Human gastric cancer SNU-16 cells (No. 00016), SNU-620 cells (No. 00620), and SNU-484 cells (No. 00484) were purchased from the Korean Cell Line Bank (KCLB, Seoul, Korea). These cells were maintained in RPMI-1640 medium (GenDEPOT, Barker, TX) supplemented with 10% fetal bovine serum (Young In Frontier, Seoul, Korea) and 1% penicillin-streptomycin (GenDEPOT). The cells were cultured in a 5% CO₂ humidified incubator at 37°C.

The cells (1 × 10⁴ cells/well) were seeded into 96-well plates. Ten nanomolar of 17β-estradiol (E2758, Sigma-Aldrich, Inc., MO, United States) in 0.1% ethanol was administered to the cells for 24 or 48 h. A solvent as 0.1% ethanol was applied to control cells. The cell numbers were determined by methyl thiazolyl tetrazolium (MTT) cell proliferation assays [22]. Briefly, 20 μl of MTT (5 mg/ml, Sigma-Aldrich, Inc.) solution was added to each well and incubated for 3 h at 37°C. Thereafter, the media were removed and 150 μl of dimethyl sulfoxide (Sigma-Aldrich, Inc.) was added into each well for 30 min. Absorbance was measured using a microplate reader (Infinite M200 PRO, TECAN, Switzerland) at 560 nm. MTT assay was repeated twice. Data were represented as the mean ± standard deviation (n = 6).

All animal experiments were approved by the Institutional Animal Care and Use Committee of Duksung Women’s University (No. 2016-003-004) in accordance with the guidelines for the care and use of laboratory animals. Male and female NCr nude mice (CrTac:NCr-Foxn1nu, 5-weeks-old) were purchased from Nara Biotech (Gyeonggi, Korea). The animals were left to acclimatize for 1 week prior to any procedural work and kept at optimal conditions (20°C, 50% humidity, and a 12/12-h light/dark cycle). Diet was provided with drinking water ad libitum.

To generate a xenograft model, SNU-16 cells (5 × 10⁶ cells/mice) were subcutaneously injected into the right hind legs of female and male NCr nude mice (n = 5/group) [23]. This experiment was repeated twice. Tumor sizes were measured three times per week using calipers. Moreover, the animals were weighed and monitored regularly for signs of distress. Tumor volume was calculated using the following equation: Tumor   volume ( m m 3 ) = ( the   longest   length ) × ( the   shortest   length ) 2 / 2 (1)

Data are presented as the mean ± standard deviation.

Female NCr nude mice (8-weeks-old) were purchased from Nara Biotech. The animals were first anesthetized before a 1-cm incision was made on their dorso-lateral abdominal walls through which the ovaries were removed [24]. After confirming that there was no bleeding, the surgical wounds were closed. To allow the uterus to regress to a minimum, stable baseline, a time of 14 days elapsed. Then, SNU-16 cells (5 × 10⁶ cells/mice) were transplanted into the right hind legs of these mice and tumor growth was measured three times per week.

After the tumor was visually checked in the ovariectomized xenograft model, the groups were randomly divided. Then, an estrogen pellet (Innovative research of America, FL, United States) was implanted into a subcutaneous pocket on the dorsal flank of mice. Estrogen was consistently released at 8.3 μg/day from the pellet for 60 days [23]. To confirm estrogen effect, mice was anesthetized by isoflurane (Forane, JW Pharmaceutical, Seoul, Korea) after the final tumor volume measurement. The blood was collected from abdominal vein of anesthetized mice and the uterus was isolated. The blood was leaved at room temperature for 30 min, and then the clot was removed by centrifuging at 3,000 rpm for 10 min in 4°C. The supernatant as a serum was taken to a new tube. Serum estrogen level was measured with electrochemiluminescence immunoassays on Roche Cobas 8,000 analyzer (Roche Diagnostics, Basel, Switzerland) by DKKorea (Seoul, Korea) [25]. The uterus weight was measured using an electronic weighing balance (OHAUS Instruments, Shanghai, China).

