Ferroptosis induction induced by ginkgetin enhances therapeutic effect of cisplatin in EGFR wild type non-small cell lung cancer

Background Cisplatin (DDP) is the rst-in-class drug for advanced and non-targetable non-small cell lung cancer (NSCLC). Recent study indicates that DDP could slightly induce non apoptotic cell death ferroptosis, and the cytotoxicity was promoted by ferroptosis inducer. The agents enhancing ferroptosis level therefore may increase anticancer effect of DDP. Several lines of evidence support the usage of phytochemicals in therapy of NSCLC. Ginkgetin, a bioavonoid derived from Ginkgo biloba leaves, showed anti-cancer effect on NSCLC both in vitro and in vivo, which could strongly trigger autophagy. Ferroptosis can be triggered by autophagy, and which regulates redox homeostasis. Thus, we aim to elucidate the possible role of ferroptosis induction in accounting for synergy of ginkgetin with DDP in cancer therapy. of SLC7A11, GPX4 and GSH/GSSG ratio. In parallel, ginkgetin disrupted redox hemeostasis in DDP-treated cells, demonstrated by the enhanced ROS formation and inactivation on Nrf2/HO-1 axis. Ginkgetin also enhanced DDP-induced mitochondrial membrane potential (MMP) loss and apoptosis in cultured NSCLC cells. Furthermore, blocking ferroptosis reversed the gingketin-induced inactivation on Nrf2/HO-1, as well as the elevation on ROS formation, MMP loss and apoptosis in DDP-treated NSCLC cells. Conclusion This study rstly reported that ginkgetin promoted DDP-induced anticancer effect, which could be accounted by induction of ferroptosis.


Background
Lung cancer is the leading cause of cancer-related death throughout the word. The global incidents of lung cancer ranked rst in 2018 among all types of cancer 1 . The mortality and incidents are climbing up quickly in recent years. In china, the mortality and incidents were much higher than other regions 2 , the mortality is expecting to increase by ~ 40% during 2015-2030 1 . Non-small lung cancer (NSCLC) is the most common type of lung cancer, which account for more than 80% of total lung cancers. For NSCLC treatment, immunotherapy usually have low response rate. While target therapy is highly dependent on oncogenic mutation, which accounts for small percentage of total NSCLC. Thus, cisplatin (DDP), a platinum-based chemotherapeutic drug, is still a standard treatment for non-targetable NSCLC, especially for EGFR wild type NSCLC patients, as well as the patients in advanced stage. Combined drug therapy is a promising strategy to treat NSCLC, especial for those classic chemo-therapeutic drugs. DDP usually combined with other chemo-drugs in treating NSCLC. However, patients with advanced stage of cancer, or poor overall health, might not tolerate the unexpected side effects, as induced by combination of chemodrugs 3 . Therefore, the focus has shifted to a combination of phytochemicals with DDP, as to increase the therapeutic effect or to eliminate side-effects. Extensive studies demonstrated that phytochemicals could enhance the sensitivity of DDP without overlapping toxicity 4 . The DDP resistance is resulted from increase of apoptosis resistance and redox homeostasis resetting: these are two key processes involving therapeutic e cacy of DDP. Thus, phytochemicals trigger non-apoptotic cell death or disrupt the redox homeostasis resetting could be effective in enhancing chemosensitivity of DDP, as well as in preventing emergence of resistance.
Ferroptosis is a mode of non-apoptotic cell death, which triggers cell death via iron-dependent lipid peroxidation 5 . Recent study demonstrated that ferroptosis is a novel anticancer action for DDP 6 . Erastin, a classic ferroptosis inducer, induces ferroptosis via system Xc − inhibition, and which has been shown to synergize with DDP in promoting cytotoxicity in different types of tumor, especially in NSCLC [6][7][8] . The drug resistance to DDP occurs via apoptotic evasion, and therefore ferroptosis is being considered as a new therapeutic way in promoting the e cacy of DDP.
The heart of ferroptosis is lipid peroxidation and iron accumulation. Ferroptosis is driven by lipid peroxidation, which typically triggers via suppression on solute carrier family 7 member 11 (SLC7A11) and glutathione peroxidase 4 (GPX4). SLC7A11 is a cystine-glutamate anti-porter. The reduction of SLC7A11 expression leads to cystine depletion, glutathione (GSH) shortage and consequent lipid peroxidation elevation 9 . GPX4, a key inhibitor on lipid-peroxidation, oxidizes GSH to glutathione disul de (GSSG). Repressing GPX4 compromised the capability in neutralizing lipid peroxidation via GSH 10 . In balancing iron, transferrin and Solute Carrier Family 40 Member 1 (SLC40A1) are crucial to maintain intracellular iron concentration. Transferrin imports iron into cells; while SLC40A1 exports iron from cells 11 . Thus, the increase on transferrin and decrease on SLC40A1 could cause intracellular iron accumulation. Cancer cells are usually adapting to iron and under persistent oxidative stress. To protect from ferroptosis-induced cell death, cancer cells can promote their antioxidant systems 5 . For instance, nuclear factor erythroid 2-related factor 2 (Nrf2), a master antioxidant regulatory transcription factor, prevents ferroptosis-induced cell death via upregulating antioxidant enzymes 12 . For instance, heme oxygenase-1 (HO-1), a key antioxidant enzyme, contains multiple number of antioxidant response element (ARE) on its promoter region 13 . In line to this notion, cancer cell having mutations in Nrf2 is able to increase transcription of antioxidant genes. The induction of ferroptosis through SLC7A11 and GPX4 could be mitigated via Nrf2/HO-1 activation via lipid oxidation elimination. Therefore, phytochemicals having suppression of Nrf2/HO-1 antioxidant system could be effective in promoting ferroptosis. Cells were treated with ginkgetin, DDP and ginkgetin + DDP. The apoptosis rates of untreated and treated cells were detected by Annexin V-FITC Apoptosis Detection Kit, as previously described 19 . Both oating and adherent cells were collected and wash 3 times by PBS. Cells were stained with Annexin V and PI for 15 min in dark, and detected apoptosis via ow cytometry with the acquisition criteria of 10,000 events for each sample.

