Galectin-7 as a potential predictive marker of chemo-and/or radio-therapy resistance in oral squamous cell carcinoma

Treatment of advanced oral squamous cell carcinoma (OSCC) requires the integration of multimodal approaches. The aim of this study was to identify predictors of tumor sensitivity to preoperative radiotherapy/chemotherapy for OSCC in order to allow oncologists to determine optimum therapeutic strategies without the associated adverse effects. Here, the protein expression profiles of formalin-fixed paraffin-embedded (FFPE) tissue samples from 18 OSCC patients, termed learning cases, who received preoperative chemotherapy and/or radiotherapy followed by surgery were analyzed by quantitative proteomics and validated by immunohistochemistry in 68 test cases as well as in the 18 learning cases. We identified galectin-7 as a potential predictive marker of chemotherapy and/or radiotherapy resistance, and the sensitivity and specificity of the galectin-7 prediction score (G7PS) in predicting this resistance was of 96.0% and 39.5%, respectively, in the 68 test cases. The cumulative 5-year disease-specific survival rate was 75.2% in patients with resistant prediction using G7PS and 100% in patients with sensitive prediction. In vitro overexpression of galectin-7 significantly decreased cell viability in OSCC cell line. Therefore, our findings suggest that galectin-7 is a potential predictive marker of chemotherapy and/or radiotherapy resistance in patients with OSCC. Identification of proteins differentially expressed in OSSC samples from patients sensitive or resistant. The samples were processed by LC-MS and analyzed with 2DICAL.


Introduction
Oral cancer is the sixth most common cancer worldwide, with an annual incidence of~275,000 cases. However, unlike many other cancers, the incidence of oral cancer is increasing [1]. In Japan, the number of patients diagnosed with oral cancer was 2100 in 1975 and 6900 in 2005, and this number is estimated to increase to 7800 by 2015, when it will represent 1% of all cancer cases and approximately 40% of all head and neck cancer cases [2]. Histopathologically, squamous cell carcinoma (SCC) is the most common cancer of the oral cavity, accounting for >90% of all oral cancer cases. Despite recent improvements in multimodal therapies, the survival rate of these patients remains poor because of frequent locoregional and/or distant recurrences. These statistics highlight the urgent need for treatment alternatives [3].
Treatment of advanced head and neck SCC requires the integration of multimodal approaches. Interest in neoadjuvant chemotherapy has recently regenerated because of its survival benefits, particularly when a taxane-cisplatin-fluorouracil regimen is applied instead of the standard cisplatin-fluorouracil regimen [4][5][6]. Usually, tumor response to neoadjuvant chemotherapy predicts its response to radiotherapy. The prognostic indicators of favorable outcome would allow oncologists to make a more rational selection of therapeutic strategies without the unnecessary toxicities of neoadjuvant chemotherapy.
Over the past decade, gene expression in oral SCC (OSCC) has been studied extensively using microarray techniques. However, gene expression is not always correlated with the level of expression of the corresponding protein [7]. Furthermore, although the use of fresh material is required for most analytical approaches, human tissue samples are not always available in sufficient quantity. As an alternative, formalin-fixed paraffin-embedded (FFPE) tissue blocks can be routinely collected and stored as samples for research purposes after pathological diagnosis. A method of extracting proteins from FFPE tissues in the form of tryptic peptides was recently developed, and the methodology is compatible with a variety of mass spectrometry (MS)-based proteomics [8].
This study aimed to identify potential predictive markers of chemotherapy and/or radiotherapy resistance in patients with OSCC using quantitative proteomic analysis of FFPE biopsy tissues.

