English

• The tumor microenvironment (TME) is a complicated ecosystem with a core formed by tumor cells surrounded by immune cells, endothelial cells, etc. [1]. Immune cells in the TME act like a double-edged sword, whose function depends on multiple factors such as the stage of tumors and patient characteristics. Efficiently reprogramming immune cells to reshape the TME is an important strategy to hinder the progress of cancer [2,3]. Recent studies have mainly focused on T cells [4,5]. Enhancing the antigen presentation efficiency of dendritic cells (DCs) subsequently enhances T-cell activation. Immunogenic cell death (ICD) is a process in which stimulated tumor cells are transferred from a nonimmunogenic state to an immunogenic state, subsequently mediating the immune responses [6]. Several signals required for ICD are termed damage-associated molecular patterns (DAMPs): Adenosine triphosphate (ATP) provides the “find me” signal, calreticulin (CRT) offers the eat me signal, and high mobility group box 1 (HMGB1) might induce DCs maturation [7–9]. With the emergence of insensitive immune checkpoints, it is acknowledged that targeting a single type of immune cell is insufficient [10]. Due to the abundance of TME, much attention should be paid to reversing the immunosuppressive circumstances [11]. Tumor-associated macrophages, the most important member of tumor immunosuppressive circumstances, can be divided into two types: (ⅰ) M1-like macrophages, promoting the release of inflammatory factors and suppressing tumors, and (ⅱ) M2-like macrophages, suppressing immune responses and promoting tumor progress [12]. Thus, lessening the M2-like macrophages and repolarizing them to M1-like macrophages might be a reasonable approach [13–15]. Ferroptosis is a form of programmed cell death (PCD) with a broken mitochondrial membrane induced by excess lipid peroxides, which helps macrophages polarize to M1-like types, amplifying their antitumor ability [16,17]. Different from other forms of PCD, the occurrence of ferroptosis is dependent on iron condensation. Three prerequisites are required in the occurrence of ferroptosis: Polyunsaturated fatty acid-phospholipid synthesis, iron metabolism, and mitochondrial metabolism [18,19]. Except for the influence on macrophages, ferroptosis was reported to promote tumor cells to release or secret DAMPs and induce ICD [20]. Recently, Liang et al. reported an isoquinoline-based cyclometalated Ir(Ⅲ) complex that showed great promise for treating triple-negative breast cancer (TNBC) by synergistically eliciting the ICD response and IDO inhibition through autophagy-dependent ferroptosis [21]. They also reported that a Cu(Ⅱ) complex triggered a ferroptosis-dependent ICD response by disturbing redox homeostasis and effectively inhibited tumor growth in vivo [22]. Overall, novel metal complexes possessing the capacity to regulate multiple immune responses seem to be a credible direction for tumor immunochemotherapy.

  Cisplatin used for the treatment of cancer has made great clinical achievements, paving a path for exploring transition metals [23–25]. Drug resistance and side effects caused by platinum (Pt) drugs have indicated that alternatives are urgently required [26]. Among them, the gold (Au) complex auranofin (AF), first used as an antirheumatoid arthritis agent, has attracted attention. Previous investigations have covered its anti-inflammatory and antitumor effects. It is worth mentioning that AF was reported to be a ferroptosis inducer. The report also covered that thioredoxin reductase (TrxR), the acknowledged target of gold complexes, was the regulator of ferroptosis [27–29]. However, AF reacts readily with the components in blood owing to its high reactivity, which limits its efficacy and leads to side effects. Therefore, stable gold complexes are in demand.

  N-Heterocyclic carbene (NHC), a bioisostere of phosphine ligands, is a strong σ-donator, which binds to metals to form stable units and is difficult to dissociate [30,31]. NHC can combine with diverse metals owing to flexible charges and structures. Changes in side chains can convert the physicochemical properties and reactivity of NHC-metal complexes. Generally, NHC is a better ligand than phosphines for AF [32,33]. Our group has previously reported several NHC-gold(I) complexes with good anticancer activity in liver cancer. Classical investigations revealed that NHC-gold(I) complexes inhibited tumor progress by inducing cell apoptosis mainly through TrxR inhibition [34–37]. In recent, Arambula et al. reported a redox-active NHC-gold(I) complex that induced ICD in vitro and in vivo [38], which stimulated explorations of the potential effects of NHC-gold(I) complexes on the immune response [39,40].

  Dinuclear metal complexes have been mainly studied as catalysts in the past [41]. The introduction of a second metal makes interactions occur between the metal and the carrier and between both metals, changing the structure of the active components of the catalyst and stabilizing the active center. The interaction between two metals is termed metallophilicity while that between two gold atoms is termed aurophilic interaction. Studies in the 1980s indicated that aurophilic interactions could produce redox potentials and fluorescence properties that differ from those of mono-gold complexes. The interaction can stabilize Au⁺, which might improve the stability, solubility, and photovoltaic properties of the complexes [42–45], providing opportunities for ligand diversity and synthesis [46]. Several studies have verified that introducing an additional gold atom in the gold(I) complex could increase the antitumor activity. Wen et al. used a diphosphine ligand to link two gold active centers to obtain the target complex, which significantly inhibited the proliferation of tumor cells compared to other synthesized mono-gold(I) complexes and AF [47]. Che et al. reported a dinuclear gold(I) complex with a mixed bridging diphosphine and bis-NHC ligands, which showed promising inhibitory effects on tumor growth and angiogenesis [48].

