Evolutionary immunology to explore original antiviral strategies

Abstract

Over the past 25 years, the field of evolutionary developmental biology (evo–devo) has used genomics and genetics to gain insight on the developmental mechanisms underlying the evolution of morphological diversity of animals. Evo–devo exploits the key insight that conserved toolkits of development (e.g. Hox genes) are used in animals to produce genetic novelties that provide adaptation to a new environment. Like development, immunity is forged by interactions with the environment, namely the microbial world. Yet, when it comes to the study of immune defence mechanisms in invertebrates, interest primarily focuses on evolutionarily conserved molecules also present in humans. Here, focusing on antiviral immunity, we argue that immune genes not conserved in humans represent an unexplored resource for the discovery of new antiviral strategies. We review recent findings on the cGAS-STING pathway and explain how cyclic dinucleotides produced by cGAS-like receptors may be used to investigate the portfolio of antiviral genes in a broad range of species. This will set the stage for evo–immuno approaches, exploiting the investment in antiviral defences made by metazoans over hundreds of millions of years of evolution.

This article is part of the theme issue ‘Sculpting the microbiome: how host factors determine and respond to microbial colonization’.

1. Evolution to understand antiviral immunity: lessons from evolutionary development biology

As we have been reminded with the recent COVID-19 crisis, viruses represent a major threat for our societies. In humans, acute or chronic viral infections are associated with many life-threatening diseases, including cancer. Innate immunity is the first line of defence that operates in all animals and, in the case of vertebrates, precedes and orients the establishment of adaptive immunity. Host cells have evolved numerous innate defences against viral infections, but the control of viruses is complicated by the high mutation rates of most viral polymerases, which promote rapid virus evolution and adaptation to antiviral mechanisms. As a result, animal genomes have been shaped by continuous challenge from viruses and the repertoire of antiviral genes in any species reflects the cumulative effects of millions of years of evolutionary investment in innate immunity defences [1,2]. At a time when we have been reminded that (i) we cannot predict what the next viral human pathogen will be and (ii) the risks of future zoonosis outbreaks are higher than ever, increasing the diversity of approaches to understand virus–host interaction should be a priority [3,4]. Studying these interactions in as many settings as possible will provide insight on a wide range of host restriction factors, opening the way to new applications in biomedicine.

The concept of evo–immuno—evolutionary immunology—is inspired by the spectacular successes of the field of evolutionary developmental biology, or evo–devo, in the past 30 years. The evo–devo approach exploited genomics and genetics to gain insight on the developmental mechanisms underlying the evolution of morphological diversity in animals [5,6]. The crucial discovery that homeobox (Hox) genes control antero-posterior patterning in both flies and humans provided the first insight into a genetic toolkit shared for the development of morphologically very different animals and paved the way for the evo–devo field [7,8]. A key aspect of the genetic theory for the evolution of animal morphology revealed by evo–devo and apparent in the hourglass scheme of Denis Duboule is that a conserved gene toolkit (e.g. Hox genes) responds to novelty and regulates novelty, hence providing adaptation to changing environments (figure 1) [7,8]. Indeed, the ontogenic trajectory is often influenced by environmental factors, and innovations in morphogenesis are generally selected if they improve adaptation of the animal to its environment or if they facilitate access to a new niche [9]. Besides embryonic development, the immune system is another facet of animal biology that is forged by interactions with the environment (e.g. the microbial world) and could benefit from evolutionary perspectives [10,11]. Of note, the model organism Drosophila melanogaster, with its extensive genetic and molecular resources, played a central role in evo–devo and other insects provided spectacular examples of emergence of striking morphological novelties from the ancestral conserved toolkit (e.g. [12–14]).

Figure 1.

Parallel between evo–devo and evo–immuno. The STING pathway can be used to probe original antiviral strategies, much like homeobox genes were used to decipher evolution of morphology traits in animals. (Online version in colour.)