Total ROS and MMP measurement
The measurement of total ROS and MMP were conducted as previously described 20 . For total ROS detection, 15 μM DCFH-DA was added in culture medium without FBS for 0.5 hour. After washing cells twice with PBS, cells were collected and analyzed by ow cytometry (Exc=488 nm, Em=530 ± 30). For MMP measurement, cells were stained with JC-1 for 30 min, then washed twice with PBS before analysis. Both JC-1 monomers (Exc=488 nm, Em=530 ± 30) and aggregates were detected (Exc=561 nm, Em=582 ± 25). Each sample met the acquisition criteria of 10,000 events, and the results were analyzed with Flowjo v7.6 software.
Lipid peroxidation measurement C11-BODIPY (10 μM) was added to drug treated and untreated cells for 0.5 hour, then cells were collected by trypsin. Oxidation of the polyunsaturated butadienyl portion of C11-BODIPY resulted in a shift of the uorescence emission peak from ∼590 nm to ∼510 nm. Cells were analyzed using ow cytometry (Exc=488 nm, Em=510) after washing twice with PBS, and the results were analyzed with Flowjo v7.6 software.

Measurement of labile iron pool (LIP)
Drug treated and untreated cells were collected and washed 2 times with PBS. Then, cells were loaded with CA-AM (0.25 μM) at the density of 0.5 × 10 6 /ml for 15 min. After washing twice with PBS, cells were incubated with iron chelator deferiprone (100 μM) for 1 hour or untreated. Measurement was conducted using uorescence microplate reader (Exc=488 nm, Em=525). The amount of LIP was re ected via difference on mean uorescence of each sample with or without deferiprone.