  Inspired by the above research, we hypothesized that the introduction of another gold atom into gold(I) NHC complexes would yield complexes with multiple antitumor mechanisms and enhanced antitumor activity. Herein, we reported the synthesis and characterization of 19 dinuclear gold(I) NHC complexes with anti-liver cancer properties. The cationic dinuclear gold(I) complex 4a (BF5-Au) with bis-NHC ligands showed outstanding inhibitory activity against liver cancer cell lines. BF5-Au preferentially accumulated in mitochondria and boosted the level of reactive oxygen species (ROS). ROS-driven endoplasmic reticulum stress (ERS) and ferroptosis further stimulated the immune response and hindered the progress of liver cancer.

  Research has demonstrated that AF suppresses cancer by interfering with the mitochondria. It indicated that gold complexes tending to target the mitochondria may have higher anticancer effects [49,50]. Cationic gold(I) complexes preferentially target mitochondria because they are negatively charged. Hence, we hypothesized that targeting negatively charged mitochondria with dinuclear gold(I) complexes including two Au⁺ would be highly effective. As shown in Scheme 1, we used NHC as the fundamental building block to synthesize a series of dinuclear gold(I) NHC complexes 3a–3j and 4a–4j. Detailed synthetic routes can be found in Scheme S1 (Supporting Information). The specific structural formulae were presented in Fig. S1 (Supporting information). The synthesis was briefly described below. The imidazolium salts 2a–2j were obtained by linking 1-ethyl-4,5-bis(4‑methoxy/fluorophenyl)−1H-imidazole via terminal dibromoalkanes (C3-C5). Then treated with silver oxide in dichloromethane (DCM) to form the corresponding intermediate Ag(Ⅰ)–NHC complexes, which were subjected to ligand exchange reactions with different equivalents of AuCl(SMe₂) to obtain 3a–3j and 4a–4j. The complex 3d was discarded because of its poor solubility which prevented its characterization. The structures of 3a–3j and 4a–4j were confirmed by ¹H- and ¹³C- nuclear magnetic resonance (NMR) (Fig. S2 in Supporting information) and mass spectroscopy (MS). High-performance liquid chromatography (HPLC) was used to assess the purity of complexes (Fig. S3 in Supporting information).

  Scheme 1

  

  Scheme 1.  Synthesis of dinuclear gold(I) NHC complexes 3a–3j and 4a–4j. Reagents and conditions: (Ⅰ) CH₃CN, reflux, 3 days, Yield: 63%–99%; (Ⅱ) Ag₂O, DCM, 18 h, darkness, then 2.0 equiv. AuCl(SMe₂), NaBr, r.t. 6 h, Yield: 33%–85%; (Ⅲ) Ag₂O, DCM, 18 h, darkness, then 1.0 equiv. AuCl(SMe₂), KPF₆, r.t. 6 h, Yield: 4%–57%.

  DownLoad: Full-Size Img PowerPoint

  To determine the structures of these complexes and obtain structural parameters, 4a was characterized through X-ray crystallography (XRC). The single crystal X-ray diffraction result of complex 4a was shown in Fig. 1, revealing a linear structure. 4a was crystallized under slow evaporation from a dichloromethane–n-hexane solution at room temperature (20–25 ℃). The crystal data and structure refinement for 4a were presented in Table S1 (Supporting information). Deposition Number 2403381 contained the supplementary crystallographic data for this paper.

  Figure 1

  

  Figure 1.  Single crystal X-ray diffraction of complex 4a.

  DownLoad: Full-Size Img PowerPoint

  The stability of 4a was analyzed in a deuterated dimethylsulfoxide (DMSO‑d₆) solution containing 10% deuterium oxide (D₂O) by ¹H NMR spectroscopy [51]. According to Fig. S4 (Supporting information), the ¹H NMR spectrum of 4a did not change after 3 days, indicating its stability. The reactivity of the dinuclear gold(I) NHC complex with nucleophilic reagents was studied by exposing 4a to an equimolar amount of glutathione (GSH) or N-acetylcysteine (NAC) for 72 h. 4a remained stable in the NAC-containing solution for 3 days, without significant changes in its ¹H NMR spectrum (Fig. S5 in Supporting information). However, it was unstable in the GSH-containing solution, indicated by the appearance of a small spurious peak at 9.95 ppm after 3 h, attributed to the hydrogen of the –NHN– (Fig. S6 in Supporting information).

  The cytotoxicity of the dinuclear gold(I) NHC complexes was measured via 3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyltetrazolium bromide (MTT) assays. Five liver cancer cell lines were used for evaluation, while AF and oxaliplatin (Oxa) were considered as the positive controls. Table S2 (Supporting information) indicated that this series of complexes generally showed favorable inhibitory activity. 4a showed a higher function on tested cells [half maximal inhibitory concentration (IC₅₀) = 1.61 ± 0.20 µmol/L in HepG2] than AF and Oxa. 4a was termed BF5-Au (B represents bis-carbene, F is derived from the substituent group fluorine, and 5 is the length of the carbon chain between two carbenes). The IC₅₀ values of BF5-Au in Huh7 and lenvatinib-resistant cell line Huh7R were 0.91 ± 0.23 µmol/L and 0.75 ± 0.16 µmol/L, respectively. Hence, we expected that BF5-Au could overcome the resistance of lenvatinib to a certain extent. A clone formation assay was used for the confirmation of the inhibitory capacity. BF5-Au prominently inhibited the cell proliferation of HepG2 in a dose-dependent manner (Fig. S7A in Supporting information). In addition, the transwell assay indicated that BF5-Au could decrease and restrain tumor invasion, its effect at 4 µmol/L was higher than the Oxa group (Fig. S7B in Supporting information).

