SCD1/FADS2 fatty acid desaturases are aberrantly upregulated in metastatic OvCa cells
This study utilized omental conditioned medium (OCM) to mimic the omental or ascites microenvironment and demonstrate that the cellular composition of UFAs, especially mono-UFAs (MUFAs), was significantly increased by approximately 12% in OvCa cell lines, such as ES-2 cells. This increase in MUFAs was intimately correlated with the loss of saturated-FAs (SFAs) in ES-2 cells cocultured in OCM (Fig. 1A, B). Under similar conditions, the composition of UFAs increased by approximately 6.5% in another OvCa cell line, OVCA433 cells, compared with the DMEM control (Fig. 1A, B). Lipidomic analysis revealed that these UFAs primarily consisted of MUFAs, including palmitoleic acid (C16:1, POA) and oleic acid (C18:1, OA), and PUFAs, such as linoleic acid (C18:2, LOA), alpha-linolenic (C18:3, α-LOA), gamma-linolenic (C18:3, γ-LOA) and arachidonic acid (C20:4, AA), in both OVCA433 and ES-2 cells (Fig. 1C). QPCR analysis generally showed that relatively higher mRNA levels of SCD1 and FADS2 were detected in samples of omental metastatic tumors than in the respective primary tumors (n = 10 pairs) (Fig. 1D). SCD1 and FADS2 levels were elevated by 16.13-fold and 11.34-fold, respectively, in OVCA433 cells with OCM (Fig. 1E). Consistently, western blot analysis showed that both SCD1 and FADS2 were commonly upregulated in OvCa cell lines compared with human immortalized epithelial ovarian cells (HOSEs) (Fig. S1A). Moreover, higher expression levels of SCD1 and FADS2 were also observed in spheroids derived from the malignant ascites of OvCa (Fig. 1F). Of note, multiparametric immunohistochemical (IHC) analysis revealed that SCD1/FADS2 were remarkably upregulated in omental metastatic OvCa compared with their paired primary tumors (n = 10 pairs) (Fig. 1G). Intriguingly, in silico analysis of the Cancer Genome Atlas (TCGA) showed that SCD1 and FADS2 were the major isoforms and were highly expressed in tumor tissues (Fig. S1B), and were associated with poor overall survival (OS) (Fig. S1C), advanced stage (Stage IV), and tumor recurrence (Fig. S1D) in OvCa. Collectively, these findings show that SCD1 and FADS2 are aberrantly overexpressed in metastatic OvCa cells, especially in the lipid-enriched microenvironment.
Upregulated SCD1 and FADS2 positively enhance fatty acid desaturase activities
Liquid chromatography with tandem mass spectrometry (LC-MS/MS) analysis demonstrated that both the SCD1 and FADS2 desaturation index were increased over ~14.33-fold and ~1.33-fold, respectively, in ES-2 cells cocultured in OCM (Fig. 1H). Similarly, in OVCA433 cells cocultured with OCM, both the SCD1 and FADS2 desaturation index were increased more than ~3.65-fold and ~1.74-fold, respectively (Fig. 1I), indicating concomitant upregulation in the expression and desaturase activities of both SCD1 and FADS2 in OvCa cells. To validate the vital functional role of both SCD1 and FADS2, selective inhibitors of SCD1, e.g., CAY10566 and FADS2, e.g., sc26196, were exploited, and the results of UFAs quantification assay (Colorimetric) showed that pharmaceutical inhibition of SCD1 or FADS2 led to 32% and 16% reductions in UFA content, respectively, in OVCA433 cells upon treatment with OCM (Fig. 1J). These outcomes substantiate that the aberrant upregulation of SCD1 and FADS2 is the prominent FADS exhibiting fatty acid desaturase activities in OvCa cells derived from the lipid-enriched microenvironment.
SCD1 and FADS2 are required for the oncogenic capacities of OvCa cells
CRISPR/Cas9-mediated SCD1/FADS2 depletion was firstly established in OvCa cell lines, such as OVCA433 and ES-2 cells, to generate SCD1low/− or FADS2low/− clones to examine their functional roles (Fig. 2A). Functionally, cell proliferation assay showed that silencing of SCD1 or FADS2 significantly hindered the cell growth of OVCA433 cells (Fig. 2B) and ES-2 cells (Fig. S2A) by ~2-fold upon three days of culture. The findings herein were in accordance with the effect of using SCD1/FADS2 inhibitors, in which OCM-mediated cell growth was remarkably repressed by at least 1.86-fold in OVCA433 cells compared with the controls (Fig. 2C). In addition, sphere-formation capability was substantially inhibited by more than 1.6-fold in both SCD1low/− and FADS2low/− clones of OVCA433 and ES-2 cells when compared with respective wild-type (WT) controls (Fig. S2B).
