The immune-regulatory effects of gold drugs have been comprehensively reviewed in the literatures concerning rheumatoid arthritis as well as other immune-related diseases, such as HIV and malaria (Griem and Gleichmann, 1996; Madeira et al., 2012; Nardon et al., 2016). Figure 2 shows the chemical structure of several common gold compounds. Several common metals (such as gold, nickel, copper, and mercury) have been found to have the ability to stimulate innate immunity (Suzuki et al., 2011; Rachmawati et al., 2015a). A series of in vitro and in vivo studies have shown that gold compounds can not only promote direct immune cell-mediated destruction, but also synergistic promote the T cell-based anticancer immunity cycle via DC (see Figure 3). The complex mechanisms of the effects of gold compounds on the innate immune system are summarized in Table 1. Gold compounds can induce cancer cells destruction through various forms of cytotoxic action, leading to the expression of proteins on the cell surface, secretion of cytokines, or rupture of the plasma membrane and release of intracellular substances. Released cytoplasmic molecules are danger signals, known as DAMPs, which enable the immune system more sensitive to the recognition of tumor antigens. Rachmawati et al. reported that the gold compound (Na₃Au (S₂O₃)2·₂H₂O, Figure 2D) induced substantial release of the pro-inflammatory mediator IL-8 from DC, PBMC, and THP-1 cells and expression of CD40 on the surface of DC, indicating DC's maturation and adaptive immune stimulatory capacity. The ability of this gold compound to induce innate immune responses can be attributed to TLR3 dependent signaling (Rachmawati et al., 2015b). I. Stern et al. observed the effects of auranofin (Figure 2A), gold sodium thiomalate (Figure 2C), and HAuCl4 [Au (III)] (Figure 2E) on the ability of LPS-induced THP1 monocytes to secrete key inflammatory cytokines (IL6, IL8, IL10, and TNF α) in vitro. The results showed that sub-lethal concentrations of the three gold compounds could differentially modulate activation of monocytes. Among them, the effect of auranofin was slightly stronger (Stern et al., 2005). Another study suggested that the activation of monocytes may be related to the success of gold compounds in the treatment of rheumatoid arthritis (Hurst et al., 1986). Mast cells play key roles in allergic and inflammatory responses. There is considerable evidence that mast cells are important for adaptive and innate immunity (Galli et al., 2005), as well as the development of autoimmune diseases (Costela-Ruiz et al., 2018). In a previous review, gold, mercury, and silver all had effects on mast cell signaling, function, and survival, inducing aberrant immunological responses. All these compounds stimulated mast cells to degranulate and secrete arachidonic acid metabolites as well as cytokines such as interleukin-4 (Suzuki et al., 2011). HAuCl4 [Au (III)] at a non-toxic concentration ( ≤ 50 μM) stimulated large amounts of degranulation and leukotriene C4 secretion in mast cells by a Ca2⁺-dependent manner (Hayama et al., 2011). It has been reported that auranofin has a dose-dependent bidirectional effect on NK cell activity, enhancing NK cell activity at low dose and inhibiting NK cell activity at high dose in vitro (Russell et al., 1982; Pedersen and Abom, 1986). However, as for TLR signaling, Youn et al. found that auranofin suppressed LPS-induced homodimerization of TLR4 and TLR4-mediated activation of key transcription factors (such as NF-κB, IRF3, and COX-2) in mice pro-B as well as monocytic cell lines. In addition, auranofin also inhibited NF-κB activation induced by MyD88-dependent downstream signaling elements of TLR4, MyD88, IKKβ, and p65 (Youn et al., 2006). Auranofin suppressed multiple steps in TLR4 and downstream signaling, thereby inhibiting immune inflammation. The experimental results of Wang et al. showed that disodium aurothiomalate could inhibit the activity of CD45, a protein-tyrosine phosphatase expressed on all white blood cells, which could enhance the signaling of B and T cell antigen receptors (Wang et al., 1997).