Western blot analysis
Western blot was conducted as described 20 . Brie y, cells were lysed, and their protein concentrations were measured using Bradford method. SDS-PAGE was used to separate the protein in each sample. Proteins were transferred from gel to membrane. Then, the membrane was blocked and incubated with indicated primary antibodies. The blots were rinsed before probed with secondary antibodies. The reactive bands were visualized by ECL and calibrated by Chemidoc Imaging System (Bio-Rad).

Immuno uorescence
Cultured A549 cells were seeded on coverslips. Ginkgetin, DDP and ginkgetin + DDP were added for 48 hours. Cells were xed with 4% formaldehyde after rinsing twice with PBS. Specimens were blocked (1X PBS/5% BSA/0.3% Triton™ X-100) for 1 hour. Then, primary antibody was incubated overnight at 4°C. Then, cells were rinsed 3 times, and probed with Alexa Fluor 555-conjugated goat anti-rabbit secondary antibody for 2 hours. After rinsing, coverslips were mounted to slide by Prolong® Gold Antifade Reagent with DAPI (CST, Danvers, MA). Images were taken by FV3000 Confocal Laser Scanning (Olympus, Japan). To analyse nuclear translocation of HO-1, the co-localization coe cients were calculated using Olympus Fluoview FV31S-DT Software.

Chromatin immunoprecipitation
Chromatin immunoprecipitation was performed using ChIP kit (Abcam, Cambridge, UK). In brief, the proteins were cross-linked to DNA by formaldehyde for 15 min at room temperature. Glycine was added to quench the formaldehyde at nal concentration of 125 mM. Cells were washed by ice-cold PBS for 3 times, then resuspended in lysis buffer. The cross-linked lysate was sonicated to shear DNA to an average fragment size of 200-800 bp, then centrifuged and transferred the supernatant for immunoprecipitation.
The sonicated chromatin (100 μg) was incubated with anti-Nrf2 antibody, or IgG, or H3 antibody overnight at 4 °C with rotation. DNA puri cation was carried out according to the manufacturer's instructions. Then, HO-1 DNA was ampli ed for 45 cycles of PCR with the following primers: 5′-TCA ATA GGC GAT CAG CAA GGG -3′ (S) and 5′-TGG AAT GCG TGG GAC ACT C -3′ (AS).

Luciferase assay
Luciferase assay was conducted as previously described 20 . In brief, drug treated and untreated cells were washed and lysed. The supernatant was collected and then analysed using a commercial kit (Thermo

Results
Ginkgetin promotes DDP-induced cytotoxicity Ginkgetin, a bi avonoid from Ginkgo biloba leaves, exhibited cytotoxicity in cultured A549 cells 15 . Here, ginkgetin was combined with DDP at various concentrations in applying onto cultured A549 cells. Majority of combinations (ginkgetin + DDP) signi cantly increased the cytotoxicity, as compared with single usage of DDP or ginkgetin (Fig. 1A). The combination index indicated that ginkgetin at 5 µM, together with different concentrations of DDP, showed the lower CI value, suggesting a better synergy (Fig. 1B). The synergistic effect of GK at 5 µM with DDP were also demonstrated in NCI-H460 and SPC-A-1 NSCLC cells (Fig. S1A&B). All these three NSCLC lines are EGFR wild type, which are not sensitive to target therapy, but more suitable for DDP treatment. Among these combinations, the mixture of ginkgetin and DDP (both at 5 µM) showed the relative lower CI value, and the lowest CI value at 0.5005 was observed in A549 cell line (Fig. 1B&C, S1B). To reveal pharmacodynamic interaction of ginkgetin and DDP on A549, a response surface tting was constructed. As expected, the combination of ginkgetin and DDP, both at 5 µM each, was located on the highest region of dose-response surface, further con rmed that this combination having better anti-cancer function (Fig. 1D). This combination ratio was chosen for further mechanistic study.