On the other hand, apoptosis was markedly promoted in SCD1low/− or FADS2low/− clones of OVCA433 cells by 10.2-fold and 12.1-fold, respectively, compared with the controls (Fig. 2D). Similar outcomes were observed in ES-2 SCD1low/− cells and ES-2 FADS2low/− cells, yielding a significant increase in cell apoptosis by 12.79-fold and 14.7-fold (Fig. S2C) compared with control cells. Furthermore, apoptotic cell death was enhanced by 1.7-fold and 1.42-fold in OVCA433 cells upon treatment with SCD1/FADS2 selective inhibitors compared with the OCM cultured controls (Fig. 2E). Conversely, migration assays showed an approximately 2-fold reduction in the migration of OVCA433 (Fig. 2F) and ES-2 cells (Fig. S2D) with SCD1/FADS2 knockout compared to the respective controls. Similarly, a 25% decrease in OCM-mediated cell migration was seen in OVCA433 cells treated with SCD1/FADS2 selective inhibitors compared with the controls (Fig. 2G). Silencing or pharmacological inhibition of SCD1/FADS2 in OVCA433 (Fig. 2H, I) and ES-2 cells (Fig. S2E) correspondingly led to a small but significant increase in the G1/S cell cycle phase compared with the respective controls. Collectively, upregulation of SCD1/FADS2 expression and FA desaturation activities are accompanied by the pathogenesis of OvCa.
Inhibition of SCD1/FADS2 impairs tumor initiation and membrane fluidity
The spheroid formation ability of ascites-derived OvCa cells was suppressed by 32% and 29% upon treatment with SCD1 and FADS2 selective inhibitors, respectively (Fig. 3A). As organoid self-renewal ability is another hallmark of stemness, treatment with SCD1 and FADS2 inhibitors led to reductions of 67% and 53% of the organoid self-renewal capability of OvCa cells, respectively (Fig. 3A). Western blot analysis further revealed that stemness markers of KLF4 and BMI1 were downregulated by silencing of SCD1 or FADS2 in OVCA433 cells (Fig. 3B). Compared with the negative controls (OCM pretreated with the lipid removal reagent, Cleanascite), OvCa cells cocultured in the lipid-riched OCM showed an increase of 18% in membrane fluidity. However, cotreatment with CAY10566 (10 nM) led to a 23% decrease in membrane fluidity, and cotreatment with sc26196 (100 nM) reduced the membrane fluidity to a level similar to that of the negative controls (Fig. 3C). Collectively, these results suggest that activated SCD1/FADS2 is required for aggressive tumor initiative potential, and UFA-mediated membrane fluidity is crucial for maintaining OvCa cell stemness within the ascites microenvironment.
Inhibition of SCD1/FADS2 impairs EMT transition in OvCa cells
Notably, silencing of SCD1 or FADS2 significantly suppressed mesenchymal markers (Vimentin) and EMT regulators (ZEB1, SNAIL, and SLUG) in ES2 SCD1low/- and FADS2low/- cells (Fig. 3D). In contrast, enforced expression of SCD1 or FADS2 reversed the EMT markers in OVCA433 cells (Fig. 3E). The data showed that Vimentin was downregulated, and E-cadherin expression was upregulated compared with the controls (Fig. 3F). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of our transcriptomics on OVCA433 SCD1low/- and FADS2low/- cells supported that SCD1 and FADS2 participated in EMT-related biological processes and molecular functions (Fig. S3A-D). In addition, upon treatment with CAY10566 (25 nM) or sc26196 (500 nM), the number and size of tumor organoids significantly decreased, and the spindle-like morphology was markedly disrupted by >1.89-fold compared with the controls (Fig. 3G), suggesting that OvCa cells in EMT status were vulnerable to SCD1/FADS2 inhibition-based therapies.
SCD1 and FADS2 are involved in protecting OvCa cells from oxidative stress
To assess ATP and ROS production levels in OvCa cells, we utilized a Cell Mito Stress assay to measure ATP production. Of note, OCM-cocultured OVCA433 cells exhibited 2.62-fold higher ATP production than DMEM (Fig. 4A). In addition, ATP-linked basal OCR and maximal OCR revealed elevations of approximately 2.1- and 2.04-fold, respectively. Furthermore, the rescue experiment showed that OVCA433 cells cocultured in lipid-depleted OCM lost the capacity for ATP production, indicating that the lipid-enriched microenvironment enhances the lipid metabolic activities of OvCa cells (Fig. 4A). Similarly, the fluorescence intensity of H2DCFDA indicated that OCM-cocultured OVCA433 cells exhibited a moderated elevation of ROS production compared with DMEM (Fig. 4B).
