Stearate Inhibits Growth Across Multiple Cancer Cell Lines
First, we evaluated the impact of palmitate and stearate on cellular function. Multiple human cancer cell lines (ovarian: OVCAR5, ES-2, SKOV3, OVCAR3, and OVCAR8; lung: H460, A549, H1650, and H1299; breast: Hs578T, MDA-MB0231, MDA-MB-453, and MCF7; and colorectal: Lovo, HCT116, HT290, DLD-1, and Caco-2) were cultured in palmitate- and stearate-supplemented media for 72 h, and their impact on cell proliferation was assessed. Overall, stearate inhibited cell growth to a greater extent than palmitate across all cell lines (p = 0.000252, Wilcoxon matched-pairs test). Furthermore, a k-means clustering analysis revealed that the cells could be segregated into three distinct groups based on their proliferative responses to the fatty acids: stearate ≈ palmitate, stearate > palmitate, and a group where neither fatty acid influenced cellular proliferation (Fig. 1a).
The ovarian cancer cell lines were exclusively categorized under the stearate > palmitate group, suggesting that a stronger inhibitory effect on ovarian cancer cell growth was seen with stearate than with palmitate. Additionally, similar experiments were conducted using the human mammary epithelial cell line MCF10A and the human ovarian surface epithelial cell line HOSE20. Despite both being immortalized normal cell lines, the sensitivity to stearate varied significantly between these two cell lines. In HOSE cells, the addition of stearate had a more pronounced impact on cell proliferation than that in MCF10A cells (p = 0.002165). To further investigate the potential anticancer effects of stearate, subsequent experiments primarily focused on the ovarian cancer cell lines classified into the stearate > palmitate group.
Conversely, in culture media supplemented with monounsaturated fatty acids (MUFA), palmitoleate and oleate did not markedly inhibit ovarian cancer cell growth. In fact, in certain cases, cell growth was enhanced, suggesting the differential effects of SFAs and MUFAs on the cancer cells (Supplementary Fig. S1a).
To identify which fatty acids significantly affect cell proliferation, we conducted experiments using long-chain fatty acids that are frequently encountered in dietary and cellular contexts7, 21. Ovarian cancer cell lines, namely OVCAR5, OVCAR8, SKOV3, ES-2, and OVCAR3, were treated with 50 µM each of palmitate, stearate, and oleate, and the cell viability was measured after 24, 48, and 72 h. Stearate markedly impeded cell growth in all lines from 24 h (Fig. 1b). Dose-response curves for each long-chain fatty acid revealed substantially lower IC50 values of stearate, at 36.96 ± 3.22 µM for OVCAR5 and 31.04 ± 1.97 µM for OVCAR8, than those of palmitate, at 1469.75 ± 74.61 µM and 74.97 ± 2.7 µM, respectively (Fig. 1c, Supplementary Fig. S1b).
Stearate Induces Apoptosis and DNA Damage in Ovarian Cancer Cells In Vitro and In Vivo
Next, we investigated whether stearate induced apoptosis like palmitate, as reported previously12, 14, 16. A flow cytometry analysis of Annexin V-positive cells showed that stearate increased apoptosis of OVCAR5 cells dose-dependently (Figs. 2a, b). As palmitate-induced apoptosis is related to DNA damage22, 23, we examined whether stearate similarly affects OVCAR5 cells. A dose-dependent increase in γH2AX expression was observed following 24 h stearate treatment (Fig. 2c). The obtained findings were corroborated through comparative experiments utilizing OVCAR8 cells; flow cytometry and western blot assays were conducted to elucidate the influence of stearate on apoptosis and DNA damage, respectively (Supplementary Figs. S3a-c).
To gain further insight into our findings, we conducted an in vivo study. We first used murine models to determine the impact of a HFD rich in oleate (O-HFD) on the growth of tumors derived from subcutaneously inoculated cancer cell lines (Supplementary Fig. S1c). Consistent with previous studies, tumor proliferation increased in mice harboring tumors derived from OVCAR5 cells that were fed an O-HFD (Supplementary Figs. S1d–g)2, 4, 24. Given the absence of notable disparities in body weight or blood insulin levels25 resulting from dietary variations, the findings prompted the hypothesis that the influence on tumor proliferation stemmed from the fatty acids themselves rather than alterations in the physiological condition of the mice. Initially, the fat in the O-HFD was mainly composed of oleate (64.3%) with a low stearate content (7.5%; Table S1).
