Emerging roles for endogenous retroviruses in immune epigenetic regulation

Summary

In recent years, there has been significant progress toward understanding the transcriptional networks underlying mammalian immune responses, fueled by advances in regulatory genomic technologies. Epigenomic studies profiling immune cells have generated detailed genome-wide maps of regulatory elements that will be key to deciphering the regulatory networks underlying cellular immune responses and autoimmune disorders. Unbiased analyses of these genomic maps have uncovered endogenous retroviruses as an unexpected ally in the regulation of human immune systems. Despite their parasitic origins, studies are finding an increasing number of examples of retroviral sequences having been co-opted for beneficial immune function and regulation by the host cell. Here we review how endogenous retroviruses have given rise to numerous regulatory elements that shape the epigenetic landscape of host immune responses. We will discuss the implications of these elements on the function, dysfunction, and evolution of innate immunity.

Introduction

The mammalian immune response functions to defend an organism against pathogens, and precise regulation of immune responses is essential for effective pathogen responses and to avoid chronic inflammation. Dysregulation of immunity is associated with increased susceptibility to infection¹, autoimmune disorders², and cancer³. It is now well-established that much of the regulation of immune responses is driven by epigenetic changes that control transcriptional changes during an infection^4–6. As a result, there have been extensive efforts to map the epigenomes of immune cells and immune responses, with the goal of identifying specific genomic loci important for healthy or disease-associated immune regulation^7–9.

A major insight to come from these studies has been the revelation that ancient retroviruses are unexpected allies in facilitating the evolution of immune regulatory networks¹⁰. Unlike other viruses, retroviruses integrate into host genomic DNA as part of their replication cycle. Retroviruses transmit as single stranded RNA genomes encapsulated within membrane enveloped particles, and replicate by reverse transcribing their genomes into double stranded DNA, which is then transported into the cellular nucleus and inserted semi-randomly into the host genome by viral integrase proteins¹¹. If a genomic integration event occurs in a germ line cell and is inherited by the host’s offspring, the retrovirus becomes an endogenous retrovirus (ERV) (Fig 1A).

FIGURE 1.

Retroviruses become endogenized and retain potential for regulatory activity. (A) Retroviruses are integrated into host cell DNA during infection. They become endogenous when they are integrated into germline DNA and passed on vertically to progeny. (B) Endogenous retroviruses disperse throughout the host cell genome. ERV families may include hundreds or thousands of individual insertion events. (C) ERVs often lose the ability to transpose and encode proteins through mutational decay, or recombination of LTRs. (D) Examples of ERV epigenetic status. Most ERVs are epigenetically silenced by DNA methylation and repressive histone modifications. Some ERVs have active promoter function, are marked by promoter-associated histone modifications, and provide the transcription start site for mRNA transcripts. Many ERVs have enhancer activity and are marked by enhancer-associated histone modifications and transcription factors. Some ERVs may function as insulators, associated with CTCF binding and are localized at boundaries between epigenetically active and silenced domains. Created with BioRender

The human genome contains over 400,000 sequence fragments derived from retroviral infections that range from 0.5 to >100 mya¹². Over the course of vertebrate evolution, retroviruses have repeatedly invaded and spread within host genomes, dispersing thousands of copies of sequences with regulatory potential (Fig 1B)¹³. New genomic integrations contain the full-length ERV genome including retroviral genes gag, pol, and env flanked by non-coding long terminal repeat (LTR) regulatory regions. New ERVs remain capable of generating new infectious retroviruses, and can generate new germline genomic insertions through either viral reinfection or intracellular retrotransposition^14–17. ERVs can disperse up to thousands of copies in host genomes before eventually accumulating mutations that inactivate their ability to replicate¹⁵. This process of retroviral genomic invasion and amplification has occurred numerous times in every vertebrate lineage, such that ERVs constitute around 6–14% of all vertebrate genomes, but originate from lineage-specific retroviruses^18,19.

Once inactivated, “fossilized” ERVs had initially been assumed to be nonfunctional DNA evolving under neutral selection^20,21. However, there is growing evidence that ERVs are an important source of novel genetic elements co-opted for host biology and evolution. After becoming integrated as part of the host genome, the viral protein-coding regions are rapidly lost through mutation or recombination, due to the deleterious effects of retrotransposition on the host (Fig 1C)²². For example, roughly 90% of ERV loci present in the human genome are represented by “solitary LTRs” which form by LTR-LTR recombination events that excise the internal coding regions. The frequency of solitary LTR events is consistent with these recombination events being adaptive for the host genome.