  The following structure-activity relationships can be deduced from the cytotoxicity data (Fig. S1 and Table S2 in Supporting information): (ⅰ) The antiproliferative activities of cationic dinuclear gold(I) NHC complexes with bis-NHC ligands (4a–4j) were generally superior to the neutral dinuclear gold(I) NHC complexes (3a–3j). (ⅱ) Effects of linker chain between two carbenes: aliphatic chain hydrocarbons > aromatic chain hydrocarbons, large steric effect between aromatic chain hydrocarbons and imidazolium salts made the corresponding dinuclear gold(I) complexes extremely difficult to solubilize, thus limiting activity. (ⅲ) In general, the IC₅₀ values of the fluorine-substituted dinuclear gold(I) complexes decreased with the increasing length of the carbon chain.

  TrxR is the target of gold complexes, and abundant gold complexes were reported to exert antitumor effects through TrxR inhibition [52]. However, BF5-Au failed to inhibit isolated TrxR or that in cells (Fig. S8 in Supporting information), which might account for its stability preventing the exposure of the gold atom to react with TrxR. Therefore, we attempted to decode its mechanisms by detecting its cellular distribution. Inductively coupled plasma-mass spectrometry (ICP-MS) illustrated the existence of free gold ions in mitochondria with a content of 40 ng/10⁶ cells (Fig. 2A). The cellular distribution of BF5-Au was attempted to visualize by introducing a fluorophore to the complex, which failed. As an alternative, a gold atom-based probe was synthesized. Fig. 2B showed that cells were concurrently stained with the gold atom probe and organelle trackers. The overlap of mitochondria with the gold atom probe was much more than others, consistent with the results of ICP-MS. Thus, we concluded that BF5-Au mainly accumulated in mitochondria to exert its anticancer effects. Its function on mitochondria was confirmed through transmission electron microscopy, and the results were shown in Fig. 2C. BF5-Au-treated mitochondria became round and the cristae decreased and even disappeared (red arrow). According to these results, we suggested that BF5-Au destroyed the morphology and impaired mitochondrial function.

  Figure 2

  

  Figure 2.  Cellular distribution of BF5-Au and the generation of H₂O₂. (A) Gold ions content of organelles in HepG2 after treatment with 4 µmol/L BF5-Au for 24 h. (B) Colocalization of the gold probe and organelle tracker in HepG2 after 4-µmol/L BF5-Au treatment for 3 h. Scale bar: 500 nm. (C) Morphology of HepG2 after 4 µmol/L BF5-Au treatment for 24 h. Scale bar: 500 nm. (D) Change in H₂O₂ observed by laser confocal microscopy after 4-µmol/L BF5-Au treatment for 3 h. Scale bar: 10 µm. Data are presented as mean ± standard deviation (SD) (n = 3).

  DownLoad: Full-Size Img PowerPoint

  As mitochondria are the primary sites of ROS generation, we speculated that BF5-Au could boost the level of ROS. 2′,7′-Dichlorodihydrofluorescein diacetate (DCFH-DA) and dihydroethidium (DHE), two ROS probes, were used to prove that a high dose of BF5-Au generated high ROS (Fig. S9 in Supporting information). The site of ROS generation was investigated using the Hyper7 sensor fabricated by our group [53]. As shown in Fig. S10 (Supporting information), the content of hydrogen peroxide (H₂O₂) in several organelles was detected at different time points and changes occurred mainly in mitochondria (~1.27-fold higher than in the control in 3 h). The changes in the endoplasmic reticulum (ER) were similar to those in mitochondria (~1.24-fold higher than that of the control in 3 h), as analyzed and confirmed using laser confocal microscopy (Fig. 2D). Consequently, BF5-Au might induce ROS generation in mitochondria and ER owing to the interplay between mitochondria and the endoplasmic reticulum.

  The ROS function in the ER was discussed first as previous investigations indicated that gold complexes could induce ICD via ROS-driven ERS. We first reviewed the results of transmission electron microscopy (Fig. 2C) and found that the treated ER became swollen and discrete, and vesicles of varying sizes appeared (green arrow). Salubrinal diminished the lethality of BF5-Au, demonstrating its effect on ERS (Fig. S11B in Supporting information). In addition, the expression levels of ERS-related proteins such as C/Emopamil-binding homologous protein (CHOP) and 78 kDa glucose-regulated protein (GRP78) changed (Figs. 3A and B, Figs. S12A and B in Supporting information). Thus, BF5-Au could induce ERS of HepG2 cells. ERS participates in the regulation of cell apoptosis. The apoptotic rate after BF5-Au treatment was thrice that of the N,N-dimethylformamide (DMF)-treated group and higher than the Oxa-treated group. Apoptotic bodies increased and apoptosis-related proteins [apoptosis inducing factor (AIF) and Bcl-2-associated X (Bax)] remarkably changed (Fig. S13 in Supporting information), indicating that BF5-Au could induce apoptosis.