Next, we investigated the effects of a S-HFD with a significantly higher per-calorie stearate content (33.35%; Table S1). Importantly, the S-HFD group exhibited a significant reduction in tumor growth compared with the normal-fat diet (NFD) group (Figs. 2d, e). Flow cytometry and IHC (immunohistochemistry) results revealed a higher degree of tumor apoptosis in the S-HFD group than in the NFD group (Annexin V-positive rate: 26.6% vs. 30.27%, cleaved caspase-3-positive area: 0.56% vs. 1.0%; Figs. 2f, g, Supplementary Figs. S2c, d). To evaluate DNA damage, IHC was performed for γH2AX, and the results showed a significant increase in the proportion of γH2AX-positive cells in the S-HFD group compared with that in the NFD group (H score: 38.2 vs. 57.5; Supplementary Figs. S2c, e). These findings were validated using SKOV3 cells (Supplementary Figs. S2f–h). Furthermore, no significant differences were observed in body weight and vital organs, including the liver and kidneys, between the S-HFD and NFD groups (Supplementary Figs. S2a, b). These data collectively demonstrate that stearate induces cytotoxicity, DNA damage, and apoptosis in ovarian cancer cells.
Stearate Induces Cytotoxicity via Endoplasmic Reticulum (ER) Stress and CHOP Activation
Next, we sought to elucidate the mechanisms underlying stearate-mediated cytotoxicity. We treated OVCAR5 cells with (i) DMSO, (ii) CAY10566 1 µM, (iii) stearate 50 µM + DMSO, (iv) stearate 50 µM + CAY10566 1 µM, (v) oleate 50 µM + DMSO, and (vi) oleate 50 µM + CAY10566 1 µM and performed RNA sequencing analysis. Principal component analysis (PCA) revealed that the presence or absence of stearate strongly contributed to PC1, whereas the presence or absence of oleate influenced PC2. However, 1 µM CAY10566 had limited effects (Fig. 4a).
Gene Ontology (GO) analysis identified 643 differentially expressed genes (DEGs), of which 401 were upregulated and 242 were downregulated between 50 µM stearate-treated and control OVCAR5 cells (false discovery rate [FDR] < 0.05, minimum fold change > 1.25; Supplementary Figs. S6a, b). The top 10 significantly upregulated GO categories were enriched in pathways associated with the unfolded protein response (UPR) and ER stress in 50 µM stearate-treated OVCAR5 cells compared with control cells (Fig. 4b; FDR < 0.05).
UPR signaling involves ATF6, IRE1α, and PERK pathways36. Our western blot analysis confirmed that stearate induced the concentration-dependent activation of UPR-related proteins, including ATF6, XBP-1 as a downstream transcription factor of IRE1α, and ATF4 as a downstream transcription factor of PERK. Moreover, the expression of pro-apoptotic transcription factor CHOP37 and apoptotic markers cleaved caspase-3 and γH2AX was upregulated (Figs. 4c, d).
We further examined whether the addition of oleate mitigated the activation of ER stress response pathways. Activation of ER stress response pathways was negated by the addition of 100 µM oleate to OVCAR5 and OVCAR8 cells (Figs. 4c, d). Furthermore, the addition of 1 µM CAY10566 enhanced the stearate-dependent activation of UPR-related proteins, CHOP, cleaved caspase-3, and γH2AX; however, this activation of the UPR pathway was almost abrogated by exogenous oleate (Figs. 4c, d).
Long-term exposure to mild ER stress or short-term exposure to severe ER stress induces CHOP-mediated apoptosis13, 38. To explore whether stearate induced apoptosis via CHOP, we generated CHOP-knockdown OVCAR5 and OVCAR8 cell lines via lentiviral infection of CHOP shRNA (Supplementary Fig. S6c, d).
Following inhibition of CHOP expression, the expression of cleaved caspase-3 and γH2AX, which was increased in a concentration-dependent manner by stearate treatment, was significantly reduced (Figs. 4e, f, Supplementary Figs. S6e, f). Moreover, CHOP knockdown significantly enhanced the resistance to stearate-induced cytotoxicity (Figs. 4g, h), indicating that stearate-induced cytotoxicity was mediated through ER stress and CHOP activation.