Gold compounds have also been shown to affect some important signaling pathways and key transcription factors. The NF-κB and protein kinase C (PKC) signaling pathways play a key role in the process of activation, differentiation, and maturation of myeloid and lymphatic cells. NF-κB is a key transcription factor involved in the expression of many inflammatory genes and plays an important role in oncogenesis, which is associated with the proliferation of multiple types of tumors (Zeligs et al., 2016). There are growing interests in exploring novel regulators to inhibit NF-κB activation, because blocking different steps of NF-κB signaling pathway may slow tumor growth, progression, and chemotherapy resistance. Auranofin suppresses nuclear translocation of NF-κB by blocking IκB kinase (IKK) activation in macrophages. This inhibitory activity may be related to the suppression of TNF-α (Jeon et al., 2000). It also has been reported that auranofin inhibits NF-κB activation by modifying Cys-179 of IKKβ subunit in monkey kidney (COS-7) cells as well as LPS-stimulated macrophages (Jeon et al., 2003). Aurothioglucose (Figure 2F) has a strong inhibitory effect on the DNA binding activity of NF-κB, which is essential for its performance (Yang et al., 1995). The PKC activity was shown to be inhibited by auranofin as well as gold sodium thiomalate in human neutrophils (Parente et al., 1989).

However, gold (I) compounds have been clinically used to treat rheumatoid arthritis due to their anti-inflammatory properties, which appears contradictory to its immune stimulatory activity. This intrigues clinicians and toxicologists for many years. Anyway, the gold paradox is not unique, steroids are also known to have both immunostimulatory and immunosuppressive effects. Radka et al. reported a series of gold(I)-triphenylphosphine compounds (Figure 2H) with hypoxanthine-derived ligands. These complexes inhibited the secretion of pro-inflammatory cytokines, such as tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β), in the lipopolysaccharide-activated macrophage-like THP-1 cell model (Krikavova et al., 2014). Gold sodium thiomalate (GST, Figure 2B) inhibited release of the key endogenous mediators of HMGB1 translocation, IFN-beta and NO, thus reducing the extracellular release of HMGB1 in murine RAW 264.7 and human THP-1 macrophages in vitro (Zetterstrom et al., 2008). The differential effects of gold compounds on the immune system may be related to drug dose, duration, cell lines, ligand composition, as well as the oxidation state of Au. The differences in the patterns of action of gold compounds between inflammation and cancer are not clear. The suppression of cancer-promoting inflammation has been presumed as one of the main mechanisms in the antitumor activity of gold compounds (Madeira et al., 2012). The anti-inflammatory and anticancer activities of gold compounds may involve similar cytokines and molecular pathways but the degree of induction varies (Yamamoto and Gaynor, 2001). In addition to the inhibition of TLR signaling and NF-κB signaling pathway mentioned above, auranofin also reduces TNF-α synthesis and secretion, decreases STAT-3 transcription activity as well as inhibits angiogenesis (Kim et al., 2007; Han et al., 2008a), which are related to tumor growth and development. Pro-inflammatory cytokines like interleukins (IL) are wildly known to be related to malignant progression and metastasis in multiple types of tumors, by regulating the expression of matrix metalloproteinases (MMP) and angiogenic proteins growth factors including VEGF (Quail and Joyce, 2013). Auranofin and the heterobimetallic Ru-Au compound (RANCE-1) have a strong inhibition of several cytokines (IL-6, IL-5, IL17A, and IL-8) in Caki-1 renal cancer cells (Elie et al., 2018). Auranofin also decreased the production of nitric oxide (NO) as well as the pro-inflammatory cytokines (TNF-α, IL-1β and IL-6) in macrophages (Han et al., 2008a). The development and evolution of tumors are associated with a series of inflammatory pathways. The blocking of several inflammatory pathways associated with tumor development has a promising potential to promote its anticancer activity, although the direct effects of gold compounds on innate immune cells have been little studied. Given the promising anticancer activity of gold compounds in many types of tumors, the role of immuno-regulatory and anti-inflammatory effects needs to be further investigated at the experimental level.