Ginkgetin induces ferroptosis in DDP-treated cells
Previous study demonstrated ginkgetin induces autophagic cell death 15 , However, this phenomenon was not further promoted in ginkgetin + DDP treated cells. As ferroptosis is considered to be a consequent event after autophagy in recent year 17,18 . Thus, we hypothesis ferroptosis might be triggered in this combination. The two key characteristics of ferroptosis are lipid peroxidation and intracellular-free iron 21 .
C11-BODIPY 581/591 could be used as a lipid peroxidation probe in mammalian cells 22 . Free iron levels could be measured via labile iron pool (LIP), stained by CA-AM 17 . Thus, C11-BODIPY 581/591 and CA-AM were employed here to observe lipid peroxidation and LIP, respectively. DDP did not alter the levels of lipid peroxidation and LIP in A549 cells; while ginkgetin signi cantly increased the levels to~2.6 and ~2.4 folds respectively (Fig. 2A&B). The combination of ginkgetin + DDP further increased the levels of lipid peroxidation and LIP to ~3.3 and ~7.1 folds, respectively (Fig. 2A&B). The enhancement of LIP was more robust than that of lipid peroxidation in ginkgetin + DDP-treated cells. Ginkgetin induced promotion on lipid peroxidation and LIP were also observed in DDP-treated NCI-H460 and SPC-A-1 cells (Fig. S2A&B).
SLC7A11 and GPX4 are main targets for ferroptosis induction. Thus, we revealed expressions of SLC7A11 and GPX4 at transcriptional and post-transcriptional level after the treatment of combined drugs. There were no signi cant changes on SLC7A11 and GPX4 mRNAs after the combined drug treatment in cultured A549 cells (Fig. 2C). In contrast to mRNA level, the protein amounts of SLC7A11 and GPX4 were markedly decreased in application of ginkgetin + DDP in A549, NCI-H460 and SPC-A-1 cells (Fig. 2D, S2C). These phenomena suggested the role of ginkgetin in increasing protein degradation of SLC7A11 and GPX4 in DDP-treated NSCLC cells. To further demonstrate the ferroptosis induction, we determined another key factor involving in iron accumulation during ferroptosis, i.e. SLC40A1 and transferrin. SLC40A1 is the sole iron exporter in mammalian cells, as well as a downstream target of Nrf2 23 ; while transferrin imports iron to cell 24 . In the cultures, DDP sharply increased the mRNA and protein levels of SLC40A1 (Fig. 2E&D). Gingketin alone did not change the mRNA level of SLC40A1; however, which could reverse the DDP-induced elevation on mRNA (Fig. 2E), as well as protein level (Fig.  2D). For the case of transferrin, DDP slightly increased the protein amount in A549 cells (Fig. 2D), while have no obvious change on NCI-H460 and SPC-A-1 cells (Fig. S2C). Ginkgetin combine with DDP sharply increased transferrin expression in all these three NSCLC cells (Fig. 2D, S2C). The decreased SLC40A1 and increased transferrin might account for LIP elevation in combination treatment.
The inhibition on SLC7A11 triggers ferroptosis via cystine/glutamate transport. The reduction of SLC7A11 is expected to decrease glutamate release and cystine uptake. In accordance with this notion, we measured the levels of cystine and glutamate in drug treated A549 cells. The intracellular cystine was signi cantly decreased, accompanied with notably increased glutamate level, after the treatment of ginkgetin (Fig. 2F). No signi cant change of glutamate was observed in DDP-treated cells; while the cystine level was notably increased, which might contribute to the increased antioxidant activity. The combined treatment sharply reversed DDP-induced elevation on cystine, and signi cantly increased glutamate level, as compared to control (Fig. 2F). These results indicated that the anti-porter function of SLC7A11 was partially reversed after application of ginkgetin in DDP-treated A549 cells.
GSH is synthesized from cystine and eliminates lipid ROS via GPX4. As the decline of cystine and GPX4 were observed here after the combined drug treatment, thus the intracellular GSH level was determined. As expected, GSH amount was signi cantly decreased by ginkgetin. However, the GSH amount was increased after application of DDP, which might be due to the redox resetting via antioxidant system. The combined drug treatment sharply reversed DDP-induced increase on GSH (Fig. 2G). GSH is highly reactive with lipid ROS to generate glutathione disul de (GSSG), and the reduced ratio of GSH/GSSG is considered to be a marker of oxidative stress. Ginkgetin + DDP application sharply decreased the ratio of GSH/GSSG (Fig. 2H), indicating elevation of oxidative stress. All these data illustrated above indicated that ferroptosis was being triggered in the drug combination.
Ginkgetin downregulates Nrf2/HO-1 axis in DDP-treated NSCLC cells Ferroptosis could be downregulated by famous antioxidant system Nrf2/HO-1 via neutralized on oxidative stress, which responsible for the compromised anticancer function of DDP 25,26 . Our previous study has demonstrated ginkgetin could reduce Nrf2 activation, thus we hypothesis that it could downregulate elevated activity on Nrf2/HO-1 axis induced by DDP. Here, neither ginkgetin nor DDP could change the expression of Nrf2, which however was sharply reduced in treatment of ginkgetin + DDP in A549, NCI-H460 and SPC-A-1 cells (Fig. 3A, S3A&B). DDP slightly increased the expression of HO-1 in A549 cells (Fig. 3A), there are no signi cant change in NCI-H460 and SPC-A-1 cells (Fig. S3A&B); while ginkgetin robustly deceased the amount of HO-1 in all these three NSCLC cells (Fig. 3A, S3A&B). Activated Nrf2 binds to ARE and upregulates transcription of HO-1. To detect the effect of ginkgetin + DDP on AREmediated transcriptional activity, a luciferase reporter pARE-Luc was applied. This construct contained four repeats of antioxidant response element (ARE) and a luciferase reporter gene luc2P. In pARE-Lucexpressed A549 cells, DDP activated ARE-mediated transcription by ~3 folds; while ginkgetin did not show activation on ARE-mediated transcription (Fig. 3B). Application of ginkgetin in DDP-treated A549 cells largely reversed DDP-induced activation on ARE-mediated transcription, i.e. counter acting the induction by DDP (Fig. 3B). CHIP assay was applied to identify the binding of Nrf2 to HO-1 promoter. DDP sharply increased the binding of Nrf2 to HO-1 promotor by over 40 folds (Fig. 3C). Ginkgetin alone showed no signi cant change on this binding. As expected, ginkgetin sharply reduced DDP-induced elevation on the binding of Nrf2 to HO-1 promoter (Fig. 3C). Binding of Nrf2 to HO-1 promoter is leading to transcription of HO-1. In cultured A549 cells, DDP increased the mRNA expression of HO-1 (Fig. 3D).
The application of ginkgetin signi cantly reversed DDP-induced upregulation of HO-1 mRNA expression (Fig. 3D). In consistent, this mRNA regulation was in line to protein level, ginkgetin reversed the DDPinduced HO-1 protein expression (Fig. 3A). These results indicated that DDP could promote the antioxidant system Nrf2/HO-1 to cope with ferroptosis induced oxidative stress, which could be reversed by ginkgetin.
The antioxidant activity, induced by Nrf2, is further enhanced by nuclear translocation of HO-1 27 . Thus, the change on HO-1 nuclear translocation in ginkgetin, DDP and ginkgetin + DDP-treated A549 cells was revealed by immunostaining. The uorescence intensity of HO-1 was observed both in cytosol and nucleus. In control group, the uorescence was mainly located in cytosol, and a faint signal was observed in nucleus, as demonstrated by co-localization of DAPI signal. Application of DDP notably increased the HO-1 uorescence intensity in nucleus; however, ginkgetin decreased signi cantly nuclear expression of HO-1 and sharply reversed DDP-induced HO-1 nuclear translocation (Fig. 3E&F). This result further con rmed that ginkgetin could reverse DDP-induced activation on Nrf2/HO-1 axis, which contribute to the mitigation on antioxidant effect in ginkgetin + DDP treated NSCLC cells.

