The aim of immunotherapies is to enable the immune system to fight cancer on its own. Drugs known as checkpoint inhibitors are already in clinical use for this purpose.
However, they are only effective in about one-third of patients. Based on analysis of human tissue samples, a team from the Technical University of Munich (TUM) has now discovered one reason why this is so:
An inactive receptor in cancer cells prevents the drugs from reactivating the immune system.
An overactive immune system can be nearly as dangerous as an inactive one, triggering inflammation that attacks the body’s own tissues.
To counter this, the immune system has what are known as checkpoint molecules, which, when switched on, act like a brake on the immune system.
However, cancer cells can exploit this mechanism: By switching on their checkpoint molecules, they are able to elude attacks by the immune system.
The weakened immune response that results is then no longer robust enough to fight off the cancer cells.
A new approach to cancer therapy therefore involves the use of drugs known as checkpoint inhibitors.
These substances release the “brake” applied by cancer cells, thus restoring the immune system’s ability to combat cancer.
Checkpoint inhibitors are already being used successfully in skin cancer and many other malignancies.
The RIG-I receptor is a key factor
“Unfortunately, checkpoint inhibitors aren’t effective in all patients.
Thanks to our recent study, we now understand why that is the case in some forms of cancer, and using experimental approaches, we have even been able to reverse the situation,” says Dr. Simon Heidegger, researcher in Medical Unit III at TUM and lead author of the paper published in Science Immunology.
RIG-I is a receptor protein that is known to play a role in the body’s defense against viruses.
Heidegger and a team headed by his colleague Dr. Hendrik Poeck have now discovered that RIG-I also plays a key role in cancer control.
In a number of mouse models for skin, pancreatic and bowel cancer, they showed that mice with cancer cells in which RIG-I is active responded much better to checkpoint inhibitors than mice with RIG-I-inactive cancer cells.
Fortunately, a drug already exists that activates RIG-I, and it is already undergoing initial tests in clinical trials with humans.
The team has used it successfully in mouse models. Mice that received the drug responded significantly better to treatment with checkpoint inhibitors.
Human skin cancer samples confirm the findings
The team then studied around 450 tissue samples from skin cancer patients to determine what effect RIG-I activity in the patients’ cancer cells had on their survival.
In cases where RIG-I was active, the patients lived significantly longer despite their cancer.
The team showed in 20 tested individuals that such patients also responded better to treatment with checkpoint inhibitors.
The researchers now plan to confirm their findings in large-scale trials on patients.
“We hope to be able to use RIG-I as a marker as well for predicting how well a patient is likely to respond to therapy.
This would avoid unnecessary treatments,” says Heidegger. In addition, they plan to test drugs that activate the RIG-I signal pathway in other mouse models and to investigate the effects of additionally administered checkpoint inhibitors.
The RIG-I like receptors (RLRs), including RIG-I, MDA5, and LGP2, detect viral infections and initiate interferon-dependent and -independent antiviral immune responses (1, 2). RIG-I is activated by the binding of an RNA substrate containing 5′-triphosphorylated short double-stranded RNA (dsRNA), although the absolute requirement for recognition is the basic duplex RNA. In contrast to RIG-I, MDA5 is activated by long double-stranded RNA (1, 3).
RIG-I recognizes viral RNA due to the presence of the triphosphorylated 5′ end, distinguishing replicating viruses from endogenous RNA that is further processed with the addition of a 5′ cap (4). In single-stranded RNA (ssRNA) viruses, the partially complementary, panhandle-structure terminal sequences are recognized by RIG-I (5,–8). In addition to 5′-triphosphorylated RNA, RIG-I also binds to 5′-diphosphorylated RNA and Cap 0 RNA (9).
Upon binding of RNA to RIG-I, the activated RIG-I binds to MAVS (alternative names are IPS-1, Cardiff, and VISA). This leads to the activation of transcription factors IRF3/IRF7 and NF-κB, which trigger the production of type I interferon (IFN) and other antiviral mechanisms (10,–13).
Type I interferons (IFN-α and IFN-β) bind to interferon-α/β receptor (IFNAR) on cell surfaces to induce JAK-STAT signaling and phosphorylation of STAT1 and STAT2 in an autocrine and paracrine manner.
Complexes of phosphorylated STAT1, STAT2, and IRF9 enter the nucleus and induce the production of interferon-stimulated genes (ISGs). MX1 is an interferon-induced protein, and its promoter is used in our study as a luciferase reporter system to quantify the amount of IFN (14,–16).