2. Evolutionary immunology to delve into the biodiversity of antiviral defence strategies in insects

Authors exploring comparative immunology have noted that, while signalling pathways tend to be conserved between species, the receptors sensing infection are less conserved and evolve rapidly, as do the effector molecules regulated by these pathways (e.g. [15–18]). Hence, innate immunity pathways exhibit an information bottleneck, akin to the bottleneck in gene expression corresponding to the establishment of the axis of the embryo during development. Although the evo–immuno concept does not entail an ontogeny aspect, it aims at exploiting central evolutionarily conserved regulatory nodes to gain insight on original solutions to adapt to environmental challenges, following in this regard the path of evo–devo (figure 1). As argued elsewhere in this issue [19], all microbes have specific weaknesses, which can be exploited by host defence mechanisms and thus drive evolution of immune systems. Evo–immuno aims at identifying evolutionarily exploited weak spots in pathogens, which could point to innovative therapeutic strategies.

We propose to explore the concept of evo–immuno in insects, taking advantage of the Drosophila model, but also of the fantastic biodiversity of this class of animals with more than 1.2 million species known and well-established phylogenetic relationships [20]. Insects are present and play vital roles in all terrestrial ecosystems, where they are exposed to a broad range of pathogens, including viruses. Indeed, recent virome analyses have revealed that insects represent an impressive reservoir of viruses, including most major viral groups found in animals and plants (e.g. poxviruses, flaviviruses, alphaviruses, rhabdoviruses, picorna-like viruses, etc.) [21–23]. Furthermore, important human viruses are transmitted by hematophagous insect vectors such as Aedes mosquitoes [24]. Hence, characterizing antiviral immunity in insects may reveal original strategies of antiviral defences. Importantly, evo–immuno aims at understanding how a conserved gene toolkit produces a diversity of non-conserved responses, just as evo–devo allowed us to understand how a conserved toolkit can build legs, wings, or fins at locations of the body. Hence, an important aspect of evo–immuno is that the study of non-conserved genes is as interesting as—or even, arguably, more interesting than—evolutionarily conserved ones. This represents an important paradigm shift as, until now, studies on insect immunity received much more support and publicity when they reported evolutionarily conserved mechanisms (e.g. Toll receptors) [25]. Yet, investigating non-conserved genes holds much promise in the context of antiviral immunity. Restriction of a virus by a host factor will put pressure on the virus to adapt, often resulting in the emergence of resistant variants. This will in turn put pressure on the host to either modify the restriction factor to recover interaction with its viral target, or to find another solution to control the virus [26]. Because this continuous evolutionary arms race between two genetic entities antagonizing each other goes on permanently, the repertoire of antiviral factors is evolving rapidly [2,27]. Central to the concept of evo–immuno is the notion that this arms race goes on in parallel in all animals, such that each animal may find its own unique solution to counter a virus (figure 2). Thus, exploring the genomes of insects for species-specific innovations in antiviral immunity may reveal novel antiviral strategies.

Figure 2.

Evolutionary arms race between hosts and viruses. When confronted by rapidly evolving viruses that escape host restriction factors, host adapts either by modifying the ancestral restriction factor (right, deep blue) or by finding another target in the virus. As a result, each animal may exhibit lineage-specific antiviral restriction factors. Modified from Daugherty & Malik [26]. (Online version in colour.)

Although antiviral immunity in insects has been a focus of attention in the past years, studies have focused on few species (flies, vector mosquitoes, honeybees) and specific viruses (e.g. [28–31]). Broader investigations have been hampered by lack of information on the viruses infecting insects, which could be used to trigger antiviral immunity. The discovery of the important role played by the Stimulator of interferon genes (STING) pathway in the antiviral defence of D. melanogaster opens new perspectives [32]. Indeed, STING-dependent signalling can be activated in vivo by injection of cyclic dinucleotides [33–36]. This provides a powerful means to trigger antiviral immunity in a range of insects, paving the way for evo–immuno (figure 1).

