Aurora-A inhibits hepatocellular carcinoma cell ferroptosis to mediate immune escape by disrupting phosphatidylethanolamine biosynthesis

Abstract

This study aims to explore Aurora-A’s role in regulating immune escape of hepatocellular carcinoma (HCC). We performed non-targeted metabolomics analysis and analyzed the impact of Aurora-A inhibitor Alisertib on anti-PD-1 therapy efficacy on xenograft tumors and co-culture models of CD8⁺ T cells and HCC cells. We determined reactive oxygen species (ROS) and malondialdehyde (MDA) production in HCC cells to evaluate lipid peroxidation. Confocal images of endoplasmic reticulum (ER) and mitochondria in HCC cells were taken to assess the role of Aurora-A and dynamin-related protein 1 (Drp-1) on mitochondria-associated endoplasmic reticulum membranes (MAMs) formation. The results showed that Aurora-A was upregulated in HCC cells and its knockdown significantly augmented phosphatidylethanolamine (PE) production while having no effect on phosphatidylserine decarboxylase (PSD). Further, Aurora-A inhibitor Alisertib enhanced the sensibility of HCC cells to anti-PD-1 therapy and CD45⁺CD8⁺ T cell infiltration in HCC tumors. To conclude, our work revealed that Aurora-A dysregulated PS/PE metabolism via facilitating Drp1-Ser616 phosphorylation to disrupt MAMs formation, resulting in suppressed ferroptosis in HCC cells to reduce their sensitivity to anti-PD-1.

Introduction

Globally, hepatocellular carcinoma (HCC) is one of the most frequent cancer types and the third leading cause of cancer-related death [1]. Generally, surgical resection is the first choice for HCC patients, but most of them are not suitable for surgery. Non-surgical therapies, including radiotherapy, chemotherapy, immunotherapy, transarterial chemoembolization, local ablation therapy, have been applied to relieve patients’ symptoms, but the overall prognosis is still not satisfactory [2]. Programmed cell death protein 1 (PD-1) is an immune checkpoint molecule that is highly related to tumor immune evasion [3]. Its inhibitors have been approved for HCC treatment [4]. However, not all HCC cases are sensitive to anti-PD-1 therapy. Hence, new and more biomarkers remain to be identified in HCC.

Aurora-A is a serine/threonine protein kinase belonging to the Aurora kinase family [5]. Aurora-A, also called STK15, has been proven to be highly expressed in HCC and be relevant to the unfavorable prognosis of HCC patients [6]. Importantly, previous reports have revealed that Aurora-A plays a pivotal role in regulating HCC cell functions and cell resistance to different therapies [7-9]. Besides, Aurora-A inhibitors have also been reported to be effective in enhancing chemosensitivities of HCC cells [10,11]. Nevertheless, the influences of Aurora-A and its inhibitors on immune therapies like anti-PD-1 remain unknown.

Metabolic reprogramming is a hallmark of tumors. The metabolic characteristics of tumors take changes during cancer development [12]. Overlapping metabolic reprogramming of tumor and immune cells is thought to be a determinant of anti-tumor immune response in malignancies [13]. Recently, Aurora-A has been indicated to regulate glycolysis in several different types of cancer cells and phospholipid remodeling in glioblastoma [14,15]. Based on these reports, we hypothesized that Aurora-A and its inhibitors might affect HCC cell metabolism to modulate immune therapy efficacy.

In this study, we aim to testify above hypothesis and investigate the underlying mechanism through which Aurora-A regulates HCC cell metabolism. According to investigations, Aurora-A dysregulates phosphatidylserine/phosphatidylethanolamine (PS/PE) metabolism through promoting the phosphorylation of dynamin-related protein 1 (Drp1) at Ser616 to impede mitochondria-associated endoplasmic reticulum membranes (MAMs) formation, finally leading to hindered ferroptosis in HCC cells to reduce sensitivity of anti-PD-1. The findings suggest Aurora-A inhibitors as a new possible target for HCC treatment.