The use of RLR-binding molecules has been proposed for antiviral prophylaxis and treatment, as cancer therapy, and as vaccine adjuvants (17, 18). Hairpin RNA molecules between 67 nucleotides long (19) and 99 nucleotides long (20) were shown to have broad antiviral activity against influenza virus, dengue virus (DENV), and chikungunya virus (CHIKV) when tested in human cell lines and in mice (2). Short 5′-triphosphorylated hairpins (10 to 14 bp) have recently been demonstrated to be able to activate RIG-I mediated production of type I interferon in mice (21, 22). The antiviral activity was mediated by the induction of antiviral programs in the cells, including the production of IFN-α and -β.
DENV is an arbovirus that is transmitted to humans through the bite of an infected Aedes mosquito. DENV is part of the Flaviridae family and is a member of the Flavivirus genus.
This family of viruses includes other viruses that are known to pose health threats to the human population globally, including yellow fever virus (YFV), West Nile virus (WNV), and Japanese encephalitis virus (JEV). DENV is an enveloped virus that contains a single-stranded, positive-sense RNA genome. This viral genome encodes a large polyprotein, which is processed by viral and host proteases into three structural proteins (capsid, prM, and envelope protein) and seven nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5).
The transmission of DENV involves the transfer of virus from the saliva of the biting mosquito to the dermal layer of human skin (23). The outermost, epidermal layer contains keratinocytes and Langerhans cells (LCs), which are skin-resident antigen-presenting cells (APCs) that are involved in detecting pathogens that penetrate the skin barrier (24).
The dermal layer, which is located below the epidermal layer, consists of fibroblasts and immune cells, including macrophages, T cells, and dendritic cells (DCs), and is innervated with blood and lymphatic vessels that enable immune cell migration to draining lymph nodes (25).
APCs are primary host cells for DENV infection (23, 26,–29). Professional APCs in the skin are particularly important in the establishment of infection due to their location at the point of virus entry into the host (23, 27, 29). We have established a human skin cell assay as a model to study DC subset infection and activation in vitro (23).
These primary skin cells are different from the conventionally used monocyte-derived dendritic cells, which are more representative of an inflammatory type of APCs and are not relevant as initial hosts. Instead, monocyte-derived dendritic cells are secondary infection targets once the infection is established (23, 29). Upon DENV infection, APCs are activated by the viral RNA binding to RIG-I and MDA5 in the cytoplasm of these cells (3).
Based on the initial work to determine the minimal RNA ligand required for interferon activation (21), we made various modifications to the original sequence and tested the ability of these newly designed immune-modulating RNAs (immRNAs) to activate the RIG-I-mediated innate immune response in host cells. We found a lead candidate immRNA, 3p10LG9, that has greater potency in activating type I interferon response than the parental construct, and we studied the protective effects of this immRNA against DENV infection both in human cell lines and in a human skin cell assay model to assess its potential as a prophylactic and therapeutic molecule.
RIG-I-like receptors (RLRs) have been known to play an important role in sensing viral infection and activating antiviral immune response, including the production of type I interferon and proinflammatory cytokines (1, 3).
Through an initial screen we found that a kink in the 9th nucleotide from the 5′ end generated by the addition of a guanine nucleotide enhanced type I interferon activation 6-fold (Fig. 1). It has been shown by others that various structural modifications to 5′-pppRNA were able to enhance RIG-I-mediated activation of type I interferon and antiviral activity. In particular, longer sequences as well as the addition of poly(U) sequences along the stem of the RNA duplex has been shown to enhance immune activation (20, 21).
An assessment of differential binding of 3p10L and 3p10LG9 to RIG-I using hydrogen-deuterium exchange coupled to mass spectrometry (HDX-MS) was carried out and will be published separately. The findings suggest that a stronger RNA-protein interaction leads to more exposed caspase activation and recruitment domains (CARDs) and thus increased RIG-I mediated signaling with 3p10LG9 (H. Y. Yong et al., submitted for publication). Our findings, together with the findings from others, suggest that the structural and sequential features of the RNA species play an important role in activating type I interferon responses in host cells. This knowledge can be used to further improve the drug-like properties of the RIG-I ligands.
Proof of activity of immRNA in human cells is crucial for a potential therapeutic application. We had previously established a human skin cell assay as a model to study the infection of various DC subsets in vitro (23). We reported that the main DC subsets susceptible to DENV infection were the CD11c+ dermal DCs, CD14+ dermal DCs, and Langerhans cells. While all skin DCs were receptive to transfection with immRNA, uptake by CD14+ cell was most efficient. The CD14+ DDC population is transcriptionally and functionally related to human monocytes and macrophages, which have a higher phagocytic capacity than conventional DCs (30). Phagocytosis activity can possibly improve the transfection efficacy. We showed that human skin cells pretreated with immRNA were effectively primed through type I interferon production and upregulation of interferon-stimulated genes (ISGs). This resulted in the inhibition of DENV replication for the three virus-susceptible DC subsets. Using U937-DC-SIGN cells, we demonstrated that the immRNA-mediated antiviral effect was both RIG-I and type I interferon dependent (Fig. 2 and and3).3). We also showed that the lack of MDA5 did not significantly attenuate the IFN activation by immRNA, suggesting that type I IFN signaling induction was MDA5 independent. These findings are in line with previous results showing that minimal-length dsRNA molecules bind to RIG-I to initiate type I interferon signaling (21, 31, 32).
