Identification of LEFTY as a molecular marker for ovarian clear cell carcinoma

To identify proteins involved in ovarian clear cell carcinoma (OCCCa), shotgun proteomics analysis was applied using formalin-fixed and paraffin-embedded samples of ovarian carcinoma. Analysis of 1521 proteins revealed that 52 were differentially expressed between four OCCCa and 12 non-OCCCa samples. Of the highly expressed proteins in OCCCa, we focused on left-right determination factor (LEFTY), a novel member of the transforming growth factor-β superfamily. In 143 cases of ovarian epithelial carcinoma including 99 OCCCas and 44 non-OCCCas, LEFTY expression at both mRNA and protein levels was significantly higher in OCCCas compared with non-OCCCas, with the mRNA expression of LEFTY1 being predominant compared to that of LEFTY2. OCCCa cells stably overexpressing LEFTY1 showed reduced cell proliferation, along with decreased pSmad2 expression, and also either displayed an activated p53/p21waf1 pathway or increased p27kip1 expression, directly or indirectly. Moreover, the treatment of stable cell lines with cisplatin led to increased apoptotic cells, together with the inhibition of protein expression of a pSmad2-mediated X-linked inhibitor of apoptosis and a decreased bcl2/bax ratio. Blocking LEFTY1 expression with a specific short hairpin RNA inhibited cisplatin-induced apoptosis, probably through the increased expression of both XIAP and bcl2, but not bax. In clinical samples, a significantly higher number of apoptotic cells and lower Ki-67 labeling indices were observed in OCCCas with a high LEFTY score relative to those with a low score. These findings suggest that LEFTY may be an excellent OCCCa-specific molecular marker, which has anti-tumor effects in altering cell proliferation and cellular susceptibility to apoptosis.

The left-right determination factor (LEFTY) is a novel member of the TGF-β superfamily, consisting of lefty1 and lefty2 in mice, which are homologous to LEFTY1 and LEFTY2 in humans, respectively. LEFTY1 is identical to LEFTYB, whereas LEFTY2 is identical to LEFTYA [13][14][15][16][17][18]. LEFTY serves as a repressor of TGF-β signaling by inhibiting Smad2 phosphorylation after activation of the TGF-β receptor, and further suppresses downstream events after R-Smad phosphorylation, including the heterodimerization of R-Smads with Smad4, and the nuclear translocation of the R-Smad-Smad4 complex [19].
This study was conducted to clarify the distinguishing factors between OCCCa and other histological subtypes of OECa based on their protein signatures derived from shotgun proteomics. We found significant up-regulation of LEFTY expression at both the mRNA and protein levels in OCCCa tissues when compared with other histological subtypes of OECa. We further elucidated the function and regulation of LEFTY, and examined TGF-β1/Smad signaling and changes in cell kinetics in OCCCa cells.

Shotgun proteomics analysis of clinical OECa samples
To extract protein for shotgun proteomics analysis, formalin-fixed and paraffin-embedded (FFPE) samples were exposed to a high concentration of Tris-buffer with boiling (Supplementary Figure 1A), which enabled the detection of many proteins with a relatively high molecular weight (Supplementary Figure 1B). This assay was used to identify differences in molecular expression between OCCCa and non-OCCCa. As shown in Figure  1A, a total of 5382 proteins, including 1267 in OCCCa, 1248 in OEmCa, 1346 in OMuCa and 1521 in OSeCa, were detected from 16 OECa tissues consisting of four samples of each subtype. Of these, 52 proteins were observed in all four OCCCa samples, but in less than three non-OCCCa samples (Table 1). Based on the values of spectral counts detected in the assay, we focused on LEFTY1 and LEFTY2, a novel member of the TGF-β superfamily, because LEFTY is a well-characterized molecule that plays an important role in the control of cell differentiation and proliferation during embryonic development [20,21]. However, the significance of its expression in tumor tissues remains to be clarified.