Figure 4 illustrates various aspects of the effect of gold compounds on the adaptive immune system. Gold compounds can enhance the antigenicity of cancer cells. The body's adaptive immune system is able to identify “non-self” antigens through MHC class I presentation of mutation derived immunogenic neoantigens or viral peptides (in virus-induced tumor) by the malignant cells. This is closely related to anticancer metal chemotherapy drugs which are believed to have mutagenesis in many cases. Indeed, the increased number of mutations as well as translocations would support adaptive immune system to identify “non-self” malignant neoantigens (Siniard and Harada, 2017). Although the interaction between gold-based drugs and DNA is weaker than that of platinum, they also induce some mutations. A panel of new Au (III) complexes with pincer type ligands (Figure 2G) had moderate binding affinity with calf thymus DNA (CT DNA), and molecular docking showed that they interacted with DNA by insertion (Radisavljevic et al., 2018). By a joint ESI MS and X ray diffraction (XRD) methods, the dicarbene gold(I) compound [Au(9-methylcaf-fein-8-ylidene)2] BF4 (Figure 2I) had a tight binding to Tel 23 DNA G-quadruplex (telomeric nucleic acid sequences with rich guanines) and formed stable adducts (Bazzicalupi et al., 2016). The complex interactions between gold compounds and DNA, including insertion, covalent binding, and even unobserved ways of acting, generate a number of damaged DNA cells and even induce the DNA mutation when the self-repair mechanisms dysfunction, thus exposing more neoantigens. According to a research by Kazuo and his co-workers, pretreatment with Au(III) complex of a model protein antigen (bovine ribonuclease A) induced novel antigen peptide recognized by CD4 ⁺ T cells (Takahashi et al., 1994) (Table 2).

Interestingly, gold compounds also stimulate the activation and proliferation of T as well as B cells (see Table 2). The mice treated with organic gold compounds showed immune stimulation, which was manifested as plaque-forming cells, rosette-forming cells, and serum antibody elevation (Measel, 1975). Some studies indicated that aurothiopropanolsulfonate (Figure 2J) could also enhance the antigen-specific IgE immune response in mice (Kermarrec et al., 1996). Gold, in the form of natrium aurothiomaleate (GSTM), had a strong B cell-stimulating effect, including T helper 1 (Th1)- and Th2- isotypes (Havarinasab et al., 2007). Moreover, Walz et al. found that the stimulatory effects of auranofin and GST on cell-mediated immunity as evidenced by oxazolone-induced contact sensitivity as well as delayed hypersensitivity to sheep red blood cells (DH-SRBC). Their results showed that gold in the form of auranofin was approximately 4 times more effective in stimulating cellular immunity than gold in the form of GST (Walz and Griswold, 1978). Gold salt (HAuCl4) stimulated CD4 ⁺ T and CD8 ⁺ T cell activation as well as promoted the secretion of IFN-γ in rats (Savignac et al., 2001). What's more, disodium aurothiomalate interfered with antigen presentation and CD4 ⁺ T cell activation by binding peptides containing cysteine residues (Griem et al., 1995). GST preferentially inhibited B cells at much lower concentrations than required for T cells interference (Hirohata, 1996).

Furthermore, it is well-known that cancer cells tend to reduce MHC class I expression to prevent tumor-specific CTL clone cells from recognizing neoantigens or “tumor-associated antigens” (TAA) during immune evasion. Several researches have shown that metal-based treatment (like platinum) might increase the expression of MHC class I in cancer cells (Ohtsukasa et al., 2003; Gameiro et al., 2012). To our surprise, there is little research with regard to the specific effects of gold compounds on MHC expression. It's reported that GST stimulated the up-regulation of MHC class II expression at low concentration, but it was inhibited at high concentration (Sanders et al., 1987). Auranofin inhibited MHC class I and MHC class II-restricted antigen presentation in dendritic cells of mice (Han et al., 2008b). In addition, various papers have showed that metal nanoparticles (NPs) including gold, silver, nickel, and iron oxide are able to increase the immunogenicity of antigens (Niikura et al., 2013; Orlowski et al., 2018). A study by Piotr and his co-workers demonstrated that tannic acid-modified silver and gold nanoparticles (TA-Ag/AuNPs) stimulated DCs maturation, TLR9 expression and memory CD8+T cells activation, and helped to overcome virus-induced suppression of DCs activation by up-regulation of MHC II and CD 86 expression (Orlowski et al., 2018). We speculate that gold compounds can also enhance antigen presentation by up-regulating MHC expression, which requires more efforts to explore. All of this suggests that gold is a promising immune-regulator whose multiple effects on the adaptive immunity remain to be discovered (Table 2).