  Figure 3

  

  Figure 3.  BF5-Au induced ERS and ICD. (A) ERS-related proteins detected through Western blot after BF5-Au (1, 2, and 4 µmol/L) and 10 µmol/L Oxa treatment for 24 h. (B) Quantitative chart of (A). (C) Expression of CRT detected by fluorescence microscopy after 4 µmol/L BF5-Au treatment for 24 h. Scale bar: 10 µm. 2-(4-Amidinophenyl)−6-indolecarbamidine dihydrochloride (DAPI), blue; CRT, green. (D) Level of cellular ATP detected after BF5-Au (1, 2, and 4 µmol/L) and 10 µmol/L Oxa treatment for 24 h. (E) Expression of HMGB1 detected through ELISA after 4-µmol/L BF5-Au treatment for different durations. (F) Maturation rate of DCs analyzed after being cocultured for 24 h with 0.5 µmol/L BF5-Au and 10 µmol/L Oxa. (G) Quantitative chart of (F). Data are presented as mean ± SD (n = 3). ns, P > 0.05. P < 0.05, **P < 0.01, ***P < 0.001.

  DownLoad: Full-Size Img PowerPoint

  ERS and apoptosis possessed the capacity to induce ICD. The key regulators of DAMPs include ATP, CRT, and HMGB1. BF5-Au prominently promoted CRT to divert the cell membrane, decreased the ATP level, and accelerated HMGB1 release to the extracellular space, inhibiting the degradation of tumor-associated antigens (Figs. 3C–E and Figs. S12C–G in Supporting information). DAMPs could bind with the corresponding molecules expressed on the surface of DCs, induce the release of immune-activated cytokines, inhibit the degradation of tumor-associated antigens, enhance the capacity of antigen presentation, and avoid the immune escape of tumor cells. Coculture assays revealed that the maturity rate of BF5-Au-treated DCs was approximately 2-fold higher than the DMF-treated group and better than the Oxa-treated group (Figs. 3F and G). In summary, BF5-Au facilitated ICD and promoted DCs for maturation.

  Excessive ROS could lead to mitochondrial dysfunction such as decreasing the mitochondrial membrane potential (MMP). A 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethyl-imidacarbocyanine iodide (JC-1) probe was used to detect the MMP of cells incubated for 12 h (Figs. 4A and B, Fig. S14A in Supporting information). The fluorescence intensity at 1 µmol/L barely showed any changes, which remarkably increased at 2 and 4 µmol/L with decreasing red fluorescence. BF5-Au reduced the MMP, indicating mitochondrial dysfunction.

  Figure 4

  

  Figure 4.  BF5-Au damaged mitochondria and induced ferroptosis. (A) MMP detected through fluorescence microscopy and flow cytometry after treatment with BF5-Au (1, 2, and 4 µmol/L) and 10 µmol/L Oxa for 12 h. (B) Quantitative chart of (A). (C) Proportion of GSH/GSSG measured after treatment with BF5-Au (1, 2, and 4 µmol/L) for 12 h. (D) Types of macrophages distinguished by flow cytometry after being cocultured with Hepa 1–6 treated with BF5-Au (0.5, 1, and 2 µmol/L). (E) Lipid peroxide via fluorescence microscopy after treatment with BF5-Au (1, 2, and 4 µmol/L) for 12 h. Data are presented as mean ± SD (n = 3). P < 0.05, **P < 0.01, ***P < 0.001. Scale bar: 100 µm.

  DownLoad: Full-Size Img PowerPoint

  Ferroptosis is a form of cell death independent of necrocytosis, apoptosis, and autophagy [54,55]. An imbalance in cellular redox homeostasis generates lipid peroxide, subsequently inducing ferroptosis. Several gold complexes exhibited anticancer activity by inducing ferroptosis. Since mitochondria metabolism is a ferroptosis prerequisite and BF5-Au could damage mitochondria, we speculated that BF5-Au could induce ferroptosis. Redox homeostasis was measured using total glutathione and oxidized glutathione (GSH/GSSG) kits. Fig. 4C showed that the ratios of GSH/GSSG in treated groups were lower than the DMF-treated group, indicating the imbalance of cellular redox activity. Subsequently, ferrostatin-1, a ferroptosis inhibitor, was used to investigate the activity of BF5-Au on ferroptosis (Fig. S11A in Supporting information). Another characteristic of ferroptosis is the generation of lipid peroxides. C11 BODIPY, a lipid peroxide probe, indicated that the intensity of green fluorescence was remarkably higher after the treatment of BF5-Au (Fig. 4E, Figs. S14B and C in Supporting information) than the DMF-treated group. An increasing dosage strengthened the fluorescence intensity. BF5-Au could reduce the GSH/GSSG ratio, and induce lipid peroxide accumulation and ferroptosis. Iron metabolism of cancer cells can regulate the macrophage types in the immune TME. Specifically, iron overloading involved in ferroptosis can prompt macrophage polarization to the M1 type, which secretes proinflammatory factors such as tumor necrosis factor (TNF)-α [56]. Fig. 4D showed the analyzed changes in macrophage types after cocultivation. The ratio of M1/M2 in treated groups was higher than the DMF-treated group. However, the effect of the BF5-Au-treated group was lower than the Oxa-treated group.