Up-regulation of LEFTY in OCCCa
Representative images of immunohistochemistry (IHC) findings for LEFTY in the four histological types of OECa are illustrated in Figure 1B (upper panels). The anti-LEFTY antibody used in this study was able to react with both LEFTY isoforms (Supplementary Figure 1C). This is because LEFTY1 and LEFTY2 are both 366 amino acids long, and differ at only 16 residues, thus they are 97% identical and share 350 identical residues. [18] Cytoplasmic LEFTY immunoreactivity, with or without nuclear staining, was frequently observed in OCCCa, in contrast to its focal positivity or absence in non-OCCCas. Average IHC scores for LEFTY showed a significant incremental decrease from OCCCa, OEmCa, OMuCa to OSeCa ( Figure 1B, lower graph). Similar findings were also evident by western blot assay, which demonstrated bands at 42, 34, and 28 kDa corresponding to the precursor peptide and polypeptides cleaved at Arg 77 and Arg 135 , respectively ( Figure 1C and Supplementary Figure 1C) [18].
In 85 OCCCa cases that were investigated, LEFTY scores were not associated with several clinicopathological factors looked at, including overall survival, progressionfree survival, clinical stage, and lymph node metastasis (Supplementary Figure 2A and 2B). Of these, six cases of postoperatively recurred tumors were available for further examination (Supplementary Table 1). The LEFTY score appeared to be decreased in the recurred OCCCas as compared to primary tumors, but the difference did not reach significance (p=0.37), probably due to the small number of cases investigated. In contrast, changes in apoptosis and cell proliferation status varied between the two lesions (Supplementary Figure 2C and 2D, and Supplementary Table 1).
Next, the association between mRNA expression of LEFTY and OECa was examined using specific primers for LEFTY1 and LEFTY2 (Supplementary Figure 1D). The expression level of LEFTY1 mRNA was significantly higher than that of LEFTY2 in OECas (Figure 2A). The expression of both LEFTY1 and LEFTY2 mRNAs was significantly higher in OCCCa compared with non-OCCCa ( Figure 2B). In the 10 OCCCa cases that were investigated, positive mRNA signals for LEFTY1, as detected by in situ hybridization, were also significantly higher than for LEFTY2 (Figure 2C and Supplementary  Table 2), in line with the shotgun proteomics data showing that 11 and 2 specific peptide fragments for LEFTY1 and LEFTY2, respectively, as well as 16 identical fragments for both isoforms, were detected in 4 OCCCa tissues (Supplementary Table 3). These were associated with LEFTY immunointensity, when cases were divided into high (score≧8) and low (score< 8) categories according to a mean value (8.6; Figure 2D).
To examine a link between LEFTY expression and the TGF-β1/Smad pathway, OCCCa cells including OVISE and TOV-21G cell lines were treated with TGF-β1. Ishikawa cells, an endometrial cancer cell line, were also used as a positive control, since treatment of the cells with TGF-β1 had previously resulted in the apparent increased expression of LEFTY and pSmad2, along with the induction of the p21 waf1 and p27 kip1 expression (Supplementary Figure 3). As shown in Figure 3A, increased expression of both pSmad2 and LEFTY in response to TGF-β1 was observed in OVISE and Ishikawa cells, whereas such effects were only minor in TOV-21G cells. Smad2 transfection or TGF-β1 treatment also led to markedly increased promoter activity of both LEFTY1 and LEFTY2 in Ishikawa, but not TOV-21G cells ( Figure 3B). In clinical samples, although LEFTY immunopositivity appeared to colocalize with pSmad2 immunoreactivity within tumor tissues of several OCCCa cases ( Figure 3C), the LEFTY score did not correlate with the pSmad2 score ( Figure 3D).
Because active demethylation is essential for regulating a subset of TGF-β1-dependent genes, [22] we further investigated changes in CpG island status within LEFTY1 in response to TGF-β1 by sodium bisulfate sequencing (Supplementary Figure 4A). Although LEFTY1 contained a region of CpG islands that were highly methylated in OVISE, TOV-21G and Ishikawa cells, demethylation by TGF-β1 treatment was rare (Supplementary Figure 4B).

