sorafenib. Biomarker for predicting the efficacy of sorafenib is a growing field and a number of candidate markers have been proposed. Low HGF levels and high c-kit levels in plasma at baseline were reported to be associated with longer survival in HCC patients treated with sorafenib (9). Several serum angiogenesis-related cytokines levels were correlated with response to sorafenib treatment (9,34). Some tissue markers, such as ␣␤-crystallin (35), FGF3/FGF4 (36), JNK (37), and pERK (38) have been reported to predict sorafenib response. A recent study indicated that a mesenchymal profile and expression of CD44 may predict lack of response to sorafenib in HCC patients (39). Although various markers have been studied, identifying predictive biomarkers to sorafenib response remains challenging and warrants further investigation. Our data showed that galectin-1, which had not previously been characterized as having a role in mediating sorafenib resistance, was identified as a protein secreted by HuH-7 R cells. Our mechanistic studies identified galectin-1 as a downstream effector of the AKT/mTOR/HIF-1␣ pathway. This is consistent with previous study showing that activation of AKT signaling mediates acquired resistance to sorafenib in HCC cells (40) and the constitutive activation of the mTOR pathway in sorafenib-resistant HCC cells by array-based pathway profiling (41). Furthermore, we also showed that down-regulation of galectin-1 suppressed migratory and invasive abilities of HuH-7 R cells, and restored sorafenib sensitivity. Several studies supported that galectin-1 associated with metastatic ability and effects of galectin-1 knockdown on drug sensitivity in different types of cancer (42)(43)(44). Taken together, our findings indicate that galectin-1 may be a component of the mechanism that promotes the progression of HCC and resistance to sorafenib. In validation studies using clinical samples, we showed that galectin-1 serum levels were markedly elevated in advanced HCC patients compared with healthy controls; in some cases, galectin-1 serum levels further increased after sorafenib treatment. We also showed that a high serum galectin-1 level was an independent factor associated with poor progress-free survival and overall survival. Additionally, HCC tissue microarray analysis showed that patients with high galectin-1 expression had a higher rate of tumor recurrence and shorter overall survival than those with lower galectin-1 expression (45). Taken together, these data may suggest that the serum levels of galectin-1 can serve as a prognostic factor for HCC. On the other hand, our data support the potential use of galectin-1 serum level as a predictive biomarker of sorafenib treatment, because high galectin-1 serum levels are associated with a low response rate and poor disease control.
In conclusion, we showed that galectin-1 is increased in in vitro and in vivo sorafenib-resistant HCC models and may promote cancer metastasis and increase tumor invasion. We also showed that high serum galectin-1 levels are associated with poor treatment efficacy and shortened survival in advanced HCC patients treated with sorafenib. These findings support the potential use of galectin-1 as a novel predictive and prognostic biomarker of HCC.