Materials and methods

Cell culture

The healthy liver-derived human cell line (THLE-2) and HCC cell lines (LM3, Huh7, HepG2, and Hep3B) were purchased from the American Type Culture Collection (ATCC. Manassas, VA, USA). Cells were all grown in DMEM (Invitrogen Life Technologies, Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS, Invitrogen) and 1% penicillin-streptomycin (Invitrogen). The cell culture plates were placed in a humidified incubator with 5% CO₂ at 37°C. The media were changed every 2-3 days.

Cell transfection

For AURKA knockdown, cells were transfected with three different siRNAs targeting AURKA (si-AURKA#1/2/3) or si-NC (negative control) using Lipofectamine™ 3000 Reagent (Lipo3000, Invitrogen, California, USA). pcDNA3.1 vectors (Invitrogen) were applied to overexpress AURKA in indicated HCC cells. After transfection for 48 h, the transfection efficiencies were evaluated via RT-qPCR.

RNA isolation and reverse transcription-real time PCR (RT-qPCR)

Total RNA was isolated using Trizol reagent (TaKaLa, Dalian, China) and then reversely transcribed into cDNA via the Reverse Transcription Kit (Takara, Dalian, China). Real-time PCR analysis was conducted with SYBR Premix Ex Taq (Takara, Dalian China). GAPDH was used as the endogenous control. The RT-qPCR was performed on an ABI 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA). Relative gene expression was evaluated with the 2^-ΔΔCt method.

Non-targeted metabolite analysis

After samples were extracted according to a published protocol by Shanghai Lu-Ming Biotech Co. Ltd. (Shanghai, China, 2021), non-targeted metabolite analysis was implemented via OE Biotech (Shanghai, China). To analyze the metabolic profiling, a Dionex Ultimate 3000 RS UHPLC system fitted with a Q-Exactive quadrupole-Orbitrap mass spectrometer and equipped with heated electrospray ionization (ESI) source (Thermo Fisher Scientific, Waltham, MA, USA) was applied. The acquired LC-MS raw datasets were analyzed through progenesis QI software (Waters Corporation, Milford, USA).

Measurement of PS, PE and phosphatidylserine decarboxylase (PSD)

The contents of PS, PE, and PSD were determined as previously described [16].

Western blotting

Protein lysates were obtained via RIPA buffer containing proteinase inhibitors (R0278, Sigma-Aldrich St. Louis, MO, USA). Proteins were isolated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transformed into nitrocellulose (NC) membrane. Next, the membranes were blocked in 5% non-fat milk, followed by incubation with primary antibodies, including anti-NCOA4 (1:1000 dilution, Cambridge, ab314553, Abcam, CA, USA), anti-ACSL4 (1:10000 dilution, ab155282, Abcam), anti-GLS2 (1:1000 dilution, ab308304, Abcam), anti-Aurora-A (1:1000 dilution, ab61114, Abcam), anti-Drp1 (1:1000 dilution, ab184247, Abcam), anti-DRP1-Ser616 (1:1000 dilution, ab314755, Abcam), anti-Drp1-Ser637 (1:1000 dilution, ab193216, Abcam), and the loading control anti-β-actin (1:5000 dilution, ab179467, Abcam) at 4°C overnight. After washing, the membranes were treated with the HRP-conjugated secondary antibody (1:2000 dilution, ab288151, Abcam) for 2 hours. The protein bands were visualized through the ECL detection system (32134, Pierce Biotechnology, Rockford, IL, USA) and then quantitated via Quantity One software (Bio-Rad, Hercules, CA, USA).