We further confirmed SCD1/FADS2 is required to protect OvCa cells against ROS overproduction and ROS-stimulated cellular damage. Treatment of either CAY10566 (10 nM) or sc26196 (100 nM) in OVCA433 cells with OCM caused increased cellular ROS production by more than 1.5-fold compared to the DMEM negative controls and OCM positive controls. Notably, suppression of SCD1/FADS2 simultaneously induced a remarkable increase of more than 4.36-fold of ROS in the OCM coculture system (Fig. 4B). In contrast, we found that ATP production dramatically decreased, indicating that abrupt oxidative phosphorylation might be unable to meet the cellular ATP demands to support cell aggressiveness (Fig. 4A). Our transcriptome profiling also supported this postulation that ROS-related pathways were significantly enriched in either SCD1- or FADS2-depleted clones, including Rho GTPases, MAPK signaling, ferroptosis, and other oxidative stress-related processes in OvCa cells (Fig. S4A-B). Together, these findings propose that ascites-derived OvCa cells obtain high ATP production from the lipid-enriched microenvironment, while SCD1 and FADS2 not only exert fatty acid desaturase activity in producing UFAs but also equilibrate redox homeostasis to avoid ROS-mediated cell death.
Given that the inhibition of SCD1/FADS2 could elevate excessive intracellular ROS, it was hypothesized that the inhibition of SCD1/FADS2 could induce lipid peroxidation in ascites-derived OvCa cells. As expected, the accumulation of lipid peroxidation was elevated by approximately 20% in OCM-cocultured OVCA433 cells upon pharmacological inhibition of SCD1/FADS2 compared with the DMEM controls (Fig. 4C). Cotreatment of Fer-1 (10 μM) with pharmaceutical inhibitors of SCD1/FADS2 in OCM-cocultured OVCA433 cells rescued the elevated lipid peroxidation to the DMEM controls (Fig. 4C). Notably, under the same condition of treatment, the cell viability was correspondingly decreased as the subsequently result of excessive cellular ROS and lipid peroxidation (Fig. 4D). The findings herein demonstrate that SCD1/FADS2 inhibition-induced excessive ROS with lipid peroxidation are severely harmful to the survival of OvCa cells in the lipid-enriched microenvironment.
Inhibition of SCD1/FADS2 promotes ferroptosis in ascites-derived OvCa cells
KEGG enrichment analysis of our transcriptional profiling revealed that the ferroptosis pathway was enriched as one of the critical signaling pathways after depletion of SCD1 or FADS2 in OvCa cells (Fig. 4E). OvCa cells in the lipid-enriched microenvironment generate ATP via FAO along with ROS production. It is thus hypothesized that a lipid-enriched microenvironment possibly enhances PUFA-based lipid peroxidation in OvCa cells when cocultured with an SCD1/FADS2 inhibitor. Accordingly, abruptly elevated ROS might trigger apoptosis and ferroptosis to suppress the survival of OvCa cells, as summarized in Fig. 4F.
Based on our transcriptome profiling, Gene Set Enrichment Analysis (GSEA) analysis indicated that ferroptosis-related genes were extensively enriched in OVCA433 SCD1low/- clones (P = 0.00086) (Fig. S5A). Mechanistically, given the crucial role of GPX4 in ferroptosis, it is necessary to evaluate the expression level of GPX4 in relation to OvCa cells. To this end, western blot analysis showed that depletion of SCD1/FADS2 led to a decline in GPX4 expression in OVCA433 cells (Fig. 5A). In contrast, forced expression of SCD1/FADS2 upregulated the level of GPX4 in OVCA433 cells (Fig. 5B). The transcriptomic analysis confirmed these findings that the level of GPX4 was significantly hampered in OVCA433 SCD1low/- and FADS2low/- cells, suggesting that SCD1/FADS2 are actively involved in the transcriptional activity of GPX4 (Fig. S5B). Moreover, pharmaceutical inhibition of SCD1/FADS2, alone or in combination, was performed to examine the expression changes of GPX4. The results showed that the combined inhibition of SCD1 and FADS2 remarkably reduced the level of GPX4, and this reduction was equivalent to treatment with the ferroptosis inducer Erastin (5 μM) (Fig. 5C). Using a glutathione assay, pharmaceutical inhibition of SCD1 and FADS2, individually or in combination, led to a reduction in the GSH/GSSG ratio by approximately 50% compared to the DMSO controls in ascites-derived OVCA433 cells (Fig. 5D). These findings suggest that SCD1 or FADS2 deficiency attenuated GPX4 expression and activity in OvCa cells.