3. STING signalling participates in insect antiviral immunity

In the course of our work on induced antiviral responses in D. melanogaster [37–40], we discovered that two components of the antibacterial immune deficiency (IMD) pathway, the kinase IKKβ and the NF-κB factor Relish, rather than the pathway as a whole [32], participated in the control of infection by two picorna-like viruses, Drosophila C virus (DCV) and Cricket paralysis virus (CrPV). It is worth reminding here that the characterization of Diedel, a gene not conserved in mammals but strongly induced in response to viral infection in D. melanogaster and hijacked by members of several families of large insect DNA viruses, attracted our attention to the possible involvement of components of the IMD pathway in antiviral immunity [41]. This illustrates how investigating non-conserved genes may reveal important evolutionarily conserved facets of immunity. We further identified the orthologue of STING, dSTING, among the genes regulated by IKKβ in the context of viral infections and showed that dSTING was acting upstream of IKKβ and Relish to regulate expression of genes induced by viral infections [32] (figure 3). Similar results were obtained in the silkworm Bombyx mori [42]. Altogether, our findings pointed to the existence of a new pathway activated by viruses and controlling expression of STING-regulated genes to curb viral infection.

Figure 3.

Evolutionary conservation of the cGLR-CDN-STING cassette. Enzymes of the CD-NTase family (cGAS, cGLRs, CdnE), the cyclic oligonucleotides they produce and STING-related molecules participate in the control of phage or viral infections in bacteria and animals, from invertebrates to mammals. The mechanism activating CD-NTase enzymes in prokaryotes in response to phage infections is still largely unknown. Note that as predicted in the hourglass model shown in figure 1, cGLRs and cGAS are not activated by the same type of nucleic acids and regulate overlapping but distinct sets of genes (e.g. no interferons in invertebrates). (Online version in colour.)

In mammals, STING is a key component of the cytosolic DNA sensing pathway [43,44]. It activates the transcription factor IRF3 through the kinase TBK1 to induce interferon (IFN) gene expression, but also regulates IKKβ and NF-κB through a less characterized mechanism [45–47]. STING acts as a signalling receptor for a second messenger, the cyclic dinucleotide 2′3′-cGAMP, which is produced by the enzyme cGAS upon sensing cytosolic DNA [43,44] (figure 3). STING can also be activated by other cyclic dinucleotides directly produced by bacteria (e.g. 3′3′-c-di-AMP, 3′3′-c-di-GMP and 3′3′-cGAMP) [48].

In Drosophila, we showed that injection of cyclic dinucleotides into flies leads to a dose-dependent induction of STING-regulated genes, in a dSTING- and Relish-dependent manner. Of note, 2′3′-cGAMP was a more potent agonist of dSTING than 3′3′-connected cyclic dinucleotides of bacterial origin, suggesting that an enzyme producing 2′3′-cGAMP was present in insects [33]. This discovery led to an analysis of the transcriptome of 2′3′-cGAMP-injected flies, which revealed more than 400 genes stimulated at least 1.5-fold 6, 12 or 24 h post injection. It is worth mentioning here that 2′3′-cGAMP-induced genes include components of the small interfering RNA (Dicer-2, Argonaute 2) and autophagy (Ref(2)P, encoding the homologue of p62) pathways, pointing to interactions between the STING pathway and other antiviral mechanisms [49–54]. Strikingly, co-injection of 2′3′-cGAMP with viruses reduced viral replication and improved the survival of wild-type flies, but not dSTING or Relish mutant flies. This effect was observed on five different viruses belonging to different families (Dicistroviridae, Nodaviridae, Alphaviridae, Rhabdoviridae, Nudiviridae), indicating that 2′3′-cGAMP triggers broad antiviral immunity in D. melanogaster [33]. Altogether, our results revealed that 2′3′-cGAMP triggers a dSTING/NF-κB-dependent antiviral transcriptional response.

4. cGLRs, an emerging family of pattern recognition receptors

Two cGAS-like receptors, cGRL1 and cGLR2, acting upstream of STING in D. melanogaster have recently been identified [35,55]. Although they synthesize cyclic dinucleotides, like cGAS, their characterization revealed intriguing differences from the mammalian enzyme (figure 3) [56]. For one, the activity of cGLR1 in vitro or in transfected mammalian cells depends on the presence of double-stranded (ds)RNA, rather than DNA. The in vitro activity of recombinant cGLR2 from the species D. bipectinata and D. pseudoananassae is also enhanced most strongly in the presence of dsRNA [34]. Another notable difference with cGAS is that Drosophila cGLRs produce at least three cyclic dinucleotides in addition to 2′3′-cGAMP, which can all activate STING signalling in vivo, albeit with different efficiencies (2′3′-c-di-GMP > 3′2′-cGAMP > 2′3′-c-di-AMP > 2′3′-cGAMP) [34].