Ferroptosis inhibition reversed ginkgetin induced promotion on cytotoxicity of DDP
To further observe the role of ferroptosis in ginkgetin + DDP induced anticancer function. Ferroptosis inhibitors UAMC 3203 and DFO were applied. Here, both UAMC 3203 and DFO markedly reversed ginkgetin + DDP induced cytotoxicity in cultured A549, NCI-H460 and SPC-A-1 NSCLC cells (Fig. 4A,  S4A&B). However, the Nrf2 activators DMF and SFN could not reverse ginkgetin + DDP induced cytotoxicity in all three NSCLC cells (Fig. 4B, S4A&B). The upregulation of DMF and SFN on Nrf2 was demonstrated by western blot, that the amount of Nrf2 were signi cantly increased by DMF and SFN in ginkgetin + DDP treated A549 cells (Fig.S4C). These phenomena indicated that the Nrf2 is not the nodal for ginkgetin + DDP induced cytotoxicity.
UAMC 3203 is a novel ferroptosis inhibitor. The reverse effect of UAMC 3203 in ginkgetin + DDP induced cytotoxicity was much obvious than DFO (Fig. 4A, S4A&B), which might be due to its better activity on ferroptosis suppression. Thus, we use UAMC 3203 for further observation. To con rm the ferroptosis suppression, the key markers lipid peroxidation, LIP, SLC7A11 and GPX4, were determined in A549 cultures. The application of UAMC 3203 moderately reversed ginkgetin-induced elevation on lipid peroxidation (Fig. 4C&D) and LIP (Fig. 4F), and this effect was much obvious in the scenario of ginkgetin + DDP (Fig. 4C, D&F). In parallel, UAMC 3203 reversed gingketin + DDP mediated decline of SLC7A11; while the reverse effect on GPX4 was identi ed in cultures being treated with gingketin or gingketin + DDP (Fig. 4E). These results indicated that ferroptosis, induced by ginkgetin + DDP, was blocked by UAMC 3203.
Considering ferroptosis induction could directly or indirectly downregulated GPX4, leading lipid peroxidation. To con rm the role of ferroptosis in ginkgetin + DDP induced cytotoxicity. We overexpressed GPX4 in A549 cells, the upregulated expression of GPX4 was con rmed by western blot in ginkgetin + DDP treated cells (Fig. 4G). To our respective, the cytotoxicity was notably decreased after GPX4 overexpression (Fig. 4H), concomitant with the downregulation on lipid peroxidation (Fig. 4I), and LIP ( Fig. 4J). This result further elucidated that ferroptosis contributes to ginkgetin + DDP induced cytotoxicity.