Relationship of LEFTY expression with cell proliferation in OCCCa
To examine whether LEFTY expression affects cell proliferation in OCCCa cells, two independent cell lines stably overexpressing LEFTY1 were established using TOV-21G cells with low endogenous LEFTY expression as well as wild-type (wt) p53, and ES-2 cells with a loss of LEFTY expression and mutant-type (mt) p53 (TOV-L1 and ES-L1, respectively; Figure 4A). Both TOV-L1 and ES-L1 stable cells showed a tendency towards a low proliferation rate, particularly in the exponential growth phase ( Figure 4B and Supplementary Figure 5A). To further examine alterations in expression of several cell cycle-related molecules during cell growth, stable cells were rendered quiescent by serum starvation and were subsequently stimulated with serum. At 6 and 24 h after release in the cell cycle, expression levels of p53 and p21 waf1 in TOV-L1 stable cells and p27 kip1 in ES-L1 stable cells were substantially increased relative to the mock cells, in contrast to the decreased expression of pSmad2 in both stable cells ( Figure 4C). www.impactjournals.com/oncotarget In clinical samples, average Ki-67 labeling indices (LIs) were significantly lower in OCCCa with high LEFTY scores (≧8) relative to tumors with low scores (<8; Figure 4D).
Two independent cell lines, with LEFTY expression blocked by a LEFTY1-specific short hairpin RNA (shRNA), were established using OVISE cells (OV-shL1), which have relatively high LEFTY expression and wt p53. The two-independent OV-shL1 cell lines (#4 and #5) showed reduced LEFTY at both mRNA and protein levels, and increased pSmad2 expression ( Figure 5A). The cell lines also showed a tendency towards a low proliferation rate ( Figure 5B and Supplementary Figure 5A), along with an increase in both p21 waf1 and p27 kip1 expression, independent of p53 status ( Figure 5C).

Relationship of LEFTY expression with susceptibility to apoptosis in OCCCa
We examined the association between LEFTY expression and susceptibility to apoptosis in response to cytotoxic effects. Treatment of TOV-L1 stable cells with CDDP resulted in an increased proportion of both sub-G1 (apoptotic cells) and G2/M fractions, and a decrease in the G1 fraction ( Figure 6A and Supplementary Figure 6A), in line with results of decreased cell viability (Supplementary Figure 5B). The expression levels of an X-linked inhibitor of apoptosis protein (XIAP), pSmad2 and p27 kip1 were apparently decreased in the stable cells as compared to mock cells, in contrast to the increased expression of p53, bax, p21 waf1 , and cleaved caspase 3 ( Figure 6B). Similar findings were also evident in ES-L1 stable cells, with the exception of changes in the expression of several cell cycle-related molecules ( Figure 6C and 6D, and Supplementary Figure 5B and 6B). Bcl2/bax ratios were also significantly decreased by CDDP treatment in both stable cells ( Figure 6E). In clinical samples, apoptotic cells were readily detected in hematoxylin and eosin (HE)stained sections according to their characteristic features ( Figure 6F, left: arrows); the values positively correlated with apoptotic LIs as detected by TdT-mediated dUTPbiotin nick end labeling (TUNEL) assay ( Figure 6F, right). As shown in Figure 6G, several apoptotic cells also showed positive immunoreactivity for LEFTY in OCCCa tissues. The average number of apoptotic cells was significantly higher in tumors with high LEFTY scores (≧8) as compared with those in a low LEFTY score (<8) category.
Although, in the absence of CDDP treatment, apoptotic cell numbers were increased in OV-shL1 cells compared with mock cells (Figure 7A), the induction of apoptosis, as well as an increase in the G1 cell cycle fraction, in response to CDDP treatment was apparently decreased in the former ( Figure 7A and 7B, and Supplementary Figure 6C); increased expression of XIAP, pSmad2 and bcl2 (increased ratio of bcl2/bax; Figure 7C) was also observed. In contrast, changes in cell viability did not differ between OV-shL1 and mock cells (Supplementary Figure 5B). Similar associations were also observed by TUNEL assay ( Figure 7D).