  The RNA sequencing was performed to investigate the antitumor mechanisms of BF5-Au toward HepG2 cells. As shown in Fig. S15 (Supporting information), the treatment with BF5-Au upregulated 2754 genes and downregulated 2976 genes (Fig. S15A). Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis revealed that BF5-Au functioned on redox homeostasis (ferroptosis), cell growth and death (apoptosis), cellular metabolism (fatty acid biosynthesis), inflammation and immune response (IL-17 signaling pathway), etc. (Fig. S15B). Gene ontology (GO) enrichment analysis indicated that BF5-Au acted on cellular processes such as response to unfolded protein, regulation of wound healing (Fig. S15C). Our attention was paid to pathways such as redox homeostasis, ERS, ferroptosis, and immune response. Venn analysis indicated that there were 38 differentially expressed genes (DEGs) related to redox, 32 to ERS, 19 to ferroptosis, and 202 to immunity (Fig. S15D). Protein-protein interaction networks were analyzed and simplified (Fig. S15E). Subsequently, the change of these genes was selected and investigated. The ERS-related targets [tribbles homolog 3 (TRIB3), ERN1, and ER protein with ubiquitin-like domain 1 (HERPUD1)] and ferroptosis-related targets [spermidine/spermine N1-acetyltransferase 1 (SAT1) and heme oxygenase 1 (HMOX1)] were more upregulated compared to the control group (Fig. S15F). BF5-Au mainly targeted the signaling of ERS and ferroptosis. We speculated that BF5-Au stimulated immune response (DCs maturation and the M1 polarization of macrophage) by inducing DAMPs (ERS-driven apoptosis and ferroptosis).

  The in vivo efficacy of BF5-Au was analyzed by designing subcutaneous tumor models. All C57BL/6 mice (male, 3–5 weeks) were maintained in pathogen-free conditions and in vivo experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, the USA) and approved by the animal ethical committee of Nanjing University of Chinese Medicine. Mice were randomly divided into three groups when tumors grew to 100 mm³ [vehicle, BF5-Au, and positive control (Oxa)] after intraperitoneal administration. During this period, the weight of mice and the volume of tumors were recorded.

  The tumor volumes of BF5-Au treatment were significantly smaller than both vehicle- and Oxa-treated groups. As shown in Fig. S16 (Supporting information), after the sacrifice, the tumors were removed and weighed, which were consistent with the results of volumes (Figs. S16A, B, D and E). During this period, the weight of mice persistently increased. However, the growth of BF5-Au-treated mice was slower than the vehicle-treated group (Fig. S16C). The major organs were removed and stained using hematoxylin-eosin (H&E), indicating that the microstructure of these organs was normal (Fig. S16F). The in vivo antitumor mechanism was further studied by separating tumors and staining with H&E. The amount of tumor cell nucleus in the BF5-Au group was less than the vehicle- and Oxa-groups. A terminal deoxynucleotidyl transferase dUTP nick end labeling assay revealed that the fluorescence intensity of tumor cells in the BF5-Au group was the strongest, indicating a high apoptosis rate. Besides, the increase in ROS generation of tumor tissues in the BF5-Au group was observed (Fig. S16G). The in vivo anticancer mechanism in the liver was studied based on the results obtained in vitro. As shown in Fig. S17 (Supporting information), the immunofluorescence results indicated that the expression of CHOP and AIF related to ERS and apoptosis elevated, while the HMGB1 expression associated with ICD diminished (Figs. S17A and B). The results illustrated that BF5-Au could induce ERS, apoptosis, and ICD in vivo.

  As BF5-Au could regulate immune responses in vitro, the changes in immune cells in tumors were further analyzed. Cluster of differentiation 4 (CD4⁺) T cells and CD8⁺ T cells were upregulated in the treatment of BF5-Au. CD4⁺ T cells were also upregulated in the Oxa group, while there was no significant change in CD8⁺ T cells (Figs. S17E–H). An immunohistochemistry assay was performed on the supplement, which exhibited consistent results (Fig. S17C). Besides, cytokines in blood were measured and the results showed that compared with the vehicle group, the level of TNF-α was elevated. Unfortunately, there was no statistical difference of TNF-α between groups (Fig. S17D). Overall, BF5-Au could activate helper and cytotoxic T cells, demonstrating its potential antitumor effects.

  ICD was reported to function similarly to a tumor vaccine. The in vivo tumor vaccination experiment was another standard crucial method to verify whether BF5-Au could facilitate ICD. The treated Hepa 1–6 cells (BF5-Au and Oxa groups) were collected and injected into the left underarm as the primary tumor. Untreated Hepa 1–6 cells were seeded after 7 days to the right underarm as the distant tumor. During the period, the body weight and the rate of tumor formation were recorded. The weight of each group showed no difference (Fig. S18A in Supporting information). After the distant tumor seeding for 15 days, the incidence of the DMF group was 100%. The incidence of the BF5-Au group was only 70%, which was lower than the Oxa group (80% incidence) (Fig. S18B in Supporting information). BF5-Au could facilitate ICD in vivo and also exhibited potential antitumor “vaccine” effects.