Overall, exogenous stearate activated ER stress response pathways, induced DNA damage, and inhibited the proliferation of ovarian cancer cells. Consistently, the addition of exogenous oleate attenuated the ER stress response pathway activated by stearate, reducing its toxicity in ovarian cancer.
Additionally, we have confirmed that the sensitivity to stearate and palmitate varies among cell lines (Fig. 1a). We investigated whether these differences were due to variations in ER stress response pathways. Regarding MCF10A cells (stearate-nonresponsive cells), we observed minimal CHOP induction by stearate, which differs significantly from the findings for HOSE and OVCAR5 cells (stearate-responsive cells) (Supplementary Figs. S7a, b). In the case of H1299 cells (stearate-nonresponsive cells), we found constant CHOP expression regardless of the addition of stearate, which was not decreased by oleate. These findings also significantly differed from those of HOSE and OVCAR5 (Supplementary Figs. S7a, b). We then investigated the responses of OVCAR5 and OVCAR8 cells to stearate and palmitate. Notably, palmitate induced CHOP expression in these cells, but to a lesser extent than stearate. Additionally, the activation of Cleaved Caspase3 by palmitate was less pronounced than that induced by stearate (Supplementary Figs. S7c, d). These findings highlight that the varying sensitivities to palmitate and stearate in different cell lines are primarily a result of their unique responses to the activation of the ER stress pathway.
Inhibition of Unsaturation Along with Dietary Supplementation of Stearate Hinders Tumor Growth, Which is Reversed by Oleate Supply
To validate our results obtained so far in vivo, we fed mice an S-HFD, O-HFD, or NFD (Supplementary Figs. S8a–c). In the S-HFD group subcutaneously injected with SCD1-knockdown (SCD1-KD) OVCAR5 cells, tumor growth was significantly inhibited compared with that in the NFD group (SCD1-KD & S-HFD vs. SCD1-KD & NFD; 0.125 g vs. 0.240 g, p = 0.006494; Figs. 5b, c). Conversely, the O-HFD group displayed significantly greater tumor growth than did the S-HFD and NFD groups (Fig. 5c). In experiments using sh-control cell lines, the S-HFD group exhibited stronger growth suppression than the NFD and O-HFD groups, although this trend was less pronounced than that observed in experiments using SCD1-KD cells (sh-control & S-HFD vs. sh-control & O-HFD; 0.2633 g vs. 0.4017 g, p = 0.006494; Fig. 5a). Furthermore, no significant differences in tumor growth were observed between sh-control and SCD1-KD cells in the O-HFD group (Fig. 5c).
In the S-HFD group subcutaneously injected with SCD1-KD OVCAR8 cells, the greatest tumor growth suppression was noted, with a significant difference compared with that in the O-HFD group (SCD1-KD & S-HFD vs. SCD1-KD & O-HFD; 0.02667 g vs. 0.0733 g, p = 0.019481; Figs. 5d–f). The same trend was observed when animals were injected with the sh-control cell line; however, no significant differences were observed between the S-HFD and O-HFD groups (sh-control & S-HFD vs. sh-control & O-HFD: 0.0433 g vs. 0.0533 g, p = 0.4848). Additionally, no significant differences were observed in tumor growth between mice injected with sh-control and SCD1-KD cells and fed on the O-HFD, as observed with OVCAR5 cells (Fig. 5f).
Next, we examined whether the UPR pathway, DNA damage, or apoptosis were modulated in vivo. IHC of OVCAR5 cell-derived tumors revealed marked upregulation of CHOP expression in the S-HFD group and the most significant upregulation in SCD1-KD cells (Figs. 5g–j). Conversely, CHOP expression was almost abrogated in the O-HFD group, regardless of whether the sh-control or SCD1-KD cells were used. We also assessed γH2AX and cleaved caspase-3 expression and observed trends consistent with those of CHOP expression. Similar results were obtained using OVCAR8 cells (Supplementary Figs. S8d–g).