Transcriptional inhibition of Smad2-dependent XIAP expression by LEFTY
Based on the findings related to apoptotic status, we examined whether the expression of XIAP, a wellstudied member of the inhibitor of apoptosis protein family, [23,24] was affected by Smad2. Treatment of Ishikawa cells with TGF-β1 led to an increase in XIAP expression at both mRNA and protein levels, accompanied by the increased expression of pSmad2 and LEFTY ( Figure 8A). Activation of the XIAP promoter was increased in a two-fold manner by Smad2 transfection, but this effect was apparently abrogated by the cotransfection of LEFTY1 ( Figure 8B), concordant with data demonstrating a considerable decrease in endogenous XIAP mRNA expression in both TOV-L1 and ES-L1 stable cells ( Figure 8C). In OCCCa tissues, cytoplasmic XIAP immunopositive cells displayed a heterogeneous distribution within tumor lesions and appeared to colocalize with pSmad2, but not LEFTY, immunoreactivity ( Figure 8D). This was in line with the significant positive correlation of the XIAP score with the pSmad2, but not LEFTY, score in OCCCa ( Figure 8E).

DISCUSSION
In the present study, using shotgun proteomics, many peptide fragments of both LEFTY1 and LEFTY2 proteins were more frequently detected in surgicallyresected OCCCa tissues compared with non-OCCCa tissues. This result was supported by IHC and western blot assays, which showed high LEFTY expression in OCCCa, regardless of several clinicopathological factors, including the survival. Moreover, the mRNA expression level of LEFTY1 was significantly higher than that of LEFTY2 in OCCCa, in line with our shotgun proteomics data showing that specific peptide fragments for LEFTY1 were more frequently detected as compared to those of LEFTY2. A similar finding was also observed in OCCCa as determined from proteomics data [25].
We found TGF-β1-induced transcriptional upregulation of LEFTY in OVISE and Ishikawa cells, probably through increased pSmad2 expression, whereas this association was lacking in TOV-21G cells. This may be due to the existence of anti-Smad proteins (Smad6 and Smad7) and repressors (SnoN and Ski) for TGF-β1-responsive genes in TOV-21G cell lines [10][11][12]26]. Given our data showing the absence of a direct correlation between LEFTY and pSmad2 expression in OCCCa tissues, it appeared that other factors may also contribute to the induction of LEFTY expression.
Epigenetic factors are considered to be a mechanism underlying the regulation of LEFTY expression mediated by TGF-β1. For example, TGF-β1 reduced LEFTY1 CpG island shore methylation by about 10 to 20%, resulting in increased LEFTY expression in human pancreatic and hepatic carcinoma cells [27]. However, we found that a reduction of LEFTY1 gene methylation in response to TGF-β1 did not occur in both OCCCa and endometrial carcinoma cells, which normally respond well to TGF-β1 stimulation, suggesting that epigenetic changes may be less important in the regulation of LEFTY1 expression in these cells.
Importantly, the overexpression of LEFTY appeared to contribute to the inhibition of aggressive growth in OCCCa, in line with a report showing that blocking TGF-β1 action by LEFTY overexpression may hinder aggressive tumor growth or lead to tumor regression. [19] This conclusion is supported by our observations of significantly decreased Ki-67 LI values in OCCCa with high LEFTY expression, as well as decreased LEFTY expression in postoperatively recurrent tumors. In addition, both TOV-L1 and ES-L1 stable cells with low pSmad2 expression showed a reduced cell proliferation rate, probably due to either an activated p53/p21 waf1 pathway or the direct or indirect induction of p27 kip1 expression.
In contrast, OV-shL1 cells also resulted in decreased cell proliferation, along with increased expression of pSmad2, p21 waf1 , and p27 kip1 . Given evidence showing that TGF-β family members can inhibit proliferation in a cell type-dependent manner through multiple mechanisms, including via the upregulation of p15 INK4b , p21 waf1 , and p27 kip1 , [28] it is possible that pSmad2-dependent p21 waf1 and p27 kip1 expression may also contribute to the inhibition of cell proliferation. In fact, changes in the expression of p21 waf1 and p27 kip1 in response to the TGF-β1/Smad2 axis were also evident in Ishikawa cells.
In line with our data showing the significantly higher number of apoptotic cells in OCCCa tissues with high LEFTY expression, the overexpression of LEFTY was closely linked with a susceptibility to apoptosis due to CDDP treatment in both TOV-L1 and ES-L1 stable cells. The apoptotic process was accompanied by a decrease in expressions of pSmad2 and XIAP and in the bcl2/bax ratio, and in an increase in cleaved caspase 3 expression. Given that the decrease in the bcl2/bax ratio results in the release of cytochrome c, activation of caspase 3, and subsequent apoptosis, [29] it is possible that an alteration in the Smad/XIAP axis by overexpressing LEFTY may be due to disruption of the mitochondrial membrane potential through changes in expression of bax and bcl2 expression.
Unexpectedly, although OV-shL1 cells in the absence of CDDP treatment also increased the numbers of apoptotic cells observed, this may be explained by the upregulation of pSmad2 due to the silencing of LEFTY1 since TGF-β family members are able to induce apoptosis [28]. In contrast, cells showed a lowered susceptibility to apoptosis in response to CDDP treatment, most likely through the increased expression of XIAP and bcl2, but not bax. In fact, XIAP protects endometrial cancer cells against various proapoptotic agents, including chemotherapeutic drugs [30]. Collectively, a complex regulatory mechanism in cell kinetics, including apoptosis and cell proliferation, may exist in OCCCa cells, which acts through signal loops involving the TGF-β/Smad/LEFTY axis, as well as XIAP and bcl family members. Several lines of evidence from the present study support the conclusion that XIAP expression is under the transcriptional control of pSmad2. Treatment of cells with TGF-β1 led to an increase in XIAP expression at both mRNA and protein levels, along with increased pSmad2 expression. Furthermore, transient transfection of Smad2 led to the transactivation of the XIAP promoter, despite the lack of consensus among Smad binding sequences in the XIAP promoter. This may be in accordance with data from a previous study showing that constitutive XIAP expression was indirectly regulated by the TGF-β1/Smad/NF-κB pathway. [31] Thirdly, XIAP immunoreactivity was positively correlated with pSmad2 status in OCCCa tissues. Finally, both TOV-L1 and ES-L1 stable cells exhibited a decrease in endogenous XIAP mRNA expression, corresponding to results showing that co-transfection of LEFTY1 abrogated Smad2-mediated transactivation of the XIAP promoter. In contrast, OV-shL1 cells resulted in the increased expression of both XIAP and pSmad2. Given that XIAP acts as an intracellular anti-apoptotic protein that functions as a direct inhibitor of caspases-3, -7 and -9, [32] it is therefore suggested that forced expression of LEFTY1 facilitates the initiation of apoptosis by caspase 3, altering the Smad2/ XIAP pathway. Upregulation of XIAP expression due to a loss of LEFTY1 causes a resistance to chemotherapy by tumor cells, resulting in tumor progression and recurrence. Although TGF-β1/Smad signaling was shown to be closely correlated with apoptosis through the downregulation of XIAP levels, [33] this discrepancy may result from cell type specificities.
Together, our observations suggest a model for the functional role of LEFTY in OCCCa (Figure 9). Up-regulation of LEFTY expression by TGF-β1/Smad signaling and/or other factors results in decreased cell proliferation, probably due to either an activated p53/ p21 waf1 pathway or the induction of p27 kip1 expression. Its overexpression is also associated with an enhancement of apoptotic features through the inhibition of Smad2mediated XIAP transcription and decreased bcl2 expression, as well as activation of the p53/bax pathway. The pSmad2 may also participate in susceptibility to apoptosis in certain conditions. Thus, LEFTY may be an excellent OCCCa-specific molecular marker and has antitumor effects through modifying cell proliferation and susceptibility to apoptosis. pathway or induces p27 kip1 expression, resulting in decreased cell proliferation. Its overexpression is also responsible for susceptibility to apoptosis due to the inhibition of the Smad2/XIAP axis and bcl2 expression, as well as the activation of the p53/bax pathway. www.impactjournals.com/oncotarget