It is currently accepted that, in certain settings, chemotherapy drugs are able to activate the entire anticancer immune cycle and even cause a persistent immunological anticancer memory by inducing tumor cells death. This ideal form of tumor chemotherapy-induced cell death is known as immunogenic cell death (ICD) (Englinger et al., 2019). The conception of ICD was first proposed in 2005 in the context of antitumor chemotherapy, and based on the observation that mice colon tumor cells with anthracycline doxorubicin treatment in vitro were able to induce an effective anticancer vaccination reaction that inhibited the growth of inoculated cancers or caused the regression of the established tumor (Casares et al., 2005). Under normal physiological conditions, the process of apoptosis is immunologically “silent” and includes an efficient engulfment by phagocytes to prevent the release of intracellular components that activate inflammation and autoimmune responses. In contrast, ICD in cancer therapy induces immune-stimulatory rather than immunosuppressive reaction. ICD is able to reverse several crucial aspects of immune evasion, thus re-inducing the immune recognition of tumor cells. Generally, apoptotic cancer cells expose “find me” and “eat me” signals to attract innate immune cells (especially DCs), leading to the phagocytosis of apoptotic cells and antigen presentation. The initiation of ICD is based on several different molecular mechanisms in the dying tumor cells, mainly by endoplasmic reticulum (ER) stress- and autophagy-mediated release of DAMP molecules, including the protein chaperones calreticulin (CRT), the nucleotide ATP, DNA damage sensitizer HMGB1, CXCL10 as well as HSP70/90 (Terenzi et al., 2016).

The generation of ER stress mediated by ROS is the core of ICD induction, thus leading to the classification of ICD inducers into two categories. Type I ICD inducers exert multiple cytotoxic effects and trigger ER stress as a secondary mode of action, while type II inducers mediates ROS-related ER-stress as the main mechanism of action (Garg et al., 2015). By far most of the ICD inducers used in clinical anticancer chemotherapy belong to type I class, such as anthracyclines and mitoxantrone, the glycopeptide bleomycin, the alkylating agent cyclophosphamide, bortezomib, as well as the metal-based agent oxaliplatin (Tesniere et al., 2010; Kroemer et al., 2013; Sun et al., 2019; Zhou et al., 2019). Although type II ICD inducers are rare, they have recently been identified in preclinical metal-containing compounds (Wong et al., 2015). Although some similarities between oxaliplatin and cisplatin, it is surprising that only the former has been reported as a true ICD inducer (type I) (Tesniere et al., 2010). The specific relationship between the ICD-inducing capacity of metal complexes and their structure and mode of action is unclear. Interestingly, multiple antitumor metal drugs, such as gold (Marmol et al., 2017; Huang et al., 2018), copper (Bortolozzi et al., 2014; Yang et al., 2017), iron (Liu and Connor, 2012; Kim et al., 2016), arsenic (Chiu et al., 2015), ruthenium (Flocke et al., 2016; Jayanthi et al., 2016), osmium (Suntharalingam et al., 2013), and iridium (Cao et al., 2013) complexes have been shown to activate a number of ICD markers involving ROS production, unfolded protein response (UPR) and ER stress. Considering the overexpression of the ROS, protein damage, UPR and ER stress also as mechanisms of action of gold compounds, it is reasonable to believe that there are more types I and II gold-based ICD inducers. Auronafin induced a concentration-dependent increase of cellular ROS in human lung cancer cell lines (A549) (Hou et al., 2018). Alkynyl gold(I) complex was able to disrupt mitochondrial normal function and induced the production of ROS, which triggered necroptosis in the colorectal adenocarcinoma (Caco-2) cell line (Marmol et al., 2017). Organometallic gold(III) complexes induced ER-stress-related apoptosis as well as pro-death autophagy in lung cancer (A549) cells, allowing lower toxicity and better antitumor activity in murine tumor model comparing with cisplatin (Huang et al., 2018). It has been reported that accumulated lethal DNA double strand breaks due to the increase of ROS render ovarian cancer cells more sensitive to auranofin in a BRCA1-deficient background. Furthermore, the antioxidant N-acetyl cysteine (NAC) protected BRCA1-deficient cells from auranofin-induced DNA damage and apoptosis (Oommen et al., 2016). Strikingly, the ICD induction ability of the experimental and clinically used gold compounds has not been systematically and fully identified so far.