However, there are several notable exceptions wherein some ERVs have retained some protein-coding capacity and have been repurposed, or co-opted, to serve host functions. One classic example is the repeated co-option of ERV envelope proteins for mediating placental development in mammals^23,24. In the context of immunity, ERV proteins have been co-opted as dominant negatives interfering in viral entry or assembly^25–27 or as other types of regulatory antiviral proteins^28–30. The co-option of ERV derived proteins and RNA has been extensively reviewed elsewhere^31,32, therefore here we will focus on the epigenetic properties of ERVs.

While there are estimated to be several dozen ERVs and transposable elements (TEs) with intact open reading frames in the human genome^33,34, the vast majority of human ERVs have completely lost their coding capacity, and have been assumed to represent inert, nonfunctional DNA. However, in recent years, epigenomic studies have revealed that ERVs are an unexpectedly substantial source of cell type-specific regulatory elements, including promoters, enhancers, chromatin boundary elements, and regulatory RNAs^35–39. These regulatory sequences often originate from the LTR region of the ERV, suggesting that the ancestral regulatory activities of integrated proviral LTRs are readily repurposed for cellular genome regulation. ERV-derived regulatory activity has now been documented in a wide range of tissues and cell types, including many immune cells. These observations have renewed interest in a decades-old hypothesis that ERVs and other repetitive elements are important drivers in the evolution of gene regulatory networks.

While the functional significance of ERV-mediated gene regulation remains largely unknown, there is a growing body of evidence supporting key roles for ERV-mediated regulation in host immune function. This review will discuss how ERVs impact epigenetic regulation in the context of mammalian immunity. We will introduce the functional properties of ERVs and how they impact epigenetic landscapes, and critically evaluate the genomic and experimental evidence supporting ERV regulatory function in immunity. Finally, we will discuss current challenges in understanding the significance of ERV-mediated regulation in human evolution and disease.

Epigenetic properties of ERVs

The retroviral origin of ERVs results in a duality of active and repressive epigenetic patterning. All infectious retroviruses contain non-coding long terminal repeat (LTR) sequences, which contain dense clusters of binding sites that hijack cellular transcription factors and drive transcription of the integrated provirus. As a defense against modern retroviruses, mammalian cells have evolved mechanisms to target and epigenetically silence proviral expression^40,41. Presumably as a counter-adaptation, retroviruses like Human Immunodeficiency Virus 1 (HIV-1) have evolved multiple interferon and NF-κB inducible binding sites that enable proviral transcription to become reactivated under conditions of innate immune stimulation^42–47. This switch in transcriptional activity depends upon epigenetic remodeling of the LTR, where acetylation of nucleosomal activating marks drives reactivation of proviral transcription^48–50. The cellular balance between epigenetically repressed and activated states contributes to stochastic reactivation from latency⁴⁹.

Most ERVs consist of sequences derived from retroviral LTRs, and exhibit similar regulatory properties as those of integrated infectious proviruses⁵¹. Consistent with their ancestral potential for pathological effects, the vast majority of ERVs are, like their exogenous counterparts, targeted for silencing by cellular epigenetic repression factors. ERVs and TEs in general are targeted by retroviral and transposon silencing pathways that deposit repressive epigenetic marks, including DNA methylation and repressive histone modifications H3K9me3 or H3K27me3^52–54, to minimize the adverse effects of rampant transposition on the host genome (Fig 1D). For example, the KRAB-ZNF family of transcriptional repressor proteins have evolved prolifically to bind specific sequence motifs associated with different families of TEs⁵⁵. Binding by KRAB-ZNF proteins results in recruitment of KAP1/TRIM28 and SETDB1 to deposit repressive epigenetic modifications including H3K9me3 and eventually DNA methylation. This targeted repression at ERVs persists long after the elements have lost their coding capacity^56–58. The KRAB-ZNF pathway is probably the most well-characterized pathway underlying epigenetic repression at TEs, but there are multiple additional mechanisms, such as the piRNA and HUSH pathways, that also result in epigenetic silencing at TEs^59,60.