Clinical cases
A total of 143 cases of OECa including 99 OCCCa, 13 OEmCa, 18 OSeCa, and 13 OMuCa, surgically resected at Kitasato University Hospital between 2000 and 2017, were selected from our patient records, according to 2014 criteria of the World Health Organization. [34] All tissues were routinely formalin-fixed (10%) and paraffin-embedded (FFPE). Of these, 16 FFPE samples (four of each subtype) were used for shotgun proteomics analysis. In addition, 42 fresh OECa samples (12 OCCCa, 11 OEmCa, 11 OMuCa, and eight OSeCa) were used for reverse transcription-polymerase chain reaction (RT-PCR) and/or western blot assays.
In OCCCa cases, none of the patients had received chemotherapy or any other preoperative treatment and most patients underwent postoperative adjuvant chemotherapy (administration of a paclitaxel, carboplatin, docetaxel, cisplatin-containing regimen). The mean age of patients was 55.3 years (range, 35 to 75 years). Forty-nine cases were subcategorized as clinical stage I, and 36 as stages II to IV, according to the criteria of the International Federation of Gynecology and Obstetrics (FIGO). [35] Ten were positive for nodal metastasis, while 75 were
The Q Exactive instrument was operated in a datadependent mode to automatically switch between full scan MS and MS/MS acquisition. Full-scan MS spectra (m/z 350−1500) were acquired in the Orbitrap with a 70,000 resolution at m/z 200 after the accumulation of ions to a 3 × 10 6 target value. The 10 most intense peaks with a charge state ≥2 from the full scan were selected with an isolation window of 2.0 Da and fragmented in the HCD collision cell with a normalized collision energy of 25%. Tandem mass spectra were acquired in the Orbitrap mass analyzer with a mass resolution of 17,500 at m/z 200 after accumulation of ions to a 2 × 10 5 target value. The ion selection threshold was 1 × 10 5 counts, and the maximum allowed ion accumulation times were 60 ms for full MS scans and 60 ms for tandem mass spectra. Typical mass spectrometric conditions were as follows: spray voltage, 2 kV; no sheath and auxiliary gas flow; heated capillary temperature, 250°C; and dynamic exclusion time, 60s.
Mass spectral data were processed, exported, and searched against a UniProt human database using SEQUEST by Proteome Discoverer (version 1.3; Thermo Fisher Scientific). Database search parameters were: peptide mass tolerance, 6 ppm; fragment tolerance, 15 ppm; enzyme was set to trypsin, allowing up to two missed cleavages; fixed modifications, carbamidomethyl (cysteine); and variable modifications, oxidation (methionine). The false discovery rate (FDR) was calculated by enabling the peptide sequence analysis using a decoy database. We used a 1% FDR as a cut-off to export results from the analysis. Quantitative analysis of shotgun proteomics data was achieved by spectral counting.