Patients and Methods
Patients This retrospective study included 86 patients diagnosed with resectable OSCC and treated at the Tokyo Medical and Dental University Hospital Faculty of Dentistry (TMDU, Tokyo, Japan) between January 2001 and December 2011 ( Table 1). The diagnosis was confirmed by histological examination of tissue biopsies surgically removed from the center of the cancerous tissue. The FFPE samples were fixed in formalin, embedded in paraffin, and stored at room temperature. Thereafter, all patients received preoperative chemotherapy and/or radiotherapy, followed by surgical primary tumor resection with or without neck dissection. After the chemotherapy and/or radiotherapy, treatment outcome was evaluated using the response evaluation criteria in the solid tumors (RECIST) guidelines. Cases of progressive disease (PD) or stable disease (SD) were assigned to the resistant group (Group R), whereas those who achieved a partial response (PR) or complete response (CR) were assigned to the sensitive group (Group S). This protocol was reviewed and approved by the Ethics Committee Board of the TMDU.
As learning cases, 18 biopsy samples were prepared (Table 1), including nine samples from Group R patients and nine samples from Group S patients. As test cases, 68 samples were prepared (Table 1), including 25 samples from Group R patients and 43 samples from Group S patients.

Preoperative chemotherapy and/or radiotherapy
Patients assigned to the S/R group received a daily fractional dose of radiotherapy (2 Gy; 5 days/week) for a total dose of 34-50 Gy using a 4MV LINAC (Varian, CA). Radiation was delivered to the primary tumor site and the cervical nodes for patients with nodal involvement. Concomitant chemotherapy with S-1 (Taiho Pharmaceutical Co., Tokyo, Japan) was orally administered twice a day after a meal for five consecutive days per week. The individual doses were calculated on the basis of body surface area (total 1000-3000 mg).
Patients assigned to the P/R group received a fractional daily dose of radiotherapy for a total dose of 50 Gy. Concomitant chemotherapy with carboplatin (CBDCA) or cisplatin (CDDP) was administered once a day using the selective intra-arterial infusion method via the superficial temporal artery [9]. The daily dose of CBDCA ranged from 10 to 30 mg, with a total dose of 495-725 mg. The daily dose of CDDP was 8 mg, with a total dose of 280 mg.
Patients in the pretreatment PF/R group received a daily fractional dose of radiotherapy for a total dose of 50 Gy. Concomitant chemotherapy with intravenous CDDP (80 mg/m 2 ), followed by 5-fluorouracil (5-Fu) (800 mg/m 2 per day) as a continuous 24-h infusion for five consecutive days, was administered.
Patients in the PF group received intravenous CDDP (60, 70, or 80 mg/m 2 ), followed by 5-Fu (600, 700, or 800 mg/m 2 per day) as a continuous 24-h infusion for five consecutive days. These patients received one or two cycles of treatment.
Patients in the pretreatment S group were administered oral S-1 twice a day after meals during the waiting period before surgery (total, 600-2100 mg).
Finally, patients in the pretreatment R group only received a daily fractional dose of radiotherapy (2.5 Gy for 4 days/week), with a total dose of 40 Gy.

Peptide extraction
The proteins were extracted from the FFPE samples in the form of tryptic peptides using the Liquid Tissue MS Protein Partitioning Kit (Expression Pathology, Rockville, MD) according to the manufacturer's protocol. The extracted peptides were desalted through a C18 ZipTip (Millipore, Billerica, MA).

Liquid chromatography-mass spectrometry
Eighteen samples (nine from Group R and nine from Group S) were blinded, randomized, and measured in duplicate with a linear gradient of 0-80% acetonitrile in 0.1% formic acid at a speed of 200 nL/min for 60 min using a nano-flow high-performance liquid chromatography (HPLC; NanoFrontier nLC; Hitachi High-technologies, Tokyo, Japan) that was connected to a triple time-of-flight mass spectrometer (5600 Triple TOF; AB Sciex, Framingham, MA). The system detected peptide peaks every 1 sec, with a mass-to-charge ratio (m/z) ranging from 400 to 1600. The MS peaks were detected, normalized, and quantified using in-house 2DICAL (twodimensional image-converted analysis of liquid chromatography and mass spectrometry) software as described previously [10]. A serial identification (ID) number was assigned to each MS peak from ID1 to ID70510.

Protein identification by MS/MS
The MS/MS spectra were aligned with a tolerance of AE0.5 m/z and a retention time (RT) of AE0.4 min. Then, Immunohistochemistry for galectin-7 FFPE sections from all 86 cases were analyzed by quantitative immunohistochemistry (IHC). The 4-lm thick sections were incubated with a rabbit monoclonal antigalectin-7 antibody (EPR4287; 1:2000; LifeSpan Biosciences, Inc., WA) for 60 min at room temperature. Then, they were incubated with biotinylated anti-primary antibodies (Histofine SAB PO kit; Nichirei, Tokyo, Japan). The signal was detected using streptavidin-peroxidase (Histofine SAB PO kit) and the diaminobenzidine tetrahydrochloride (DAB) substrate. Finally, the sections were counterstained with Mayer's hematoxylin and dehydrated.