Ferroptosis suppression mitigated attenuation on Nrf2/HO-1 activation and ROS promotion induced by
ginkgetin Redox homeostasis is governed by the balance of antioxidant system and ROS formation. The downregulation on antioxidant system Nrf2/HO-1 induced by ginkgetin in DDP treated NSCLC has driven us to found if ROS was further increased to disrupt the redox homeostasis. In cultured NSCLC cells, DDPinduced ROS formation, and which was sharply promoted by ginkgetin in A549, NCI-H460 and SPC-A-1 cells (Fig. 5A&B, S5A). However, blocking ROS formation via N-acetylcysteine failed to reverse ginkgetin + DDP induced cytotoxicity (Fig. S5B).
Since ferroptosis inhibitors, not Nrf2 activators, could largely reversed ginkgetin + DDP induced cytotoxicity. In addition, the unchanged mRNA level of SCL7A11 and GPX4 partial indicated that these two ferroptosis genes were not transcriptionally regulated by Nrf2. Thus, we hypothesis that Nrf2/HO-1 antioxidant inactivation and ROS enhancement could be a consequent event of ferroptosis. Here, application of UAMC 3203 sharply rescued gingketin + DDP induced decline of Nrf2 (Fig. 5C), as well as ARE-mediated transcription activity (Fig. 5D). In addition, UAMC 3203 application reversed the ginkgetin + DDP suppressed expressions of mRNA and protein of HO-1 (Fig. 5G&C). Consistent with this, ginkgetin + DDP induced ROS increasement was sharply reversed by the application of UAMC 3203 in A549, NCI-H460 and SPC-A-1 cells (Fig. 5H, S5C). These results indicated that ferroptosis suppression mitigated attenuation on Nrf2/HO-1 activation and promotion on ROS formation induced by ginkgetin in DDP treated NSCLC cells.