cGLRs represent an emerging family of pattern recognition receptors (PRRs) present in all metazoan phyla and producing an array of cyclic di-purine and purine-pyrimidine signals in response to binding DNA, RNA and probably also other signals [57]. The biological significance of this diversity of cyclic dinucleotide products is still unclear but could reflect the existence of alternative receptors for these nucleotide signals. For example, the stony coral Stylophora pistillata encodes 42 cGLRs and 7 STING paralogues, which exhibit different affinity and selectivity for cyclic dinucleotides: Sp-STING3 preferentially binds 3′3′-linked cyclic dinucleotides, while Sp-STING5 is highly selective for 2′3′-cGAMP [57]. Other receptors for cyclic dinucleotides exist in animals, e.g. RECON (reductase controlling NF-κB) in mice [58,59]. Interestingly, flies synthesize an unusual cGAMP isomer, 3′2′-cGAMP, which has so far only been identified in Drosophila and the bacteria Asticcacaulis sp. [34,35,55,60]. This Drosophila innovation may have been driven by a family of viral suppressors of cGAS-STING signalling, the Poxins [35]. Poxins were initially identified in vaccinia virus, a member of the family Poxviridae, and function as 2′3′-cGAMP-specific nucleases [61]. Strikingly, poxin homologues can be found in the genomes of several large DNA viruses (baculoviruses, entomopoxviruses) infecting lepidopteran insects, but also in the genome of the moths and butterflies that host these viruses, pointing to a likely insect origin for poxin genes [62]. Poxin efficiently cleaves 2′3′-cGAMP but fails to cleave 3′2′-cGAMP, suggesting that the isomeric switch in phosphodiester linkage specificity in Drosophila may have occurred to evade its action [35]. It is worth noting here that genomic analysis of different poxvirus genera indicates that Poxins are widely found in insect poxviruses and are present in some bat and rodent poxviruses but are missing in many other vertebrate poxviruses. This suggests that Poxin was acquired by horizontal transfer from insect viruses, possibly favoured by the insectivorous nature of bats and rodents [63]. Overall, these findings illustrate the relevance of taking a broad look at the repertoire of antiviral defences present in animals, rather than focusing on only a few species.

5. The cGLR-CDN-STING cassette in the animal toolkit of antiviral defences

It is now clear that the connection of the STING pathway to the regulation of interferon genes is a late addition to an ancestral pathway, which has been associated with the control of viral infections long before the onset of vertebrates. Indeed, the three key components of the pathway—cGAS-like enzymes, the cyclic oligonucleotides they produce and STING-related molecules—were all inherited from prokaryotes, where they function in the control of phage infections (reviewed in [64]). Accordingly, genes encoding cGLRs and STING are present in the genomes of invertebrates, although they were lost in some species, such as worms or, within insects, mosquitoes. In addition, an array of distinct cyclic dinucleotide signals controlling discrete STING signalling pathways has recently been identified in a set of invertebrates [57]. Although the exact biological function of cGLRs in most animals remains unknown, the function of cGAS/DncV-like nucleotidyltransferase (CD-NTase) enzymes in the control of highly divergent anti-phage defence signalling pathways in bacteria [65]; the conserved function of cGAS and cGLRs in antiviral immunity in flies and mammals [56]; and the fact that most cGLRs can be activated in vitro by nucleic acids [34,35,55,57]—a molecular pattern characteristic of viral infections—altogether strongly suggest that the cGLR/STING signalling axis belongs to an ancestral toolkit associated with the control of viral infections [64]. Therefore, this pathway can be harnessed to gain insight on the diversity of induced antiviral defences in animals, much like the way Hox genes were used for evo–devo (figure 1).