Therapeutic treatment appeared to only moderately decrease DENV replication in Langerhans cells but not in the other skin APC subsets. LCs are located in the epidermis, the most superficial layer of the skin. It is still unclear to what extent Langerhans cells come into contact with the virus once the host is bitten by an infected mosquito, since the probing for blood vessels is in the dermis (23, 27, 33). Regardless, our results suggest that Langerhans cells are possibly more sensitive to RIG-I-mediated innate immune activation than CD11c+ DDCs or CD14+ DDCs after DENV infection has been established. More work is required to determine the factors involved in the responsiveness of Langerhans cells to RIG-I-mediated immune activation. One limitation of the assay is the relatively high MOI (5) required to achieve detectable levels of DENV E protein by flow cytometry. More-sensitive approaches for virus detection, such as viral RNA sequencing on sorted DC subsets, could give us better insight into the therapeutic effects of immRNA and RIG-I-mediated immune activation, as well as determining the factors involved in the responsiveness of Langerhans cells to RIG-I-mediated immune activation.
The concentration-dependent activity of 3p10LG9 observed in U937-DC-SIGN cells could indicate negative feedback inhibition of interferon signaling as a result of overstimulation at high immRNA concentrations. Using a mouse model with transgenic expression of picornaviral RNA-dependent RNA polymerase (RdRP), Painter et al. (34) found that the interferon-stimulated genes (ISGs) were up to 300-fold elevated in mouse tissues and that this elevated ISG profile protected RdRP mice from viral infection. Interestingly, these RdRP mice were healthy with normal longevity despite life-long, constitutive MDA5-mediated innate immune system activation caused by the presence of endogenous long dsRNA. Genes involved in the negative regulation of type I IFN signaling, such as USP18, NLRC5 and LGP2, were upregulated in the gene expression data (34). In addition to the results with U937-DC-SIGN cells in our study, a potential negative regulation of the antiviral effects after the establishment of a viral infection was observed for primary cells. CD11c+ DCs showed a trend for higher infection when treated with immRNA only 24 h after infection (Fig. 7). This time- and concentration-dependent negative feedback loop could be important to limit inflammation and cell death. Accordingly, no cell death was observed in the U937-DC-SIGN cells, even at the highest immRNA concentration. Similarly, primary skin APCs were unaffected by high concentrations of immRNA. Death receptor signaling-related DEGs were not among the top hits in the RNAseq analysis (Fig. 5F). This was different for the A549 fibroblast cell line, which did not survive transfection with immRNA at high concentrations. The induction of apoptosis and cell death is similar to what others have described when using a RIG-I agonist in A549 cells (2, 19). The wider active window of 3p10LG9 compared to 3p10L shown in U937 cells (Fig. 1) could be a key advantage for a potential therapeutic application.
It was surprising that 3p10LG9 appeared to activate skin APCs more efficiently than poly(I·C) (Fig. 5E), given that poly(I·C) can bind to multiple receptors (RIG-I, MDA5, and TLR3), whereas 3p10LG9 binds only to RIG-I. However, it is difficult to directly compare the efficacy of short RNA molecules like 3p10LG9 with that of poly(I·C), given the large difference in their molecular weight and potential differences in transfection efficiency. Nevertheless, it is worth noting that 10 nM 3p10LG9 is equivalent to a concentration of about 0.08 μg/ml, much less than the concentration of 0.5 μg/ml poly(I·C) used in the cell line experiments (Fig. 2 and and3).3). Given the key role of RIG-I signaling activation at the interface of innate and adaptive immune responses (35, 36), RIG-I signaling in tissue-resident APCs as a physiologically relevant model should be further studied.
In summary, we have shown that the minimal RNA ligands are capable of generating an effective innate immune response in host cells with natural infection, and this response inhibits DENV replication in primary cells efficiently when used as a prophylaxis. Beyond dengue infection, our findings could be relevant for topical or systemic application of RNA-based ligands targeting RIG-I and for the ensuing responses in general.
More information: Simon Heidegger et al, RIG-I activation is critical for responsiveness to checkpoint blockade, Science Immunology (2019). DOI: 10.1126/sciimmunol.aau8943
Journal information: Science Immunology
Provided by Technical University Munich