Iron quantification assays revealed that both Erastin (5 μM) and the combination treatment of CAY10566 (10 nM) and sc26196 (100 nM) in OVCA433 cells elevated the overall intracellular ferrous ion concentration (Fe2+) by approximately 1.3-fold compared with the DMSO negative control and Cleanascite-pretreated OCM (Fig. 5E). This finding suggests that the suppression of SCD1/FADS2 activities led to the elevation of redox-active labile iron levels that, in turn, resulted in deregulated ROS accumulation and subsequent lipid peroxidation.
Inhibition of SCD1/FADS2-induced ROS is associated with ferroptosis
The inhibition of SCD1/FADS2 downregulated GPX4 and elevated intracellular ROS simultaneously, it caused the level of lipid peroxidation was elevated by 20% in OCM-cocultured OVCA433 cells upon cotreatment with Erastin and/or in combination with SCD1low/-/FADS2low/- (Fig. 5F). Consistently, the accumulation of lipid peroxidation was observed in OvCa cells upon pharmaceutical inhibition of SCD1/FADS2. The effect was equivalent to that of the ferroptosis inducer Erastin. Fer-1 rescued lipid peroxidation and led to a shift in fluorescence from green (oxidation) to red (nonoxidation) (Fig. 5G). To investigate the mechanism of SCD1/FADS2 in preventing lipid peroxidation in metastatic OvCa cells, paired primary and metastatic ovarian tumor tissues were recruited. Western blot analysis showed that the expression levels of SCD1, FADS2, GPX4, and TFR1 in omental metastatic tumors were relatively higher than their counterpart primary tumor tissues (Fig. 5H). Multiparametric IHC analysis also confirmed that GPX4 was highly expressed in omental metastatic tumor cells (Fig. 5I). To further substantiate these findings, analysis of a publicly available dataset revealed that GPX4 was highly expressed in the advanced stage of OvCa, accompanied by high recurrence and platinum resistance (Fig. S5C). Moreover, a positive correlation between SCD1/FADS2 and SCD1/GPX4 in OvCa tumor tissues was observed by regression analysis in TCGA-OV (Fig. S5D). These data suggest that metastatic OvCa cells in the lipid-enriched microenvironment exhibiting high lipid metabolic activities to support their aggressive oncogenic properties and protect against ROS-mediated cell death or lipid peroxidation through the activated SCD1/FADS2/GPX4/TFR1 signaling axis.
The combination of cisplatin with SCD1/FADS2 inhibitors exhibits inapparent cell toxicity in noncancerous cells
Cisplatin is one of the first-line chemotherapeutic drugs for OvCa 12. The 50% inhibitory concentration (IC50) values of cisplatin in OVCA433 cells and ES-2 cells were 5.17 μM and 3.13 μM, respectively (Fig. S6A), in line with other publications stating that a dose of cisplatin (0.5-5 μM) could abolish cancer cell growth 13, 14. Thus, a low dose of cisplatin (2 μM) was selected to perform the following combination treatment. Accordingly, a low dose of CAY10566 (5 nM) and sc26196 (100 nM) was chosen to perform the combinational treatment (Fig. S6B). The XTT cell proliferation assay supported the combination of cisplatin with SCD1/FADS2 inhibitors exhibit unobvious cell toxicity in noncancerous cells (Fig. S6C).