Induction of antiviral gene expression can be achieved by analysing the transcriptome of cells or animals after viral infection. However, it is difficult in these studies to distinguish between the stress reaction triggered by cell lysis or tissue damage and the immune response. The analysis is further complicated by the fact that cell infections are not synchronized when animals are used instead of cell lines, preventing the distinction between immediate early responses and late responses to the infection. Finally, viruses are notorious for hijacking cellular functions and suppressing host defence mechanisms. The image provided by these transcriptomic studies can therefore be imprecise. This caveat can be avoided by stimulating cells with molecules mimicking the molecular patterns sensed by PRRs to trigger antiviral immunity. However, this is not trivial in the case of viral infections, since the nucleic acid mimics known to trigger antiviral immunity have to be delivered into cells, to meet PRRs residing in the cytosol or the endosomes [66]. In this context, the ability to activate STING signalling by injection of cyclic dinucleotides provides a powerful shortcut to visualize the modifications of the transcriptome associated with induction of antiviral immunity, in Drosophila and also other invertebrates [33]. The emerging diversity of cyclic dinucleotide signals produced by cGLRs represents a caveat but should not be limiting. Indeed, although cGLRs from invertebrates can produce several cyclic dinucleotides, 2′3′-cGAMP is produced by cGLRs from (i) several insects (e.g. Drosophila flies [34,35,55], the beetle Tribolium castaneum [35], the thrip Frankliniella occidentalis and the flea Ctenocephalides felis [57]); (ii) the sea anemone Nematostella vectensis [52,67]; (iii) the coral Pocillopora damicornis [57]; and (iv) the fresh-water polyp Hydra vulgaris [57]. Furthermore, as described below, 2′3′-cGAMP can activate STING signalling in many species, including Drosophila flies. If necessary (e.g. lack of significant response to 2′3′-cGAMP), other cyclic dinucleotides may be used, as shown recently for the Drosophila species D. serrata and 2′3′-c-diGMP [34]. Indeed, approaches suitable for chemical synthesis of cyclic dinucleotides with any combination of nucleobases and their purification have been described [68].

6. Cyclic dinucleotide injection provides access to a repertoire of antiviral genes

Injection of 2′3′-cGAMP or other cyclic dinucleotides has now been reported to activate an extensive transcriptional programme not only in Drosophila, but also the sea anemone N. vectensis, the coral S. pistillata, the Eastern oyster Crassostrea virginica and even the marine chanoflagellate Monosiga brevicollis, a free living unicellular and colonial eukaryote that is the closest living relative of metazoans [33,57,69,70]. Strikingly, comparison of the genes differentially expressed in response to CDN injection in these organisms reveals a common pool of immune genes—including transcription factors (e.g. NF-κB, IRF), PRRs (e.g. cGLRs; RIG-I-like receptors; nucleotide binding domain, leucine-rich repeats containing receptors (NLRs)) or effectors (e.g. viperin, OAS, RNaseL)—pointing to an ancestral set of immunity factors that have been conserved over more than 500 Myr of evolution. A number of these genes (e.g. viperin, Argonaute or cGLRs and STING themselves, as mentioned above) have been inherited from the arsenal of antiphage defences in prokaryotes [71]. The function of the mammalian orthologues of several of these factors has already been characterized [72]. However, besides this evolutionarily ancient and conserved programme, many non-conserved genes are also induced in all species. These genes are evidence of divergent evolution in some organisms to adapt to the threat of viruses, which may have resulted in unique strategies to restrict viral replication (figure 2). Examples of these clade-specific innovations include the onset of the IFN family of cytokines in vertebrates, or the fly-specific gene pastrel, which encodes a potent restriction factor for picorna-like viruses and whose expression is induced by 2′3′-cGAMP in all 10 Drosophila species analysed [73–75]. In invertebrates, it appears that, at least in some species (e.g. N. vectensis, M. brevicollis), the STING pathway regulates expression of antibacterial genes, in addition to antiviral genes, which may represent an early innovation in the animal lineage of multicellular eukaryotes to connect the pathway to infectious agents beyond viruses [69,70].