After the final tumor measurement, the SNU-16-derived tumor tissues were isolated from sacrificed mice, immersed, and stored in liquid nitrogen until analysis. Total RNA from the tumor tissues was extracted with the RNeasy Mini Kit (Qiagen, MD, United States) according to the manufacturer’s instructions. In brief, 30 mg of frozen tumor tissues were submerged in 600 μl of buffer RLT and then immediately homogenized using Tissueruptor II (Qiagen) on ice for 30 s. The lysate was centrifuged at 15,000 rpm for 3 min in 4°C. Ethanol was added to the supernatant (lysate). The lysate was loaded into an RNeasy mini spin column. Contaminants including DNA were removed, and then RNA was eluted in water. Then, the quality and concentration of the RNA samples were determined based on an electropherogram and RNA integrity was calculated using the Agilent 2,100 BioAnalyzer (Agilent, CA, United States). Samples with an RNA integrity number greater than six were sequenced using an Illumina HiSeq 2,500 (Illumina, CA, United States). Based on the fragments per kilobase million (FPKM) value of each gene, differentially expressed genes (DEGs) were selected. DEG analysis was performed using the Cuffdiff tool with p-value ≤ 0.00005 and q-value (a multiple-test corrected p-value) ≤ 0.05, whereas the cut-off for the gene ontology analysis was p < 0.0001 [26, 27].

Publicly available data (Supplementary Table S1) in Kaplan-Meier Plotter (kmplot.com) were used in this study; patient clinical data were obtained from the Gene Expression Omnibus database (GEO database: GSE15459, GSE22377, GSE29272, GSE38749, and GSE62254 datasets). The data contained mRNA expression profiles of 112 female patients and 127 male patients with diffuse type-gastric cancer based on microarray. Overall survival (OS) analysis was shown as a Kaplan-Meier curve [28]. Data were divided into two groups based on the gene expression levels. A cut-off refers to the best performing threshold. The hazard ratios (HR) with 95% confidence intervals and p-values were automatically calculated in Kaplan-Meier Plotter.

Tumor tissues were lyzed in radioimmunoprecipitation assay buffer (RIPA buffer, GenDEPOT) with Xpert protease inhibitor cocktail solution (P3100, GenDEPOT) and Xpert phosphatase inhibitor cocktail solution (P3200, GenDEPOT). The proteins (10 μg) were separated by 10% or 12% SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (Millipore, Darmstadt, Germany). The membranes were then blocked with 5% skimmed milk in TBST (Tris-buffered saline with 0.1% tween 20) and incubated with primary antibodies for artemin (GeneTex, CA, United States, 1:1,000), ERα (Santa Cruz Biotechnology, sc-130072, 1:1,000), ERβ (Santa Cruz Biotechnology, sc-53494, 1:1,000), GPER (Abcam, ab188999, Cambridge, United Kingdom, 1:1,000) and β-actin (Sigma-Aldrich, Darmstadt, Germany, 1:5,000) overnight at 4°C. The next day, the membranes were incubated with secondary antibodies (goat anti-mouse IgG-HRP conjugated or goat anti-rabbit IgG-HRP conjugated, 1:3,000) for 3 h at room temperature. Proteins were visualized using enhanced chemiluminescent solution and detected with a Chemi-Doc (FluorChem E system, San Jose, California, United States).

Statistical analysis was performed using GraphPad Prism 7 (GraphPad Software Inc., CA, United States). The data were analyzed using Student’s t-test or one-way ANOVA test followed by Tukey’s multiple comparisons test as post hoc test. Significant differences were assumed for p-values less than 0.05.

Xenograft mouse models were established to investigate whether tumor progression varied depending on sex differences. The female gastric cancer cell line SNU-16 was used to induce tumors in female and male mice and the resulting tumors were examined; as shown in Figure 1, tumor progression was significantly faster in female mice than in male mice. It was further investigated whether other diffuse-type gastric cancer cells are affected by the sex of mice. As shown in Supplementary Figure S1A, results of SNU-620 female diffuse gastric cancer cells were in accordance with those of SNU-16 cells although the statistical significance was not shown. However, SNU-484 male diffuse gastric cancer cells showed no difference in tumor growth between male and female xenograft models (Supplementary Figure S1B). Only female-originating cancer cells showed enhanced tumor growth in female mice. To elucidate the involvement of female sex hormones in tumor progression, the expression level of ERs was investigated in these gastric cancer cell lines. There was no difference in the expression level of ERs according to sex or cancer type in gastric cancer cell lines (Supplementary Figure S2). These results suggested that endogenous factors in female mouse xenograft models affect tumor growth rather than a direct ER-mediated pathway in female diffuse-type gastric cancer cells.