Despite the fact that the majority of ERVs in a cell are in a repressed epigenetic state, there are nonetheless hundreds to thousands of ERVs that show some biochemical evidence of epigenetic or transcriptional activity in mammalian cells^61–65. The epigenetic activation of ERVs, as with silenced modern proviruses, is driven primarily by cellular transcription factors that bind to sequences within the LTR and recruit additional transcriptional machinery^38,63,66. Epigenetically activated ERVs show all the canonical hallmarks of enhancer elements: increased accessibility, enhancer-associated histone marks including H3K27ac and H3K4me3, binding of coactivators p300, and nascent transcription by RNA POL2 (Fig 1D)^37,67,68. Different ERV families harbor different repertoires of transcription factor binding sites, resulting in different patterns of cell type-specific or context-dependent activation⁶², and individual insertions of a particular family may be mutated, duplicated, or truncated to expand or reduce their capacity for activation. These genomic findings ran contrary to the expectation that most ERVs represented silenced parasitic sequences, and implicated a role for ERVs and TEs in the evolution of gene regulatory networks.

While the potential for ERVs to act as enhancers has only recently been established, the potential for ERVs to regulate cellular gene expression was first made clear due to their activity as primary or alternative promoters for gene transcription (Fig 2A)^69–71. These ERV promoters are readily identified based on cDNA sequencing data, and were initially appreciated in specific gene expression patterns of the human placenta, but have been observed in many contexts, tissues, and ontologies. ERV-derived promoters can give rise to alternate isoforms, convey tissue specificity, or even serve as the dominant promoter of a gene. An example locus in primates has even been seen to serve as a promoter in both sense and antisense directions to abutting genes Down syndrome critical region 4 (DSCR4) and DSCR8, regulating their co-expression⁷².

FIGURE 2.

Examples of co-opted ERV and transposon elements have been identified with a variety of regulatory functions. (A) An interferon-inducible alternative promoter in the ACE2 gene in humans derived from an LTR16 and a MIRb element gives rise to a nonfunctional truncated isoform during IFN signaling 85. (B) A MER41-derived enhancer upstream of the AIM2 gene in humans is bound by the interferon signaling transcription factor STAT1 and is necessary for interferon-dependent expression of AIM2 10. (C) The non-ERV retrotransposon family MIR has been attributed with insulator activity enriched in T cell–specific gene regions. MIRs recruit RNA Polymerase III (PolIII) complexes and delineate regions of active and repressive epigenetic regulation 35

In summary, ERVs have dispersed many sequence elements in mammalian genomes that possess potent epigenetic and regulatory activities that originally evolved for retroviral replication. As a class of elements, they are subject to both silencing and active transcriptional regulation, and their activity in any given cell results from the balance of these opposing epigenetic forces.

ERV-derived regulatory activity in immune contexts

Genomic evidence of pervasive ERV-derived regulatory activity has renewed interest in a decades-old idea that mobile elements may play an important role in genome regulation and evolution. In her 1950 paper describing the nature of the mutable loci we now know as transposons, Barbara McClintock observed their association with plant variegation through altered expression of nearby genes, and even predicted that this process was basic to diverse organisms⁷³. Britten and Davidson expanded on this idea, suggesting that the genome-wide amplification of transposons could facilitate the evolution of regulatory networks⁷⁴. While compelling, this hypothesis has long been challenging to investigate due to the difficulties in identifying non-coding regulatory elements and characterizing their functions.

Owing to the development of epigenomic sequencing-based technologies, a growing body of research points to an important role for ERVs in driving the evolution of immune-related gene regulatory networks. Based on unbiased global analysis of chromatin profiling data, it is now clear that ERVs constitute a significant fraction (15–30%) of enhancers in immune cells^{10,64,75–78}. Intriguingly, chromatin landscapes of immune cells and immune responses are particularly enriched for ERV-derived regulatory elements^75,79, consistent with a functional role for ERVs in immune gene regulation.

In a study focusing on the interferon response in human cells, we discovered that ERVs are a source of ~15% of binding sites for interferon-associated transcription factors STAT1 and IRF1 in human macrophages¹⁰. Based on chromatin profiling analysis of interferon-stimulated primary macrophages, thousands of ERVs exhibit properties of interferon-stimulated enhancer elements. The primate-specific MER41 ERV family was identified as a major source of ERVs with interferon-inducible activity, which was recapitulated in reporter assays. These observations suggest that ERVs act as inducible enhancers that contribute to the regulation of interferon-stimulated genes.