  In summary, we reported a series of dinuclear gold(I) NHC complexes with potential antitumor activity against liver cancer cells. The complexes were characterized via NMR and MS, and HPLC for purity. The novel cationic dinuclear gold(I) complex 4a (BF5-Au) with bis-NHC ligands was the most active complex with a linear geometry, indicated through XRC. This series of complexes showed promising inhibitory activity and BF5-Au was selected for further investigation based on its superior antitumor activity. Although BF5-Au showed no inhibitory ability toward TrxR, it exhibited good antitumor activity by preferentially accumulating in mitochondria and boosting the cellular ROS level.

  The cellular H₂O₂ detector Hyper7 sensor was first used to study the site of ROS generation of BF5-Au. ROS were preferably generated in mitochondria and the ER after exposure to BF5-Au. ROS in the ER could cause excess unfolding or misfolding of proteins and subsequently induce ERS, ultimately leading to apoptosis. ERS and apoptosis facilitated the secretion and release of DAMPs, inducing ICD. It could promote the decrease of the MMP, affecting the function of mitochondria. Ferroptosis occurred with the cumulative increase of lipid peroxide. Cancer cells during ferroptosis could direct macrophages to differentiate into M1-like macrophages, which secrete proinflammatory cytokines that enhance the immune response to destroy cancer cells. BF5-Au was responsible for inducing these effects. As supplements, RNA-sequencing analysis verified these effects from the perspective of genes. The subcutaneous tumor model was designed to confirm the in vivo efficacy of BF5-Au, and the tumor vaccine model was established to confirm the ICD-inducing effects. Overall, this in-depth study indicated that BF5-Au could suppress the tumor growth and remodel the TME by enhancing the immune response via ROS-driven ERS and ferroptosis, providing valued information for designing future gold(I) drug candidates.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  CRediT authorship contribution statement

  Qiuyue Lu: Writing – review & editing, Writing – original draft, Formal analysis, Data curation, Conceptualization. Min Shan: Writing – review & editing, Writing – original draft, Data curation, Conceptualization. Jiaqi Yang: Writing – original draft, Data curation, Conceptualization. Zhongren Xu: Writing – original draft, Data curation, Conceptualization. Yueyue Lei: Writing – original draft, Formal analysis, Data curation. Wukun Liu: Writing – review & editing, Writing – original draft, Supervision, Data curation, Conceptualization.

  Acknowledgments

  This study was supported by the National Natural Science Foundation of China (No. 82173684), the Priority Academic Program Development of Jiangsu Higher Education Institutions (Integration of Chinese and Western Medicine), High level key discipline construction project of the National Administration of Traditional Chinese Medicine-Resource Chemistry of Chinese Medicinal Materials (No. zyyzdxk-2023083), the Key R&D Program of Jiangsu Province (No. BE2023840), and Yunnan Provincial Science and Technology Talent and Platform Program (No. 202405AF140031).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.110940.

  
  
• 1. [1]

     K.E. de Visser, J.A. Joyce, Cancer Cell 41 (2023) 374–403.

  2. [2]

     L. Bejarano, M.J.C. Jordāo, J.A. Joyce, Cancer Discov. 11 (2021) 933–959.

  3. [3]

     A. Dominiak, B. Chełstowska, W. Olejarz, G. Nowicka, Cancers 12 (2020) 1232. doi: 10.3390/cancers12051232

  4. [4]

     R. Zappasodi, T. Merghoub, J.D. Wolchok, Cancer Cell 33 (2018) 581–598.

  5. [5]

     P. Sharma, B.A. Siddiqui, S. Anandhan, et al., Cancer Discov. 11 (2021) 838–857. doi: 10.1158/2159-8290.cd-20-1680

  6. [6]

     D.V. Krysko, A.D. Garg, A. Kaczmarek, et al., Nat. Rev. Cancer 12 (2012) 860–875. doi: 10.1038/nrc3380

  7. [7]

     M. Obeid, A. Tesniere, F. Ghiringhelli, et al., Nat. Med. 13 (2007) 54–61. doi: 10.1038/nm1523

  8. [8]

     M.R. Elliott, F.B. Chekeni, P.C. Trampont, et al., Nature 461 (2009) 282–286. doi: 10.1038/nature08296

  9. [9]

     L. Apetoh, F. Ghiringhelli, A. Tesniere, et al., Nat. Med. 13 (2007) 1050–1059. doi: 10.1038/nm1622

  10. [10]

      S. Kumagai, Y. Togashi, T. Kamada, et al., Nat. Immunol. 21 (2020) 1346–1358. doi: 10.1038/s41590-020-0769-3

  11. [11]

      D. Wang, X. Wu, Y. Sun, Signal. Transduct. Target. Ther. 7 (2022) 331.

  12. [12]

      A.J. Petty, Y. Yang, Immunotherapy 9 (2017) 289–302. doi: 10.2217/imt-2016-0135

  13. [13]

      P. Pathria, T.L. Louis, J.A. Varner, Trends. Immunol. 40 (2019) 310–327.

  14. [14]

      Y. Xia, L. Rao, H. Yao et al., Adv. Mater. 32 (2020) 2002054.

  15. [15]