ROS elevation induced by ginkgetin in DDP-treated cells might result in increasing cell sensitivity to ROS.
One consequent event of ROS elevation is the loss of mitochondria membrane potential (MMP). Here, we demonstrated that ginkgetin notably increased the MMP loss in DDP-treated A549 cells (Fig. 5E&F). MMP loss could lead to activation on caspase-9, consequently activate caspase-3, -7, leading apoptosis. As apoptosis is the key mechanism for DDP-induced anticancer effect. Thus, we observed if apoptosis was increased after ferroptosis induction. Here, we found that DDP at 5 µM slightly increased the apoptosis rate at ~15%: while ginkgetin at 5 µM induced apoptosis at ~30% (Fig. 6A&B). The combined DDP and ginkgetin sharply increased the apoptosis rate to over 50%, which was con rmed by increased apoptotic markers, i.e. cleaved-caspase 3, cleaved-caspase 7 and cleaved caspase 9, as revealed by western blotting (Fig. 6C&D).
While ferroptosis suppression signi cantly reversed apoptosis in ginkgetin+DDP treated cells (Fig. 6F). Consistent with this, ferroptosis suppression attenuated ginkgetin + DDP induced MMP loss (Fig. 5I), characterized with the sharply increase on the mean FITC uorescence and decline on PerCP-Cy5-5 uorescence (Fig. 5J), which indicated the reduced MMP loss. These results indicated that ferroptosis might contribute to ginkgetin induced promotion on DDP triggered apoptosis.
Ginkgetin enhanced anticancer effect of DDP is compromised by ferroptosis suppression in xenograft nude mice model To further con rm the ginkgetin induced promotion on anticancer function of DDP, we applied A549 xenograft nude mice model. After the treatment of 31 days, the mean body weight of DDP treated group signi cantly declined; while ginkgetin group showed no signi cant change, as compared with control mice (Fig. 7A). Combined administration of ginkgetin + DDP signi cantly increased the mean body weight since day 25, as compared with DDP group (Fig. 7A), which might indicate that ginkgetin treatment could relieve DDP-induced toxicity. The mean tumor volumes in DDP, ginkgetin, and ginkgetin + DDP group were decreased: the best reduction was revealed in the combined administration group (Fig.   7B). When combined with UAMC 3203 administration, the mean tumor volume was not statistically signi cantly changed in control group, as well as in DDP group. Ginkgetin group showed moderately increase on mean tumor volume in the presence of UAMC 3203. However, a notably increase was identi ed in ginkgetin + DDP group after UAMC 3203 treatment (Fig. 7B). Consistent with the change on tumor volume, the mean tumor weight was smallest in ginkgetin + DDP group (Fig. 7C&D). UAMC treatment statistically signi cantly reversed the tumor shrink in ginkgetin group, however, the reversed effect was more robustly in ginkgetin + DDP group (Fig. 7C&D).These results consistent with in vitro study in supporting the notion that ginkgetin induced promotion on anticancer effect of DDP in NSCLC could be mediated by ferroptosis.