To further support this hypothesis, an example locus was tested experimentally. We found that one of these MER41-derived enhancers was located within 200 bp upstream of a gene AIM2, a sensor for cytoplasmic double-stranded DNA that regulates inflammasome assembly. AIM2 is known to be an interferon-stimulated gene, but its regulatory interactions were not known. Based on our genomic analyses suggesting that AIM2 was regulated by MER41, we used CRISPR to create a modified human cell line harboring a deletion of MER41, with the AIM2 gene intact. Stimulation of these cells with interferon revealed no induction of AIM2, demonstrating that its expression was regulated by the ERV (Fig 2B). Moreover, we showed that AIM2-dependent inflammasome activity was completely abrogated in these cells infected with Vaccinia virus. This study was the first to directly demonstrate an ERV-derived enhancer. In addition to acting as inducible enhancers during direct interferon exposure, ERV-derived enhancers and promoters have since been shown to be directly activated by viral or bacterial infection^80,81, indicating that ERVs have been co-opted to regulate gene expression during infection.

ERV-derived promoters are also enriched in genes related to immunity and stress response, and depleted in genes that function in metabolism and other basic housekeeping processes^71,82–84. One recently discovered example identified a truncated transposon-derived isoform of the human angiotensin-converting enzyme 2 (ACE2) gene. This isoform arises from a composite promoter derived from an ERV of the LTR16 family and another retrotransposon, MIRb (Fig 2A). The truncated transcript, MIRb-ACE2, is thought to be nonfunctional, but highly responsive to interferon signaling^85–87. This particular instance of ERV co-option is particularly notable recently due to the critical role ACE2 plays as a target of the SARS-CoV-2 virus during entry⁸⁸. These studies suggest that ERVs and other transposons may facilitate the evolution of nonfunctional alternative isoforms of the genes they promote, potentially allowing for downstream regulation of protein function through the regulation of decoy isoforms^89,90.

It is worth noting that not all cis-regulatory ERVs act as promoters or enhancers. There is evidence that they also possess insulator function in some contexts, defining the boundaries of regulatory effects within the genome. In Drosophila, the retrovirus-derived Gypsy elements are well-characterized to exhibit insulator activity⁹¹. In mammals, other retrotransposons exhibit the same capacity. Families of short interspersed nuclear elements (SINEs) have shown insulator activity in mice^36,92–94, and elements of the mammalian-wide interspersed repeat (MIR) family have been shown, both through correlation and experimental testing, to act as insulators with particular enrichment in T-cell specific networks (Fig 2C)³⁵. There is some evidence in non-immune cells that certain ERV families, like LTR13 and HERVH, have insulator activity in primates^36,95,96. Therefore, by affecting the organization of regulatory domains in mammalian genomes, it is conceivable that new ERV insertions could single-handedly alter the regulation of entire neighborhoods of genes.

In addition to regulating genes through cis-acting mechanisms, ERVs can also act as non-coding regulatory RNAs such as long non-coding RNAs^39,97. One example, the HCP5 lncRNA, has been seen to have associations with innate immunity, adaptive immunity, autoimmune diseases and even cancer since the 1990’s, and was later confirmed to be derived from an ERV insertion⁹⁸. The ERV derived lncRNA lnc-EPAV in mouse was recently found to be crucial to a proper antiviral response, regulating NF-κB expression by competing with its repressor for binding⁸⁴. As the majority of long non-coding RNAs are derived at least partially from TEs³⁹, it is likely that the many recently characterized immune related non-coding RNAs are derived from ERVs as well.

ERV-derived RNAs may also serve a non sequence-specific function in activating immune responses. In addition to the specific activity of functional lncRNAs, there is evidence that ERV derepression gives rise to double-stranded RNAs which are nonspecifically recognized by RIG-I and the cytoplasmic dsRNA sensing pathway. These dsRNA species are recognized as pathogenic and trigger activation of the innate immune response. This phenomenon has been termed ‘viral mimicry’, and it has been observed to help stimulate interferon signaling and immune responses during viral infection as well as certain cancer treatments that rely on DNA demethylation^99–101.