To better understand the difference reported in Figure 1, tumor tissues obtained from these xenograft models were compared using total RNA sequencing (RNA seq). The data of RNA seq were analyzed based on human-originating genes excluding mouse-originated genes. We observed DEGs between female and male mice (Figure 2A); transcripts of DEGs with a log₂fold_change (fc) > absolute value of 1.5, p-values ≤ 0.00005, and q-values ≤ 0.05 are summarized in Table 1. As shown in Table 1, interferon gamma receptor 2 (IFNGR2) had lower expression levels in tissues from female mice than in male tissues, whereas IQ motif containing E (IQCE), transient receptor potential cation channel subfamily M member 4 (TRPM4) and structure-specific endonuclease subunit SLX4 (SLX4) exhibited higher expression in females than in males. To investigate the association between our in vivo findings and clinical data, we compared genes obtained from our RNAseq data with those from clinical databases. As shown in Figure 2B, the OS of patients with diffuse-type gastric cancer obtained from the GEO database was reanalyzed depending on expression levels of DEGs. There is a trend toward better survival at high IFNGR2, IQCE, and TRPM4 expression levels and at low SLX4 expression level but cannot claim there was any association in women. Only IFNGR2 has a positive prognostic value, and only among all patients.

Considering that SNU-16-derived tumors progressed faster in female mice than in male mice during the evaluation of cancer xenograft models (Figure 1), we investigated whether estrogen could directly affect the proliferation of these cells. As shown in Figure 3, cell proliferation was stagnant after 24 h of 10 nM of 17β-estradiol (E2 as estrogen) treatment, but there was no significant difference compared to that in the control group.

In vitro, E2 showed no direct effect on cell proliferation (Figure 3). Therefore, we investigated whether E2 deficiency and exogenous E2 treatment could affect tumor progression in xenograft models. Ovariectomized mice were used to better understand the factors involved during tumor progression in the female xenograft model. To ensure an accurate comparison, we confirmed that estrogen levels in the blood and the uterus weight were decreased in ovariectomized mice, and then estrogen treatment was restored (Supplementary Figure S3). In the ovariectomized xenograft mouse (OVX) model, SNU-16-derived tumor growth was detected (Figure 4A). The OVX group showed slightly delayed tumor growth compared to the normal xenograft group (control group). However, no significant difference between the two groups was noted. Interestingly, the estrogen-treated group (OVX with E2 group) had significantly enhanced tumor growth when compared to that in either OVX or control groups (Figure 4A). To elucidate the mechanism of estrogen-induced tumor growth, the expression of ER subtypes was determined by western blotting (Figure 4B). ERα and ERβ expression levels were slightly increased in ovariectomized mice to generate a sensitive response to low estrogen levels. Surprisingly, GPER expression levels were induced in the OVX with E2 group more than in the OVX group. These results suggested that E2 might affect tumor growth through the regulation of ER expression levels, and especially GPER. To elucidate the effect of E2 on SNU-16-derived tumor growth in ovariectomized mice, we performed RNA seq and summarized DEGs among these groups (Supplementary Table S2). Among genes expressed only in OVX with E2 group, artemin (ARTN) was known as a factor involved in the cancer progression [29]. Thus, we first investigated the change in the level of ARTN expression by western blotting. ARTN levels were significantly decreased in the OVX group; however, E2 treatment significantly induced ARTN expression (Figure 4C). Based on the Kaplan-Meier curve using the GEO database, the OS of diffuse-type gastric cancer patients showed a negative association with GPER expression levels. A trend of OS with ARTN expression was shown, but no statistically effects were measured (Figure 4D).