      C. Ngambenjawong, H.H. Gustafson, S.H. Pun, Adv. Drug. Deliv. Rev. 114 (2017) 206–221.

  16. [16]

      I. Efimova, E. Catanzaro, L. Van der Meeren, et al., J. Immunother. Cancer 8 (2020) e001369. doi: 10.1136/jitc-2020-001369

  17. [17]

      B. Yu, B. Choi, W. Li, D.H. Kim, Nat. Commun. 11 (2020) 3637.

  18. [18]

      B. Gan, J. Cell Biol. 20 (2021) e202105043.

  19. [19]

      X. Jiang, B.R. Stockwell, M. Conrad, Nat. Rev. Mol. Cell Biol. 22 (2021) 266–282. doi: 10.1038/s41580-020-00324-8

  20. [20]

      Q. Wen, J. Liu, R. Kang, B. Zhou, D. Tang, Biochem. Bioph. Res. Commun. 510 (2019) 278–283.

  21. [21]

      Y. Lu, S. Wang, M. Li, et al., Acta Pharm. Sin. B 15 (2025) 424–437.

  22. [22]

      K. Huang, F. Wang, Y. Lu, et al., Proc. Natl. Acad. Sci. U. S. A. 121 (2024) e2404668121.

  23. [23]

      S. Ghosh, Bioorg. Chem. 88 (2019) 102925.

  24. [24]

      A. Johnson, C. Olelewe, J.H. Kim, et al., Chem. Sci. 14 (2023) 557–565. doi: 10.1039/d2sc04707a

  25. [25]

      X. Shao, F. Xing, Y. Zhang, C.N. Lok, C.M. Che, Chem. Sci. 15 (2024) 5349–5359. doi: 10.1039/d3sc04834a

  26. [26]

      X. Wang, X. Wang, S. Jin, N. Muhammad, Z. Guo, Chem. Rev. 119 (2019) 1138–1192. doi: 10.1021/acs.chemrev.8b00209

  27. [27]

      R.T. Mertens, S. Gukathasan, A.S. Arojojoye, C. Olelewe, S.G. Awuah, Chem. Rev. 123 (2023) 6612–6667. doi: 10.1021/acs.chemrev.2c00649

  28. [28]

      L. Yang, P. Shi, G. Zhao, et al., Signal. Transduct. Target. Ther. 5 (2020) 8. doi: 10.1089/acm.2019.0105

  29. [29]

      L.F. Boullosa, J. Van Loenhout, T. Flieswasser, et al., Redox Biol. 42 (2021) 101949.

  30. [30]

      S.P. Nolan, N-Heterocyclic Carbenes in Synthesis, John Wiley & Sons, Weinheim, 2006.

  31. [31]

      D. Nemcsok, K. Wichmann, G. Frenking, Organometallics 23 (2004) 3640–3646.

  32. [32]

      L. Cavallo, A. Correa, C. Costabile, H. Jacobsen, J. Organomet. Chem. 690 (2005) 5407–5413.

  33. [33]

      B.C. Lee, C.F. Liu, L.Q.H. Lin, et al., Chem. Soc. Rev. 52 (2023) 2946–2991. doi: 10.1039/d2cs00972b

  34. [34]

      M. Bian, R. Fan, G. Jiang, et al., J. Med. Chem. 63 (2020) 9197–9211. doi: 10.1021/acs.jmedchem.0c00257

  35. [35]

      R. McCall, M. Miles, P. Lascuna, et al., Chem. Sci. 8 (2017) 5918–5929.

  36. [36]

      F. Papi, C. Bazzicalupi, M. Ferraroni, et al., Chem. Eur. J. 23 (2017) 13784–13791. doi: 10.1002/chem.201702854

  37. [37]

      Y. Lu, Y. Liu, Z. Liang, et al., Chin Chem. Lett. 34 (2023)108413.

  38. [38]

      S. Sen, S. Hufnagel, E.Y. Maier, et al., J. Am. Chem. Soc. 142 (2020) 20536–20541. doi: 10.1021/jacs.0c09753

  39. [39]

      R.D. Mule, A. Kumar, S.P. Sancheti, et al., Chem. Sci. 13 (2022) 10779–10785. doi: 10.1039/d2sc03756d

  40. [40]

      Z. Yang, M. Bian, L. Lv, et al., J. Med. Chem. 66 (2023) 3934–3952. doi: 10.1021/acs.jmedchem.2c01798

  41. [41]

      T. Miyake, T. Asakawa, Appl. Catal. A: Gen. 280 (2005) 47–53.

  42. [42]

      H. Schmidbaur, A. Schier, Chem. Soc. Rev. 41 (2012) 370–412.

  43. [43]

      P.J. Barnard, M.V. Baker, S.J. Berners-Price, B.W. Skelton, A.H. White, Dalton Trans. 7 (2004) 1038–1047.