In vivo experiments

Male C57B/L6 mice (8 weeks, 18±2 g) were brought from The Jackson Laboratory. All mice were reared under a specific pathogen-free environment with 50-60% humidity and a temperature of 22°C. All mice were exposed to 12-hour light/dark cycle and had free access to food and water. All mice were allowed 3-7 days for adaptation after arrival before experiments. The xenograft model was established via subcutaneous injection of indicated HCC cells into the left backs of mice. Tumor volumes were measured every 3 days based on the formula: V = 0.5 × D × d2 (V: volume; D: the longitudinal diameter; d: the latitudinal diameter). After the tumor volume exceeds 50 mm³, mice were treated with Aurora-A inhibitor Alisertib or anti-PD-1 which is described in Figure 2A. Four weeks later, the tumors were obtained for subsequent analyses. Humane endpoints were strictly enforced when tumor-bearing mice exhibit either ≥ 20% body weight loss or inability to access food/water independently, requiring prompt euthanasia. To humanely euthanize C57BL/6 mice, CO₂ inhalation was used at a controlled flow rate of 28% chamber volume per minute, ensuring minimal distress and rapid unconsciousness.

Figure 2.

Combination treatment of Alisertib elevates anti-PD-1 therapy efficacy for HCC tumors. A. The injection project of mice loading xenograft HCC tumors. B. Images of tumors isolated from indicated groups of mice. C, D. Tumor growth curves and tumor weights were presented. E. Changes in the infiltration of CD45⁺CD8⁺ T cells in xenograft tumors were analyzed by flow cytometry. F. IHC analyses of Ki67 staining in xenograft tumors. Scale bar = 200 μm. G, H. The iron and GSH contents in the lysates of each group of tumors were evaluated. I, J. The expression levels of ferroptosis-related genes (including NCOA4, ACSL4 and GLS2) and AURKA were detected in four groups of tumors via RT-qPCR and western blotting. K. PE concentrations in the lysates of four groups of xenograft tumors were measured. *P < 0.05, **P < 0.01.

Flow cytometry analysis of T cells

The live T cells were selected by staining with Fixable Viability Dye eFluor 450 (eBioscience) for 15 mins at 4°C and then incubated with primary antibodies against CD45 and CD8 (BioLegend). The stained T cells were captured by the BD FACSCanto II Flow Cytometer using BD FACSDiva software (BD Biosciences), and the generated data were processed with FlowJo software.

Immunohistochemistry (IHC) assay

The Ki67 staining in xenograft tumors was assessed via IHC. In brief, tumors after being fixed by formalin and embedded via paraffin were cut into sections. Then, the sections were treated with primary Ki67 antibodies for half an hour, followed by incubation with secondary antibody (DakoCytomation, Denmark) for 30 min. After being counterstained with hematoxylin, sections were processed by 3, 30-diaminobenzidine chromogen to visualize the signals.

Measurement of reactive oxygen species (ROS), malondialdehyde (MDA), glutathione (GSH) and iron contents

The lysates from HCC cells and xenograft tumors were attained by RIPA Lysis Buffer (Beyotime, China). Flow cytometry was applied for ROS analysis. MDA content was tested via the Lipid Peroxidation MDA Assay Kit (Beyotime, China). GSH level was determined with The Glutathione Assay Kit (Cayman Chemical). The iron contents in the lysates of xenograft tumors were estimated by using a Tissue Iron Assay Kit (Nanjing Jincheng, China).

Confocal fluorescence microscopy

Confocal fluorescence microscopy was used to measure the co-localization of ER and mitochondria as previously described [17]. Besides, the co-localization of Drp1 and mitochondria was also observed by confocal fluorescence microscopy, with the mitochondria stained by MitoTracker Red CMXRos (Beyotime Biotechnology) and Drp1 recognized by GFP-labeled specific antibodies. Also, the co-localization of PE and CD8 in T cells was observed via confocal fluorescence microscopy. In this experiment, PE-labeled by duramycin-LC-biotin (Molecular Targeting Technologies) and CD8 antibodies were applied. The nuclei were stained using DAPI.