Analysis of galectin-7 immunostaining
The galectin-7-stained area (G7S) was quantified by dividing the staining intensity of that area over the staining intensity of the whole tissue section using ImageJ signals were visualized by SuperSignal â west dura extended duration (Thermo Scientific, Waltham, MA), using Lightcapture â (ATTO, Bioscience and Biotechnology, Tokyo, Japan).

In vitro cell viability assay
Cell viability was assessed by the colorimetric water-soluble tetrazolium salt (WST) assay (Cell counting kit-8; Dojindo Laboratories, Kumamoto, Japan) as described by the manufacturer. Briefly, cells (1 9 10 4 /well) were seeded into 96-well plates and cultured for 12 h. Then, cells were incubated with adenovirus infection for 12 h with or without 1 lg/mL doxycycline, and further cultured with or without doxycycline in fresh medium with the absence or presence of anticancer drug, 5 lg/mL cisplatin or 50 lg/mL 5-fluorouracil, for an additional 24 h. At the end of the culture period, 10 lL of a formazan-generating reagent, WST-8, was added to each well for 60 min at 37°C. The absorbance of each well at 450 nm was then measured using a microplate reader.

Statistical Analysis
The MS data and cell viability were analyzed by receiver operating characteristic (ROC) curve analysis, including area under the curve (AUC), stepwise discriminant analysis, T-tests, Mann-Whitney U-tests, and two-sided log-rank tests with PASW Statistics, version 18 (SPSS Inc., Chicago, IL). P < 0.05 was considered statistically significant.

Identification of a predictive marker of chemotherapy and/or radiotherapy resistance in OSCC by LC-MS
To identify a predictive marker of chemotherapy and/or radiotherapy resistance in OSCC, 18 biopsy samples, including nine samples from Group R and nine samples from Group S, were analyzed by LC-MS (Fig. 1). A total of 70,510 peaks per sample were readily detected and quantified. Among them, 10,869 MS peaks differed in intensity between Group R and Group S (P < 0.05; AUC >0.7). We further selected 105 peaks with Mascot scores >50 and excluded abundant proteins, including keratin, fibrinogen, collagen, and histone. Among them, we selected 20 peaks with Group R/Group S peak intensity ratios ≤0.5. We further limited our selection to the peaks with high predictive power on the basis of discriminant analysis (stepwise method), which identified two candidate peaks (ID1181 and ID2504). We chose the peak with the highest Mascot score: ID1181 (Table 2). MS/MS spectra of the ID1181 peaks matched the amino acid sequence of the galectin-7 protein.

Expression of galectin-7 in OSCC by IHC
The protein expression of galectin-7 in OSCC tissue was confirmed by IHC analysis of the FFPE samples used for LC-MS analysis ( Fig. 2A). Pearson's correlation analysis was conducted between G7S and each of the peaks listed in Table 2 analysis was conducted for the G7S values of Group R and Group S (Fig. 4A). The median G7S was significantly lower for Group R (0.455) than for Group S (0.666; Mann-Whitney U-test; P = 0.003), suggesting that oral SSC patients with low galectin-7 expression are more likely to exhibit chemotherapy and/or radiotherapy resistance. The sensitivity of this prediction was 88.9% and specificity was 88.9% (cutoff point: 0.526). A careful observation of the IHC findings revealed that galectin-7 was expressed in both the cytosolic and nuclear compartments; we observed strong nuclear staining in Group R and mostly cytosolic staining in Group S (Fig. 2). Median G7N was 10-fold higher for Group R than for Group S, with 0.549 and 0.042, respectively (Mann-Whitney U-test; P = 0.003; Fig. 4B). These data show that chemotherapy and/or radiotherapy resistance is associated with a nuclear concentration of galectin-7. Therefore, we conducted a discriminant analysis using G7S and G7NL for the 18 learning samples analyzed by LC-MS and IHC and obtained the following predictive formula for chemotherapy and/or radiotherapy resistance: Galectin-7 prediction score (G7PS) ¼ 5:812G7S À 1:246G7NC À 1:96 ðcutoff point : 0Þ Based on this formula, the sensitivity of prediction was 100% and specificity was 88.9%, indicating that sensitivity was increased in the 18 learning cases (Fig. 5A). When this formula was used to analyze the remaining 68 test cases, the sensitivity of prediction was 96.0% and specificity was 39.5% (Fig. 5B).