Immunohistochemistry
IHC was performed using a combination of microwave-oven heating and polymer immunocomplex (Envision, Dako) methods, as described previously [39,40].
For the evaluation of IHC findings, the scoring of nuclear and/or cytoplasmic immunoreactivity for LEFTY, pSmad2, and XIAP was performed, as described previously [39,40]. Briefly, the proportion of immunopositive cells among the total number of counted cells was subdivided into five categories as follows: 0, all negative; 1, <10%; 2, 10-30%; 3, 30-50%; and 4, >50% positive cells. Immunointensity was also subclassified into four groups: 0, negative; 1, weak; 2, moderate; and 3, strong immunointensity. IHC scores were generated by the multiplication of values of the two parameters. In addition, cases were divided into high and low LEFTY categories on the basis of the mean±SD values of the LEFTY score. Nuclear immunopositivity for Ki-67 was also counted in at least 1000 cells in five randomly selected fields. LIs were calculated as the number per 100 cells, as described previously [39,40]. www.impactjournals.com/oncotarget

In situ hybridization
Riboprobes for LEFTY1 and LEFTY2 containing nucleotides 1022 to 1274 and 1186 to 1467 of the LEFTY1 and LEFTY2 genes (Table 2), respectively, were generated by in vitro transcription, and ISH assays were performed using the GenPoint Tyramide Signal Amplification System (Dako), as described previously [41]. The ISH signal score was determined on the basis of the percentage of ISH signal-positive cells (1, less than 10% positive cells; 2, 10-30%; 3, 30-50%; 4, more than 50%) and the ISH signal-intensity (0, none; 1, weak; 2, moderate; 3, strong) with the multiplication of values of the two parameters.

Apoptosis and TdT-mediated dUTP-biotin nick end labeling assay
Apoptotic cells were identified in HE-stained sections, according to the criteria of Kerr et al [42]. A total of 20 fields were randomly selected, and the number of apoptotic cells was calculated by counting the mean number of apoptotic figures per 10 high power fields (HPFs), as described previously [39].
A TUNEL assay for the detection of apoptotic cells was also conducted using the In Situ Cell Death Detection Kit (Roche, Tokyo, Japan), according to the manufacturer's instructions. The number of positive cells was also analyzed by counting the mean number of TUNEL-positive cells per 10 HPFs.