Co-culture system of HCC cells and T cells

The transwell inserts were used for the co-culture model. Before co-culture, LM3 cells were treated with Alisertib or DMSO for 48 h. The media were replaced, together with a transwell (0.4 μm) inserted into a 6-well plate to create two spaces. In the co-culture model, LM3 cells were cultured in a 6-well plate while CD8⁺ T cells were grown on the surface of the transwell insert.

Statistical analysis

Data obtained from independent experiments repeated at least three times were expressed as mean ± SD. Differences between two groups were assessed via the student’s t-test, whereas those among more than two groups were evaluated by using one-way ANOVA with Tukey post-hoc test. The statistical analyses were conducted by using SPSS v17.0 software. P < 0.05 indicated that the difference was statistically significant.

Results

Aurora-A modulates PS/PE metabolism in HCC cells

We first tested the expression of AURKA gene encoding Aurora kinase A and found that it was highly expressed in four HCC cells (HepG2, LM3, Huh7, and Hep3B) when compared to the healthy liver-derived human cell line THLE-2 (Figure 1A). Next, we separately knocked down and overexpressed the AURKA expression to further analyze its function in HCC (Figure 1B). Since Aurora-A has been revealed to be related to metabolic reprogramming-modulated cancer development [14], we analyzed its impact on HCC metabolism. Hence, we performed non-targeted metabolomics analysis after knocking down AURKA in LM3 cells. The results indicated that 54 metabolites were significantly up-regulated and 4 metabolites were down-regulated after AURKA knockdown, among which the most evidently up-regulated metabolite was PE (Figure S1). The KEGG pathway enrichment results uncovered that repressing Aurora-A expression had the significant influence on the glycerol phospholipid metabolic pathway (Figure S2). Importantly, knockdown of Aurora-A induced the most evident changes in the PE metabolite of this pathway (Figure S3). Based on these data, we further explored the impact of Aurora-A on PE metabolism in HCC cells. As expected, PE levels in HCC cells were enhanced after AURKA knockdown, while reduced by AURKA overexpression (Figure 1C, 1D). PE is transformed from PS, and its synthesis is exclusively mediated by PSD (Figure 1E). Moreover, PS levels were increased in AURKA-downregulated HCC cells (Figure 1F). However, no changes were observed in PSD levels after AURKA knockdown (Figure 1G). In conclusion, Aurora-A prevents PE production in HCC cells by modulating PS/PE metabolism.

Figure 1.

AURKA is overexpressed in HCC cells and suppresses PE production. A. AURKA mRNA expression in four HCC cells and the healthy THLE-2 cells were measured by RT-qPCR. B. The knockdown and overexpression efficiencies of AURKA in HCC cells was analyzed by RT-qPCR. C, D. PE contents were estimated in HCC cells under AURKA knockdown or overexpression. E. The mechanistic graph of PE biosynthesis pathway. F, G. The changes in PS and PSD contents were evaluated in HCC cells after AURKA knockdown. *P < 0.05, **P < 0.01.

Aurora-A inhibitors promote ferroptosis to reinforce the efficacy of anti-PD-1 therapy for HCC