Immune checkpoints can negatively regulate T-cell immunity. The discovery of immune checkpoint inhibitors has opened up new clinical possibilities for anticancer immunotherapy. Among them, programmed death receptor 1 (PD-1) and cytotoxic T lymphocyte antigen 4 (CTLA-4) are the most commonly studied. PD-1 is mainly expressed on the surface of activated T cells, B cells, as well as NK cells. The activation of PD-1 suppresses the phosphatidylinositol 3-kinase (PI3K)/Akt signaling pathway, leading to the inhibition of survival and proliferation of T cells (Pedoeem et al., 2014). PD-1 has two main ligands: PD-L1 and PD-L2, which are involved in inducing T-cell exhaustion. Targeting PD-L1 in some early clinical cancer studies has shown beneficial effects, which has been welcomed by researchers. PD-L1 is widely expressed and can be detected in both hematopoietic cells and non-hematopoietic healthy tissue cells. Numerous studies have confirmed that the expression of PD-L1 gene is controlled by inflammatory signaling. Many soluble factors secreted by immune cells have been described as inducers of PD-L1 in the past few years. INF can regulate the expression of PD-L1 in many types of tumors and immune cells, as well as healthy tissues (Brown et al., 2003). The binding of IFN-γ to its receptor activates the classical JAK/STAT signaling pathway, inducing an increase in the expression of a series of transcription factors, known as interferon-response factors (IRFs). Among these factors, IRF1 is prerequisite to the IFN-γ-mediated upregulation of PD-L1 (Lee et al., 2006). In addition, studies have reported that the up-regulation of PD-L1 depends on TLR4/STAT1 pathway, while the expression of PD-L2 depends on IL-4/STAT6 pathway. As discussed above, it makes sense for immunotherapy considering that STAT signaling is key to activating immune checkpoint molecules such as PD-L1 (Loke and Allison, 2003). It was reported that auranofin inhibited IL-6-induced activation of JAK2/STAT3 signaling pathway and activation of NF-κB in human multiple myeloma cell line (U266, RPM8226, and IM9). The phosphorylation of STAT3 was inhibited by auranofin through IL-6, leading to down-regulation of the anti-apoptotic proteins Mcl-1 and apoptosis of myeloma cells (Nakaya et al., 2011). In some similar experiments, auranofin also blocked the IL-6-mediated JAK1/STAT3 signaling pathway in HepG2 human hepatoma cells and primary cells (i.e., human umbilical vein endothelial cells, fibroblast-like synoviocytes and rat astrocytes) (Kim et al., 2007). In human breast cancer cells (MDA-MB 231), STAT3 phosphorylation and telomerase activity were also decreased by auranofin, but N-acetyl-L-cysteine (a scavenger of ROS) pretreatment restored STAT3 and telomerase activity (Kim et al., 2013). What's more, in a recent study, Raninga et al. for the first time found that auranofin could increase CD8 ^+Ve T- cells tumor infiltration and upregulate the expression of PD-L1 by an ERK1/2-MYC-dependent manner in mice models of triple-negative breast cancer (TNBC). Their study provides a novel combination of cancer therapy using auranofin in combination with anti-PD-L1 targeting therapy, which has limited clinical efficacy in TNBC patients with monotherapy (Raninga et al., 2020). As far as we know, gold derivatives have no effect on other immune checkpoints. Although there is very little research on the direct effects of gold compounds on immune checkpoints, the evidence above suggests a close relationship between gold compounds and the regulation of immune checkpoints.