Plasmids and cell lines
Full-length cDNA for human LEFTY1, LEFTY2, and Smad2, which were from Open Biosystems (Huntsville, AL, USA), were subcloned into pcDNA3.1 (Invitrogen, Carlsbad, CA, USA). The human LEFTY1 promoter (GeneBank accession number NM020997) between -3533 to +33 bp, the human LEFTY2 promoter (NM003240) between -4805 to +42 bp, and the human XIAP promoter (NG007264) encompassing -432 to -67 bp (where +1 represents the transcription start site) were amplified by PCR and subcloned into the pGL-3B vector (Promega). The identity of all constructs was confirmed by sequencing prior to use. The sequences of PCR primers employed in this study are listed in Table 2. TOV-21G and ES-2 cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA), and the OVISE and Ishikawa cell lines were also from the National Institute of Biomedical Innovation (Osaka, Japan).
The LEFTY1 expression plasmid or empty vector was transfected into TOV-21G and ES-2 cells, and stable overexpressing clones were established as described previously [39,40].
LEFTY1 specific shRNA oligonucleotides were designed using siDirect version 2 software. Single-stranded Lefty oligonucleotides were annealed and then cloned into BamHl-EcoRV sites of the RNAi-Ready pSIREN-RetroQ vector (Takara, Shiga, Japan), according to the manufacturer's instructions. The shLEFTY1 plasmid or non-specific shRNA constructs were transfected into OVISE cells, and stable knockdown clones were established.

Reverse transcription-and real time-polymerase chain reaction
To start, cDNA was synthesized from 2 μg of total RNA. Amplification by conventional RT-PCR was carried out in the exponential phase to allow comparison among cDNAs synthesized from identical reactions using specific primers ( Table 2). The intensity of individual signals was measured using ImageJ software version 1.41 (NIH, Bethesda, MD, USA). For quantitative analysis, real-time PCR was also conducted using a Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Fluorescent signals were detected using the ABI 7500 real-time PCR System, and data were analyzed using the associated ABI 7500 System SDS Software (Applied Biosystems). Primers for the GAPDH gene were also applied as described previously [39,40].

Flow cytometry
Cells were fixed using 70% alcohol and stained with propidium iodide (Sigma-Aldrich) for analysis of apoptotic cells. The prepared cells were analyzed by flow cytometry using BD FACS Calibur flow cytometer (BD Biosciences) and CellQuest Pro software (BD Biosciences), as described previously [39,40]. www.impactjournals.com/oncotarget

Cell counting Kit-8 assay
The quantitation of viable cell number in proliferation and after CDDP treatment was carried out using a Cell Counting Kit-8 (CCK-8; Dojindo Lab, Kumamoto, Japan), according to the manufacturers' instructions.

Methylation analysis of LEFTY1
The methylation status of the LEFTY1 was analyzed as described previously. [27] Briefly, genomic DNA extracted from cell lines using a Wizard Genomic DNA Purification kit (Promega) was treated by bisulfate using an EZ DNA Methylation-Gold kit (ZYMO Research, Orange, CA, USA). Bisulfate-treated DNA was amplified by PCR using specific primers as described previously. Amplicons were ligated into a pCR2.1 vector using the TA cloning kit (Invitrogen). Four positive clones were identified and sequenced from each sample. The DNA sequence generated was subjected to bioinformatics analysis to describe the CpG methylation status of these clones.

Mutation analysis for p53
Exons 5 to 9 of the p53 in TOV-21G and OVISE cells were amplified by PCR, and the products were subjected to direct sequencing PCR as described previously. [43] Statistics Comparative data were analyzed using Kruskal-Wallis, Mann-Whitney U-and Chi-square tests, and Spearman's correlation coefficient, as appropriate. The cutoff for statistical significance was set as p<0.05.

Author contributions
MA, MY and MS performed the majority of experiments, data analysis and drafted the manuscript. TM, SK, DK, RC, HT, YK, AY, YO, and YK provided support. All authors reviewed and approved the final manuscript.