SCD1/FADS2 inhibitors attenuate cisplatin resistance in OvCa cells
Upon cotreatment of CAY10566 (5 nM) or sc26196 (100 nM) with cisplatin (2 μM) in OVCA433 cells, cell viability was synergistically retarded in a dose-dependent manner with a combination index (CI) less than 1 (Fig. 6A). Sole treatment with cisplatin (2 μM) and pharmacological inhibition of SCD1 and FADS2 in OVCA433 cells merely caused a 2-fold enhancement of the apoptotic rate (Fig. 6B), whereas cotreatment of these inhibitors with cisplatin strengthened the apoptotic rate to 3.28-fold (Fig. 6B). In addition, migration analysis demonstrated that the combined treatment of SCD1 and FADS2 inhibitors caused an approximately 2-fold reduction in cell migration, while a combination of cisplatin further reduced the migration of OVCA433 cells by 32% (Fig. 6C). Treatment with SCD1/FADS2 inhibitors of OVCA433 cells led to G1/S cell cycle arrest compared with the controls (89.15% vs. 79.14%) (Fig. 6D), while combinational treatment of these inhibitors with cisplatin generated substantial G1/S cell cycle arrest compared to the controls (91.06% vs. 79.14%) (Fig. 6D). These findings suggest that cotreatment with SCD1/FADS2 inhibitors could synergistically inhibit cell growth and cell aggressiveness by promoting G1/S cell cycle arrest in OvCa cells. In addition to in vitro studies, we also provide a comprehensive preclinical evaluation of combining CAY10566, sc26196, and cisplatin in OvCa patient-derived organoids. Combination treatment with CAY10566 (25 nM) and sc26196 (500 nM), with or without cisplatin (2 μM), significantly disrupted the spindle-like morphology of the patient-derived organoids by more than 2.43-fold (Fig. 6E). As apparent from the IF assay, Vimentin was downregulated, and E-cadherin was upregulated after combination treatment with CAY10566, sc26196, and cisplatin (Fig. 6F), indicating that OvCa cells with a mesenchymal phenotype during malignant transformation are susceptible to this combination regimen.
An isogenic paired OvCa cell line, PEO1 (cisplatin-sensitive) and PEO4 (cisplatin-resistant), was chosen to identify this assumption. Combination treatment of PEO1 and PEO4 cells with CAY10566 (5 nM), sc26196 (100 nM), and cisplatin (2 μM) synergistically displayed the maximal inhibitory effect on cell viability with increasing values of fraction affected (Fa), and a combination index (CI) less than 1 (Fig. 6G). Upon combination treatment of PEO1 and PEO4 cells, the expression of the mesenchymal marker Vimentin, the migration protein MMP2, and the ferroptosis marker GPX4 was severely decreased, whereas the expression of the epithelial marker E-cadherin was apparently elevated compared with either treatment with cisplatin or SCD1/FADS2 inhibitors (Fig. 6H). Collectively, these data provide solid preclinical evidence supporting the potential clinical use of SCD1/FADS2 inhibitors in combination with cisplatin for eradicating OvCa metastasis.
SCD1/FADS2 inhibitors and cisplatin synergistically enhance the in vivo anticancer effect
To further evaluate the synergistic, therapeutic effect of combination drugs in OvCa progression and peritoneal metastasis, we first identified them in a murine omentum culture ex vivo model (Fig. 7A) 6. C57BL/6 mouse omental tissues with metastatic GFP-ID8 cell colonies were treated for 30 days with vehicle, CAY10566 (25 nM), sc26196 (500 nM), cisplatin (5 μg/mL), or a combination. According to the fluorescence microscopy results, in each treatment group compared with the vehicle group, the number of metastatic tumor colonization had a decrease of 43–59%, whereas the two inhibitors combined with cisplatin had a dramatic inhibition of 87%; in the rescue group, Fer-1 (20 μM) neutralized the efficacy of the three-drug combinations, and the colony number showed a decrease of only 75% (Fig. 7B).
Next, we established an in vivo omental metastasis mouse model by intraperitoneal (i.p.) GFP and luciferase dual-labeled ES-2 cells (Fig. S7A). The schematic diagram shows the experimental timeline of the treatment (Fig. 7C). Peritoneal tumor-bearing SCID mice were treated with vehicle, CAY10566, sc26196, cisplatin, separately or in combination for two weeks. Notably, severe peritoneal metastasis was solely observed in the control group (Fig. 7D and S7B), the luminescence quantification value revealed a decrease of approximately 32.7% to 61.4% (Fig. 7D), and the GFP fluorescence value exhibited a decrease of approximately 33.2% to 81.4% (Fig. S7B). The control group mice showed widespread metastatic dissemination among the kidneys, spleen, stomach, and liver. In contrast, the three-drug combination group only showed small lesions of liver metastasis (Fig. 7E). The control group mice had noticeable ascites, which was 2.33- fold more than the treatment groups. Notably, the mice treated with the combination of three drugs only had less than 0.5 mL of ascites in total (Fig. 7F). The control group mice had a rapidly deteriorating body weight, whereas there was no significant change in body weight within the treatment groups, suggesting that all of the drugs were well tolerated in vivo (Fig. 7G). Histopathologic examination revealed no significant injury to mouse organs after a long therapy duration (Fig. S7C). The IHC results confirmed that the combination of three drugs significantly suppressed EMT (Fig. 7H). These results indicate that clinically applicable cisplatin synergizes with CAY10566 and sc26196 to suppress OvCa peritoneal metastasis in vivo.