We conducted further experiments using CAY10566 (Supplementary Fig. S9a). In mice injected with OVCAR5 and OVCAR8 cells, the CAY10566-treated group exhibited the most significant tumor growth suppression when fed the S-HFD compared with the NFD and O-HFD groups (Supplementary Figs. S9b–f, S10a–c). Tumor growth in the vehicle-treated group, those fed on the S-HFD, was the lowest, but this trend was less pronounced than that in the CAY10566 group. Moreover, no significant differences were observed between the vehicle and CAY10566 groups when fed on the O-HFD. The expression levels of γH2AX and cleaved caspase-3 were most significantly upregulated in the CAY10566 + S-HFD group, whereas almost no expression was observed in the O-HFD groups, irrespective of whether they were in the vehicle or CAY10566 group (Supplementary Fig. S9g–j, S10d–g). Assessments of stearate and oleate concentrations within tumor tissues demonstrated a stearate increment of 1.5- to 2-fold in the S-HFD-, O-HFD-, or CAY10566-administered group compared with that in the NFD + vehicle group. Despite the administration of CAY10566, O-HFD elevated oleate levels by approximately 1.5-fold, correlating with an actual proliferation enhancement in the tumors. Conversely, S-HFD in combination with CAY10566 administration resulted in a significant elevation of stearate to 185 pmol/mg while maintaining oleate levels at 50 pmol/mg, which was lower than that in the NFD-vehicle group, thus exhibiting a pronounced inhibitory effect on tumor growth (Supplementary Fig. S10h).
Overall, robust tumor-suppressive effects were achieved in vivo by increasing tumor stearate levels via S-HFD feeding, coupled with oleate inhibition mediated via SCD inhibition. Additionally, excessive intake of oleate through the O-HFD significantly diminished this effect.
Supply of Stearate along with Inhibition of Unsaturation Shows Significant Anti-proliferative Effects on Ovarian Cancer Patient-derived Xenograft (PDX) Models
To evaluate the applicability of our findings in the clinical setting, we next conducted experiments using PDXs. Conducting large-scale interventions to assess the effects of dietary changes is challenging; however, drug responses in PDXs have been suggested to correlate with patient clinical outcomes39. Therefore, we utilized two PDXs from distinct clinical backgrounds (PDX72 and PDX82; Supplementary Texts) that were established from patients treated at our institution. PDX82 was sourced from a 38-year-old female patient with stage IIIC HGSC harboring a BRCA2 mutation. This patient was sensitive to platinum-based chemotherapy and maintained no long-term evidence of disease under poly (ADP-ribose) polymerase inhibitor (PARPi)40, 41 treatment (Figs. 6a–c, Supplementary Figs. S11a–c). In PDX82 experiments, while treatment with CAY10566 alone showed limited effectiveness, tumor growth was significantly inhibited when these mice were fed S-HFD (NFD-CAY10566: 2685 mg vs. S-HFD-CAY10566: 970 mg, p = 0.0285; Figs. 6d–f). However, feeding mice on O-HFD led to significantly larger tumor sizes compared to the S-HFD-fed group, even with CAY10566 administration (S-HFD-CAY10566: 970 mg vs. O-HFD-CAY10566: 970 mg, p = 0.0285).
Another PDX, PDX72, was sourced from a 43-year-old female who developed platinum-resistant recurrent HGSC. The tumor was collected during secondary debulking surgery (SDS). Despite surgery, the patient relapsed quickly, and neither platinum-based chemotherapy nor anti-VEGF antibodies42 were effective, resulting in a poor prognosis (Figs. 6g–i, Supplementary Figs. S12a–c). Studies using PDX72 revealed that CAY10566 administration alone inhibited tumor growth, and this effect was further enhanced by feeding the S-HFD to mice (NFD-vehicle: 678.3 mg vs. NFD-CAY10566: 245.0 mg vs. S-HFD-CAY10566: 150 mg, p = 0.0021, 0.0043, respectively; Figs. 6j–l). However, despite CAY10566 treatment, mice fed on the O-HFD developed significantly larger tumors than those fed on the S-HFD (S-HFD-CAY10566: 150 mg vs. O-HFD-CAY10566: 798.3 mg, p = 0.0021).
IHC analysis results of these two PDX models regarding UPR, DNA damage, and apoptosis markers were consistent; the highest expression levels of CHOP, γH2AX, and cleaved caspase-3 were observed in the CAY10566 + S-HFD group, whereas these markers were significantly inhibited in the O-HFD group (Supplementary Figs. S11d–g, S12d–g).
Overall, combined administration of CAY10566 and S-HFD significantly suppressed tumor growth in two distinct PDX models with different clinical backgrounds and outcomes. Furthermore, even in cases sensitive to CAY10566 alone, tumor proliferation was enhanced when the O-HFD was consumed, suggesting that the antitumor effect of CAY10566 can be compromised by an O-HFD.