Galectin-7 prediction score correlates with poor prognosis in patients with OSCC
Five-year cumulative survival rates in Group S and Group R were estimated by Kaplan-Meier analysis using galectin-7 as a predictor of chemotherapy and/or radiotherapy resistance. The cumulative 5-year disease-specific survival rate was 75.2% in patients with resistant prediction using galectin-7 prediction score (G7PS) (<0) and 100% in patients with sensitive prediction (G7PS ≥0; Fig. 6). There was a significant positive correlation between resistant prediction using G7PS and survival parameter (log-rank test; P = 0.027; Fig. 6).

Galectin-7 decreases cell viability
To investigate the roles of galectin-7 in OSCC cells, the expression status of galectin-7 in six human OSCC cell lines was detected by Western blot analysis. A low endogenous expression of galectin-7 was detected in all OSCC cell lines except SKN3 (Fig. 7A). Next, we examined the effect of overexpressed galectin-7 in OSCC cells. HSC3 cells were infected with recombinant adenovirus encoding FLAG-tagged galectin-7 (Ad-FLAG-GAL7). The expression of galectin-7 was detected in a MOI-dependent manner with 1 lg/mL doxycycline by Western blot analysis (Fig. 7B), and we confirmed that FLAG-tagged galectin-7 was strongly expressed in HSC3 cells than endogenous expressions of galectin-7 in SKN3 or HSC2 cells (Fig. 7C). To examine the infection efficiency and intracellular distribution of Ad-FLAG-GAL7, we performed immunofluorescence labeling for overexpressed galectin-7. The infection efficiency of Ad-FLAG-GAL7 in HSC3 cells at MOI 50 was~80% (Fig. S1A). The intracellular distribution of Ad-FLAG-GAL7 was similar to the IHC staining pattern of galectin-7 (Fig. 8B). Moreover, by Western blot analysis, we confirmed that analysis of supernatants from HSC3 cells infected with Ad-FLAG-GAL7 or other OSCC cell lines have failed to provide evidence for a secreted form of galectin-7 (data not shown). To examine the effect of galectin-7 on cell viability, HSC3 cells infected with Ad-FLAG-GAL7 (MOI of 50) were cultured for 24 h with or without doxycycline. The overexpression of galectin-7 significantly decreased cell viability in normal culture conditions (Fig. 8A). Furthermore, similar results were observed when we treated the cells with 5 lg/mL cisplatin or 50 lg/mL 5-fluorouracil (Fig. 8A). These results indicate that galectin-7 may be involved in tumor cell proliferation/viability rather than chemosensitivity. We also investigated the role of galectin-7 using antisense galectin-7 oligonucleotides in OSCC cell lines. Unfortunately, the results showed no effects of galectin-7 knockdown on cell viability (Fig. S2). To determine whether the decreased cell viability was because of apoptosis, Ad-FLAG-GAL7infected HSC3 cells were cultured with or without doxycycline. We observed weak activation and cleavage of caspase-3 induced by the overexpression of galectin-7 was observed (Fig. 8B) indicating that the decreased cell viability by overexpression of galectin-7 may be because of growth arrest rather than apoptosis.