It has been reported that PE is one of the responsible factors for ferroptosis [18], which is highly related to cancer immunotherapy [19]. Besides, Aurora-A has been unveiled as a ferroptosis modulator in cancer [20]. Hence, we investigated that whether Aurora-A could affect ferroptosis, regulate HCC development and immune therapy by influencing PE metabolism. At first, we analyzed the impact of Aurora-A inhibitor Alisertib on xenograft HCC tumors treated with or without anti-PD-1 therapy. The details of treatments on xenograft tumors were shown in Figure 2A. Data showed that tumors were distinctly lessened after treatment with Alisertib or anti-PD-1 alone, but further suppressed when both Alisertib and anti-PD-1 were given (Figure 2B-D). Besides, flow cytometry analysis revealed that compared to the control group, the accumulation of CD45⁺CD8⁺ T cells was enhanced in the groups treated with Alisertib or anti-PD-1, and this trend was more significant in the group treated with both Alisertib and anti-PD-1 (Figure 2E). Meanwhile, IHC data indicated that the positivity of Ki67 was lower in the groups treated with Alisertib or anti-PD-1 than that in control group, and it was further reduced in the group treated with both Alisertib and anti-PD-1 (Figure 2F). These findings suggested that Aurora-A inhibition suppressed HCC tumor development and enhanced the efficacy of immune therapy. Next, we tested the changes in ferroptosis-related factors and PE content in these groups of tumors. It showed that Alisertib led to the augmentation of the iron content and the reduction of the GSH level in mice treated with or without anti-PD-1 compared with control or anti-PD-1 group (Figure 2G, 2H). Additionally, the expression levels of AURKA and ferroptosis-related genes, including nuclear receptor coactivator 4 (NCOA4), acyl-CoA synthetase long-chain family member 4 (ACSL4), and glutaminase-2 (GLS2), were detected in four groups of tumor samples. As shown in Figure 2I, 2J, the mRNA and protein levels of AURKA were decreased in tumor tissues of mice treated with Alisertib compared to control group. Similarly, both levels of AURKA in mice treated with both Alisertib and anti-PD-1 were significantly lower than that in mice treated with anti-PD-1 alone. However, the mRNA and protein levels of three ferroptosis-related genes in four groups of mice showed a completely opposite trend compared to AURKA (Figure 2I, 2J). Moreover, Alisertib treatment could increase PE content in mice treated with or without anti-PD-1 (Figure 2K). To sum up, inhibiting Aurora-A strengthens ferroptosis to improve the efficacy of immune therapy for HCC.

Aurora-A suppresses ferroptosis in HCC cells

Next, we evaluated the effects of Aurora-A on the ferroptosis of HCC cells. RT-qPCR and western blotting results indicated that both mRNA and protein levels of NCOA4, ACSL4, and GLS2 were markedly upregulated by the knockdown of AURKA, whereas they were all downregulated upon AURKA overexpression (Figure 3A, 3B). Meanwhile, we observed the changes of ROS and MDA production in HCC cells to evaluate lipid peroxidation, a feature of ferroptosis. The data showed that AURKA knockdown led to an increase in ROS, while AURKA overexpression resulted in a decrease in ROS (Figure 3C). Similarly, MDA content was augmented by AURKA knockdown, while it was lessened after AURKA overexpression (Figure 3D). In summary, Aurora-A acts as a ferroptosis inhibitor in HCC cells.

Figure 3.

AURKA negatively regulates ferroptosis in HCC cells. A, B. The expression of NCOA4, ACSL4 and GLS2 in HCC cells with AURKA knockdown or overexpression was examined by RT-qPCR and western blotting. C. The effect of AURKA knockdown or overexpression on the ROS level in HCC cells was evaluated. D. The changes in MDA contents in HCC cells with AURKA knockdown or overexpression were tested. **P < 0.01.