Discussion
DDP combines with phytochemicals is a promising strategy to enhance its anticancer effect in NSCLC. In this study, we revealed the synergistic effect of gingketin with DDP on cytotoxicity in EGFR wild type NSCLC. The synergy of combined drugs was further con rmed in animal study. Apoptosis resistance and redox resetting are key factors in accounting for failure of DDP therapy, and therefore the strategy in triggering non-apoptotic cell death and disruption on redox balance could be promising methods to enhance anticancer function of DDP. Here, ferroptosis, a non-apoptotic cell death, was robustly triggered by gingketin application in DDP treated EGFR wild type NSCLC cells, concomitant by inactivation on Nrf2/HO-1 axis and promotion on total ROS formation. In addition, DDP-induced MMP loss and apoptosis were robustly ampli ed under ginkgetin application. Furthermore, the suppression on ferroptosis could diminish the synergy of gingketin + DDP, as well as reverse the inactivation on Nrf2/HO-1, ROS enhancement, MMP loss and apoptosis. This is the rst time to demonstrate that ferroptosis could account for increased therapeutic effect of DDP induced by ginkgetin both in vitro and in vivo systems.
Ferroptosis is dependent on iron and characterized by lipid peroxidation, which could ultimately cause oxidative cell death 28 . Recent report has suggested that autophagy could trigger ferroptosis, and some of them believed that ferroptosis is a part of autophagy process 5 . In line to this notion, ginkgetin, which robustly triggers autophagy, could has potential effect on ferroptosis induction.
Ferroptosis induced by direct or directly suppression on GPX4 to decline GSH-medicated antioxidant activity, contributing to elevation on lipid peroxidation. GPX4 is a selenoenzyme, catalyzing GSH to GSSG, as to neutralize production of lipid peroxidation. As GPX4 is the most functional selenoenzymes in reducing esteri ed lipid hydroperoxide, its reduction notably increases lipid peroxidation 29 . SLC7A11 is a cystine/glutamate antiporter. Suppression on SLC7A11 could induce autophagy via activation on lysosomal-associated membrane protein 2a, which in turn degrades GPX4 30 . In addition, SLC7A11 inactivation leads to reduced intracellular concentration of cystine and increased the glutamate level.
Cystine is an essential substrate for GSH synthesis; while the reduction of cystine leads to decline of GSH. This could disrupt the equilibrium of antioxidant system and thereafter enhance lipid peroxidation 31,32 . Consistent with this theory, expression of SLC7A11 and GPX4 expression, GSH level, and GSH/GSSG ratio, were sharply decreased, which might contribute to robustly increase on lipid peroxidation in combined drug treatment. Furthermore, SLC40A1 and transferrin contribute to export and import of iron, respectively. The decreased SLC40A1 and increased transferrin were observed here, which contributed to elevation of LIP (Fig. 8). Thus, the phenomena illustrated here indicated that ferroptosis was triggered by gingketin + DDP application.
The crucial role of ferroptosis on anticancer effect of ginkgetin + DDP combination was further demonstrated by the ferroptosis inhibition. Blocking ferroptosis signi cantly reverse ginkgetin + DDP induced anticancer effect both in vitro and in vivo. Intriguingly, the ferroptosis key factors SLC7A11, GPX4, and SLC40A1 are all transcriptionally regulated by Nrf2. However, only the mRNA expression of SLC40A1 was signi cantly changed. In addition, ferroptosis blocking could largely reverse Ginkgetin + DDP induced decrease on Nrf2. The inhibition on Nrf2 could not signi cantly rescued ginkgetin + DDP induced cytotoxicity. These phenomena might be due to the promoted degradation on SLC7A11 and GPX4, which evoked ferroptosis, consequently downregulated Nrf2 mediated suppression on SLC40A1, nally increase the transcriptional and post-transcriptional level of SLC40A1. Our previously study has demonstrated that ginkgetin induces autophagy via p62. The decline of p62 could downregulate Nrf2, we do nd the decline on p62 in drug combination (Fig. S6), however, the detailed mechanism needs to be further studied.
Nrf2 is a master regular on antioxidant and detoxi cation, which is responsible for ferroptosis resistance via binding to ARE of its downstream genes 33 . Among these, HO-1 has most abundant sites for ARE on its promoter region 13  contribute to redox homeostasis disruption, nally enhanced DDP-induced cell death (Fig. 8).

Conclusions
In short, ferroptosis contributes to the increased cytotoxicity induced by ginkgetin in DDP-treated EGFR wild type NSCLC. Several studies reported the bene ts for combination of ferroptosis and apoptosis inducers on cancer treatment. This phenomenon was con rmed by our study that ginkgetin-induced ferroptosis and DDP-induced apoptosis were both ampli ed under the drug combination.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. 35Supplementarydata.docx