Discussion
The identification of tumor response predictors applying to a variety of radiotherapy and chemotherapy regimens will allow oncologists to customize strategies aimed at minimizing exposure to high-dose adjuvant therapy. To identify candidate biomarkers from the proteomics data, we used the 2DICAL analysis platform that performs a quantitative comparison of unlabeled shotgun proteomics data generated by LC-MS [11]. This approach was successfully used to identify blood biomarkers in pancreatic and colorectal cancer patients [12][13][14][15][16][17]. The use of proteins extracted from FFPE tissues in the form of tryptic peptides is compatible with a variety of MS-based proteomics [8]. However, we present the first report, as per our knowledge, on tumor response predictors in patients with OSCC.
Galectins constitute a family of evolutionary-conserved carbohydrate-binding proteins. They are widely distrib- uted in all organisms and are implicated in many essential functions such as development, differentiation, cell-cell adhesion, cell-matrix interaction, growth regulation, and apoptosis [18]. The various roles of galectins in cancer invasion and metastasis have been well documented [19,20]. However, very little is known about how galectin-7 expression affects cancer progression [18]. The expression level of galectin-7 varies widely among tumor types, from completely downregulated to highly upregulated [21][22][23][24].
De novo galectin-7 expression by p53 is associated with apoptosis, which inhibits tumor growth [25]. Alternatively, galectin-7 mediates the nuclear export of Smad3 through complex formation, which is essential for the tumor-suppressive effects of hepatocyte growth factor on the transforming growth factor-beta signaling pathway [26]. Paradoxically, galectin-7 may also induce the expression of genes known to promote cancer progression, including matrix metalloproteinase-9 (MMP-9) [27,28]. In this case, galectin-7 expression could be induced by nuclear factor-kappa B, a transcription factor and a positive regulator of MMP-9 known to be expressed in highly aggressive tumor cells [18]. This study provides in vitro evidence suggesting not a promoting effect of galectin-7 on the chemosensitivity but a potential of galectin-7 on decreasing cell proliferation/viability, which may be an interesting topic for future studies on OSCC therapy. We also investigated the role of galectin-7 using antisense galectin-7 oligonucleotides, however, the results showed no effects of galectin-7 knockdown on cell viability. A plausible reason for this is that strongly overexpressed galectin-7 affected cell viability, whereas physiologically expressed galectin-7 did not. Therefore, the knockdown of physiologically expressed galectin-7 did not elicit a noticeable effect. Moreover, because the function of galectin-7 in the nucleus remains unclear, further studies are needed to clarify this issue. This study suggests that low galectin-7 protein expression translates into low cumulative survival rate in patients with OSCC via tumor resistance to preoperative chemotherapy and/or radiotherapy. These data are not consistent with the findings of Saussez et al. [29], who reported that galectin-7 protein expression increases during tumor progression in hypopharyngeal and laryngeal SCC. However, this research team also reported that the percentage of cells immunopositive for galectin-7 decreased significantly with the loss of histological differentiation in hypopharyngeal SCC [30]. Furthermore, tumor progression was associated with a translocation of galectin-7 from the nucleus to the cytoplasm [29]. These data suggest that the sequestration of galectin-7 into the nucleus would prevent this protein from interfering with tumor progression or regression, leading to treatment resistance. Accordingly, a formula combining the expression level and nucleic/cytoplasmic expression ratio of galectin-7 would constitute a rational predictor of chemotherapy and/or radiotherapy resistance in patient with advanced OSCC.
Our findings suggest that the specificity of the prediction of chemotherapy and/or radiotherapy resistance using G7PS was 39.5%; therefore, there could be many false-positive results. However, one case with a response evaluation of PR who was predicted as resistant (G7PS = À0.25) died of uncontrolled neck node metastases despite surgery followed by additional radiotherapy. Therefore, patients with resistance prediction using G7PS (<0) should be considered for other treatments or drugs that were not used in this study (e.g., docetaxel or a molecular target drug). Moreover, a synergistic effect of galectin-7 and S100-A9 on cervical squamous carcinoma patients was recently reported [31]. We also identified S100-A9 as one of other potential predictive markers; therefore, identification of other useful combination markers that were not analyzed in this study may be important.
In conclusion, we identified galectin-7 as a predictor of tumor resistance and developed a predictive formula for patient survival. Accordingly, measurement of galectin-7 protein expression in fresh tissue biopsy samples at the time of diagnosis can provide invaluable information for the design of customized preoperative treatment regimens with advanced OSCC. Future challenges include the identification of other useful combination markers that were not analyzed in this study in order to improve diagnostic accuracy.