Aurora-A inhibitors boost PE production to promote T cell activation

The above data indicate that Aurora-A prevents PE production to inhibit ferroptosis, and Aurora-A inhibitor strengthens ferroptosis to improve the efficacy of immune therapy for HCC. Here, we continued to explore whether Aurora-A inhibitors affected T cell activation and immune therapy through regulating PE. As expected, the addition of Aurora-A inhibitor Alisertib increased PE content in HCC cells (Figure 4A). Next, the co-culture model of CD8⁺ T cells and LM3 cells treated with or without Alisertib was established (Figure 4B). It manifested that CD8⁺ T cells co-cultured with Alisertib-treated HCC cells absorbed more PE compared to those co-cultured with control HCC cells (Figure 4C). Moreover, the accumulation of CD8⁺ T cells was suppressed by overexpression of AURKA, while this phenomenon was reversed upon further addition of PE (Figure 4D). These results suggest that Aurora-A inhibitor activates T cells by enhancing PE in HCC cells. Thereafter, we analyzed the PE-induced changes in HCC tumors that received immune therapy like anti-PD-1 treatment. It showed that the addition of PE or anti-PD-1 led to tumor inhibition, while the combination therapy induced much more suppression of tumor growth (Figure 4E-G). In contrast, CD45⁺CD8⁺ T cell rate in xenograft tumors was partially increased by either PE or anti-PD-1 therapy, but remarkably enhanced after combined treatment of PE and anti-PD-1 (Figure 4H). Also, the positive expression of Ki67 was reduced by PE or anti-PD-1 addition and further decreased after the tumors were processed simultaneously by PE and anti-PD-1 (Figure 4I). Based on these data, we summarized that upregulation of PE enhances the therapeutic effect of anti-PD-1 therapy on HCC downregulated upon AURKA overexpression. These data reveals that inhibiting Aurora-A induces PE production to promote T cell activation in HCC.

Figure 4.

Aurora-A affects T cell infiltration and immune therapy efficacy in a PE-dependent manner. A. The impact of Alisertib on PE production in HCC cells was estimated. B. The co-culture model of CD8⁺ T cells and LM3 cells treated with DMSO or Alisertib. C. The co-localization of PE and CD8 in T cells was observed by confocal fluorescence microscope. Scale bar = 5 or 10 μm. D. Changes in the accumulation of CD45⁺CD8⁺ T cells were assessed via flow cytometry. E. Images of tumors isolated from indicated groups of mice. F, G. Tumor growth curves and tumor weights were presented. H. Changes in the infiltration of CD45⁺CD8⁺ T cells in xenograft tumors were analyzed by flow cytometry. I. IHC analysis of Ki67 positive expression in xenograft tumors. Scale bar = 200 μm. *P < 0.05, **P < 0.01.

Aurora-A facilitates Drp1-Ser616 phosphorylation to hinder MAMs formation, disrupt PS/PE metabolism and impede ferroptosis

Previously, we discovered that Aurora-A disrupts PS/PE metabolism while having no significant effect on mitochondrial PSD, indicating Aurora-A might affect PS transport to affect PE synthesis. Thus, we explored how Aurora-A regulated PS at MAMs, the primary site of PS localization prior to mitochondrial import. The data of confocal fluorescence microscopy manifested that overexpression of AURKA hindered the co-localization of mitochondria with ER in HCC cells (Figure 5A). Drp1 is a pivotal protein involved in mitochondrial fission, and its activated status has recently proposed to participate in MAMs formation [21]. Hence, we detected the role of Aurora-A in regulating Drp1. Results indicated that AURKA overexpression had no effects on the levels of total Drp1 and Drp1 phosphorylated at Ser637 (named as Drp1-Ser637), but induced an obvious increase in the level of Drp1 phosphorylated at Ser616 (named Drp1-Ser616) (Figure 5B). In addition, more activated Drp1 (namely Drp1-Ser616) proteins were found to localize at mitochondria in HCC cells under AURKA overexpression (Figure 5C), indicating AURKA accelerated mitochondrial fission in HCC cells. Then, Drp1 inhibitor Mdivi-1 was applied to verify whether Aurora-A could affect MAMs formation, PE synthesis, and ferroptosis in a Drp1-dependent manner. As a result, the co-localization of mitochondria with ER suppressed by AURKA overexpression was reversed by Mdivi-1 (Figure 5D). Moreover, the repressive effects of AURKA overexpression on PE production, ROS, and MDA content were all counteracted by Mdivi-1-mediated Drp1 inhibition (Figure 5E-G). Similarly, AURKA overexpression-induced downregulation of ferroptosis-related genes at both mRNA and protein levels was also reversed by Mdivi-1 treatment (Figure 5H, 5I). Therefore, we confirm that Aurora-A activates Drp1 to impede MAMs formation, PE production, and ferroptosis.