A few years ago, Palmarini, Wilson and colleagues investigated in a pioneer study the ‘type-I interferomes’ of fibroblasts from 10 species of vertebrates and identified a conserved core of 62 IFN-stimulated genes (ISG), thus highlighting the ancestral functions of the IFN response [76]. Notably, this study also revealed that each animal possessed ISGs unique to their species or their phylogenetic lineage. This attests to the constant expansion of ISGs in vertebrates, most likely driven by the constant arms race between host and viruses [2]. Of note, many ISGs are also induced directly upon sensing viral infection by PRRs and predate the evolution of interferons, such that they have homologs in invertebrates and even in prokaryotes [71,77,78]. The possibility to investigate the ‘STINGome’ of invertebrates using the cyclic dinucleotide injection assay now provides the opportunity to investigate evolution of antiviral responses in these animals as well, which offer access to a much broader biodiversity. In particular, with the tools now at hand, insects represent powerful models to identify new restriction factors against families of viruses with broad host tropisms, such as Picornaviridae, Poxviridae, Flaviviridae and Togaviridae.

7. Concluding remarks

It is now acknowledged that most animals control viral infection through the induction of dedicated gene expression programmes. Yet only a fraction of mammalian ISGs have been functionally characterized and recent studies reveal that a substantial number of genes induced by viral infection are not conserved between animals [79]. Furthermore, recent insight on the emerging role of the evolutionarily conserved cGLR/STING pathway positions it as a central component of the toolkit used to fight viral infections in animals beyond vertebrates. Induction of STING-dependent antiviral immunity upon cyclic dinucleotide injection allows rapid identification in different species of large sets of genes regulated in the context of viral infections, paving the way for the exploration of the putative antiviral functions of non-conserved genes. It is noteworthy that, although we focus here on viruses, the concept of evo–immuno also applies to other infections. Indeed, pioneering studies in innate immunity revealed that, although Toll signalling to NF-κB is conserved between insects and mammals, Toll functions as a cytokine receptor in Drosophila, unlike Toll-like receptors, which function as PRRs [80–83]. Furthermore, recent studies indicate that the antimicrobial peptides regulated by NF-κB in different Drosophila species are subjected to diversification and specialization driven by differences in the ecologies of the flies [84].

In summary, the breakthrough discovery that broad antiviral immunity can be triggered in Drosophila and other invertebrates by injection of CDNs provides a rapid and reliable way to access the repertoire of antiviral genes [33–35,57,69,70]. The next step will be to exploit these lists of genes to identify novel restriction factors. This will be challenging, but large-scale biochemical analysis aided by structural predictions [85] and evolution-guided pipelines [86] to select non conserved candidate genes with interesting features, represents a powerful strategy for discovery. Mechanistic studies on the molecular function of these non-conserved genes may lead to the discovery of potentially unique strategies of antiviral defence, revealing discrete weak spots in the targeted viruses. These could be exploited to develop innovative antiviral therapies. As we will continue to be confronted with emerging viruses that ignore phylogeny when they jump into new hosts, it is time to extend the search for antiviral restriction mechanisms beyond humans or mammals and to acknowledge that what is not conserved can be as important as what is conserved, as superbly illustrated by the evo–devo field.

Data accessibility

This article has no additional data.

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors' contributions

J.L.I.: conceptualization, writing—original draft; H.C.: conceptualization, writing—original draft; C.M.: conceptualization, writing—original draft; N.M.: conceptualization, writing—original draft.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

H.C. was supported by the Natural Science Foundation (grant no. 32000662), Guangdong Provincial Science Fund for Distinguished Young Scholars (grant no. 2023B1515020098), Guangdong Provincial Young Scholars academic exchange programme (grant no. 2022A0505030018), Youth Talent Support Programme of Guangdong Provincial Association for Science and Technology (grant no. SKXRC202229) and Science and Technology Programme of Guangzhou (grant no. 202102020090). J.L.I. acknowledges financial support from CNRS, the Hoffmann Infinitus Program and grants from ANR (grant nos ANR-11-EQPX-0022, ANR-22-CE15-0019), IdEx Unistra (grant no. ANR-10-IDEX-0002), SFRI-STRAT'US (grant no. ANR-20-SFRI0012) and EUR IMCBio (grant no. ANR-17-EURE-0023). J.L.I. and H.C. acknowledge support by the Chinese National Overseas Expertise Introduction Center for Discipline Innovation (grant no. Project 111_D18010) and the National Key R&D Programme of China (grant no. 2023YFE010770).