  44. [44]

      G. Trevisan, V. Vitali, C. Tubaro, et al., Dalton Trans. 50 (2021) 13554–13560. doi: 10.1039/d1dt02444b

  45. [45]

      P.J. Barnard, L.E. Wedlock, M.V. Baker, et al., Angew. Chem. Int. Ed. 45 (2006) 5966–5970. doi: 10.1002/anie.200601526

  46. [46]

      M. De Franco, L. Biancalana, C. Zappelli, et al., J. Med. Chem. 67 (2024) 11138–11151. doi: 10.1021/acs.jmedchem.4c00641

  47. [47]

      Q. Cui, W. Ding, P. Liu, et al., Int. J. Mol. Sci. 23 (2022) 12169. doi: 10.3390/ijms232012169

  48. [48]

      T. Zou, C. Lum, C. Lok, et al., Angew. Chem. Int. Ed. 53 (2014) 5810–5814. doi: 10.1002/anie.201400142

  49. [49]

      P. Zou, M. Chen, J. Ji, et al., Oncotarget 7 (2016) 18050–18064. doi: 10.18632/oncotarget.7633

  50. [50]

      T. Gamberi, T. Fiaschi, A. Modesti, et al., Int. J. Biochem. Cell. Biol. 65 (2015) 61–71.

  51. [51]

      S.M. Meier-Menches, B. Neuditschko, K. Zappe, et al., Chem. Eur. J. 26 (2020)15528–15537. doi: 10.1002/chem.202003495

  52. [52]

      Y. Lu, X. Ma, X. Chang, et al., Chem. Soc. Rev. 51 (2022) 5518–5556. doi: 10.1039/d1cs00933h

  53. [53]

      Z. Xu, J. Xu, S. Sun, et al., Redox Biol. 54 (2022) 102351.

  54. [54]

      X. Chen, R. Kang, G. Kroemer, D. Tang, Nat. Rev. Clin. Oncol. 18 (2021) 280–296. doi: 10.1038/s41571-020-00462-0

  55. [55]

      Z. Gao, S. Zheng, K. Kamei, C. Tian, Acta Mater. Med. 1 (2022) 411–426.

  56. [56]

      A.A. Kapralov, Q. Yang, H.H. Dar, et al., Nat. Chem. Biol. 16 (2020) 278–290. doi: 10.1038/s41589-019-0462-8

• 

  Scheme 1  Synthesis of dinuclear gold(I) NHC complexes 3a–3j and 4a–4j. Reagents and conditions: (Ⅰ) CH₃CN, reflux, 3 days, Yield: 63%–99%; (Ⅱ) Ag₂O, DCM, 18 h, darkness, then 2.0 equiv. AuCl(SMe₂), NaBr, r.t. 6 h, Yield: 33%–85%; (Ⅲ) Ag₂O, DCM, 18 h, darkness, then 1.0 equiv. AuCl(SMe₂), KPF₆, r.t. 6 h, Yield: 4%–57%.

  下载: 全尺寸图片 幻灯片
  

  Figure 1  Single crystal X-ray diffraction of complex 4a.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  Cellular distribution of BF5-Au and the generation of H₂O₂. (A) Gold ions content of organelles in HepG2 after treatment with 4 µmol/L BF5-Au for 24 h. (B) Colocalization of the gold probe and organelle tracker in HepG2 after 4-µmol/L BF5-Au treatment for 3 h. Scale bar: 500 nm. (C) Morphology of HepG2 after 4 µmol/L BF5-Au treatment for 24 h. Scale bar: 500 nm. (D) Change in H₂O₂ observed by laser confocal microscopy after 4-µmol/L BF5-Au treatment for 3 h. Scale bar: 10 µm. Data are presented as mean ± standard deviation (SD) (n = 3).

  下载: 全尺寸图片 幻灯片
  

  Figure 3  BF5-Au induced ERS and ICD. (A) ERS-related proteins detected through Western blot after BF5-Au (1, 2, and 4 µmol/L) and 10 µmol/L Oxa treatment for 24 h. (B) Quantitative chart of (A). (C) Expression of CRT detected by fluorescence microscopy after 4 µmol/L BF5-Au treatment for 24 h. Scale bar: 10 µm. 2-(4-Amidinophenyl)−6-indolecarbamidine dihydrochloride (DAPI), blue; CRT, green. (D) Level of cellular ATP detected after BF5-Au (1, 2, and 4 µmol/L) and 10 µmol/L Oxa treatment for 24 h. (E) Expression of HMGB1 detected through ELISA after 4-µmol/L BF5-Au treatment for different durations. (F) Maturation rate of DCs analyzed after being cocultured for 24 h with 0.5 µmol/L BF5-Au and 10 µmol/L Oxa. (G) Quantitative chart of (F). Data are presented as mean ± SD (n = 3). ns, P > 0.05. P < 0.05, **P < 0.01, ***P < 0.001.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  BF5-Au damaged mitochondria and induced ferroptosis. (A) MMP detected through fluorescence microscopy and flow cytometry after treatment with BF5-Au (1, 2, and 4 µmol/L) and 10 µmol/L Oxa for 12 h. (B) Quantitative chart of (A). (C) Proportion of GSH/GSSG measured after treatment with BF5-Au (1, 2, and 4 µmol/L) for 12 h. (D) Types of macrophages distinguished by flow cytometry after being cocultured with Hepa 1–6 treated with BF5-Au (0.5, 1, and 2 µmol/L). (E) Lipid peroxide via fluorescence microscopy after treatment with BF5-Au (1, 2, and 4 µmol/L) for 12 h. Data are presented as mean ± SD (n = 3). P < 0.05, **P < 0.01, ***P < 0.001. Scale bar: 100 µm.

  下载: 全尺寸图片 幻灯片
•