Figure 5.

Aurora-A facilitates Drp1-Ser616 phosphorylation to hinder MAMs formation, PE synthesis and ferroptosis. A. Confocal images of ER and mitochondria in HCC cells under AURKA overexpression. Scale bar = 10 μm. B. Western blotting tested the impact of AURKA upregulation on the protein levels of total Drp1, Drp1-Ser637 and Drp1-Ser616 in HCC cells. C. Confocal images of the co-localization of Drp1 and mitochondria in indicated HCC cells. Scale bar = 10 μm. D. Confocal images of ER and mitochondria in HCC cells in response to different treatments. Scale bar = 10 μm. E. PE contents under different conditions were assessed. F. ROS levels in indicated HCC cells were evaluated via flow cytometry. G. MDA contents under different conditions were assessed. H, I. The alterations in the expressions of NCOA4, ACSL4 and GLS2 in indicated HCC cells were measured using RT-qPCR and western blotting. **P < 0.01.

Aurora-A inhibitors contribute to MAMs formation and ferroptosis via inactivating Drp1

Finally, we tested whether Aurora-A inhibitor Alisertib exerted functions by affecting Drp1-Ser616 phosphorylation. We found that the level of Drp1-Ser616 was decreased by Alisertib (Figure 6A). Moreover, Alisertib-induced Aurora-A inhibition could promote the co-localization of ER and mitochondria, while this trend was reversed by additional Mdivi-1 treatment (Figure 6B). We then evaluated whether Aurora-A inhibitor affect MAMs formation, PE metabolism, and ferroptosis in a Drp1-dependent manner. Results showed that the co-localization of Drp1-Ser616 with mitochondria was lessened by Alisertib, while the reduced tendency was recovered after treatment with the Drp1 inhibitor Mdivi-1 (Figure 6C). In addition, Alisertib-caused enhancement on PE production was inversed after Mdivi-1 co-addition (Figure 6D). The stimulating effects of Alisertib on ROS, MDA, and the expression of ferroptosis-related genes were all counteracted after Mdivii-1-induced Drp1 inhibition (Figure 6E-H). To conclude, Aurora-A inhibitors accelerate MAMs formation, PE synthesis, and ferroptosis through activating Drp1.

Figure 6.

Aurora-A inhibitors affect MAMs formation, PE synthesis and ferroptosis via Drp1. A. The effects of Alisertib on the levels of Drp1-Ser616 in HCC cells were evaluated by western blotting. B. Confocal images of ER and mitochondria in HCC cells treated with DMSO, Alisertib or Alisertib+Mdivi-1. Scale bar = 10 μm. C. Confocal images of Drp1 and mitochondria in HCC cells with different treatments. Scale bar = 10 μm. D-F. The changes in PE production, ROS and MDA content under different conditions were evaluated. G, H. The expressions of NCOA4, ACSL4 and GLS2 in indicated HCC cells were determined by RT-qPCR and western blotting. **P < 0.01, n.s.: not significant.

Discussion

HCC ranks the third highest mortality rate among all tumor types worldwide. Recently, the FDA has approved two immune-checkpoint inhibitors (ICIs) targeting PD-1, namely pembrolizumab and nivolumab, as second-line treatments for advanced HCC [22]. However, only a small percentage of patients respond to immunotherapy, and most of them finally develop resistance [23]. Hence, searching for new biomarkers to alleviate the resistance is vital. Here, we focused on Aurora-A, a serine/threonine protein kinase that is overexpressed in HCC. Aurora-A has been proven to be overexpressed and highly related to poor prognosis in several malignancies including HCC [24,25]. According to previous studies, Aurora-A and its inhibitors can modulate the radioresistance and chemoresistance of HCC cells [8,11]. Presently, we demonstrated that Aurora-A inhibitor markedly enhanced the anti-tumor effect of anti-PD-1 therapy on HCC. The results indicated Aurora-A as a potential target for improving immunotherapy on HCC.

The occurrence of HCC is often accompanied by a large amount of metabolic reprogramming to create a suitable microenvironment for the growth and proliferation of cancer cells [26]. Through metabolomics analysis, we demonstrated that knockdown AURKA significantly augmented PE biosynthesis. As a phospholipid abundant in membranes, PE plays pivotal parts in multifarious membrane functions [27]. A previous report unveiled lower PE concentrations in HCC patients than in controls [28]. Furthermore, PE is reported to participate in regulating autophagy [29], and changes in its synthesis can affect cancer cell proliferation [16]. Significantly, PE biosynthesis is proposed as a new target for cancer chemotherapy [30]. Besides, PE has high relevance to immune function [31-33]. In this study, we unveiled that inhibiting Aurora-A targeted PE biosynthesis to enhance the sensitivity of HCC cells and tumors to anti-PD-1.

Ferroptosis is a kind of necrosis that occurs in cellular membranes. Usually, it is an iron-dependent process resulting from lipid peroxidation and featured by the accumulation of lipid peroxides [34]. PE is one of phospholipids, which is responsible for lipid peroxidation-induced ferroptosis [18]. Additionally, direct PE oxygenation is a ferroptosis signal [35]. Here, we uncovered that Aurora-A hampered ferroptosis by inhibiting PE production. Meanwhile, targeting ferroptosis is also suggested as a potential pathway for cancer therapy [36,37], including immunotherapy [38]. These reports further support our findings that targeting Aurora-A strengthened the efficacy of anti-PD-1 in HCC via facilitating PE biosynthesis and subsequent ferroptosis.

MAMs are dynamic membrane coupling regions formed by the connection of the mitochondrial outer membrane and ER [39]. Generally, PS is transported from ER to mitochondria via MAMs and then transformed to PE via PSD-mediated decarboxylation in mitochondria [40]. The current study verified that Aurora-A modulated PS/PE metabolism in HCC cells by suppressing MAMs formation. Aberrant MAMs formation induces various pathological events [41]. Drp1 is a main controller of mitochondrial fission through being phosphorylated at Ser616 and Ser637 [42,43]. More importantly, excessive mitochondrial fission hinders MAMs formation [44]. In the present study, we found that Aurora-A facilitated Drp1 phosphorylation at Ser616, thus contributing to mitochondrial fission and reducing MAMs formation.

In conclusion, our findings indicated that Aurora-A impeded PE synthesis to improve anti-PD-1 efficacy by facilitating Drp1-Ser616 phosphorylation to disrupt MAM formation and suppress ferroptosis. The results hint that Aurora-A inhibitors might be suitable for combination with immune therapy in HCC. However, the current study has some limitations. Our experimental results showed that AURKA knockdown led to an overall increase in cellular PS levels, suggesting that AURKA might act as an upstream regulator of PS. Therefore, our findings indicated that AURKA likely influences PS/PE metabolism through two distinct pathways. On one hand, it may exert acute effect by regulating PS transport from the ER to mitochondria; on the other hand, it may exert chronical effect through modulating PS synthesis/degradation. The precise upstream regulatory mechanisms of AURKA on PS will be a key focus for our future research.

Disclosure of conflict of interest

None.

Abbreviations

Aurora-A

Aurora Kinase A Protein

HCC

Hepatocellular carcinoma

PE

phosphatidylethanolamine

PSD

phosphatidylserine decarboxylase

Drp1

dynein-related protein

MAM

mitochondrial-associated membrane

PS

phosphatidylserine

PD-1

Programmed cell death protein 1

THLE2

The healthy liver-derived human cell line

ESI

electrospray ionization

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

NC

nitrocellulose

IHC

Immunohistochemistry