Recent studies appear to paint a picture where ERVs occupy diverse critical roles in the regulation and function of our immune systems. However, while some of these beneficial roles have been experimentally validated, most findings related to TEs remain correlative. As we will now discuss, there are many potential pitfalls related to investigating the functional impact of ERVs and other TEs on host immune function.

Establishing bona fide immune function

There is a growing body of evidence supporting multifaceted roles for ERVs in the epigenetic regulation of immunity. However, caution is warranted for interpreting these studies. While many studies have documented pervasive ERV-derived regulatory activity in cells, the impact of this activity on host biology remains poorly understood. Most previous research in this area employed correlation analyses or reporter assays to test the regulatory activities of ERVs, and comparably fewer elements have been directly characterized for functional importance in their endogenous genomic context. Therefore, direct functional evidence for the biological significance of ERV-derived regulatory elements remains slim.

Many studies have identified ERV-derived regulatory elements that are located physically near cellular genes, which is suggestive of cis-regulatory function. However, it is increasingly appreciated that many predicted regulatory elements--both ERV and non-ERV derived--have no or minimal gene expression consequences, due to regulatory redundancies or unfavorable 3D conformations¹⁰². For example, recent studies have shown that removing or silencing some, but not all ERVs actually results in a change in expression of nearby genes¹⁰³. The vast majority of ERV-derived elements have not been functionally characterized. Thus, while it is established that ERVs and TEs alter genome-wide epigenetic landscapes, the full extent to which they impact cellular gene regulation and function remains a major question.

Experimental validation using genome or epigenome editing via CRISPR-based tools is currently the most direct and robust method to determine whether an ERV has a regulatory function. The most direct genetic approach is to use CRISPR tools to target a particular element for genetic knockout, allowing one to test the function of the element in its endogenous context. However, isolating homozygous knockout clones can be time consuming and low-throughput, limiting even an ambitious study to the assessment of a handful of putative regulatory ERVs. In addition, their repetitive nature can make some ERVs difficult to target at a single locus without the potential for unpredictable off-target effects.

CRISPR-based epigenome editing tools have also been applied to study regulatory elements. CRISPR interference (CRISPRi) applies a cleavage incompetant mutant Cas9 fused to a transcriptional repressor such as the KRAB domain^104,105. This results in transient blockage of transcription rather than the permanent genetic alteration of traditional CRISPR, and is therefore quicker and easier to implement. However, for the same reasons, on-target silencing effects are often incomplete. It is also less appropriate for certain small loci due to the large footprint of blockage. For instance, the regulatory effect of an intronic ERV may be indistinguishable from steric interference imposed by the CRISPRi machinery on RNA Polymerase II.

In a method that cleverly leverages the repetitive nature of ERVs, the chimeric array of gRNA oligonucleotides (CARGO) approach uses guides designed to target an entire family for simultaneous epigenetic activation or repression¹⁰⁶. CARGO coupled CRISPR allows for screening of potentially functional transposable elements in parallel, but this leaves the difficulty of isolating the importance of a single perturbation among many to an observed effect.

While the vast majority of putative ERV-derived regulatory elements remain functionally uncharacterized, recent CRISPR-mediated experiments have revealed a general trend: that some, but not all, ERVs have an impact on gene expression. While these negative results have the caveat that not all physiological contents have been tested, they suggest that many ERV-derived regulatory elements are likely nonfunctional in terms of regulating gene expression, despite exhibiting epigenetic marks predictive of enhancer activity.

Our relatively poor ability to predict the regulatory effects of transposons is linked to a greater challenge of predicting gene regulatory effects of any regulatory elements in genomic data. The high redundancy and effect overlap of genomic regulatory networks makes it difficult to simply assume a crucial function for a particular putative regulatory element based solely on observed epigenetic landscape¹⁰². Even an element with all the markers of enhancer activity may be inconsequential in practice, or may affect distal genes rather than proximal ones, interactions which are difficult or impossible to predict. Incorporation of 3D chromatin data^105,107 has improved our ability to predict gene targets of regulatory elements, and the continued development of high-throughput methods for functional screening^108,109 will eventually allow researchers to better predict which ERV-derived enhancers have biologically significant regulatory functions worth testing experimentally.

Finally, while some cellular functions and signaling pathways can be assayed in vitro, the ultimate test is in vivo. While studies on mouse ERVs can naturally be tested using mouse models, it is much less straightforward to investigate the function of ERVs in the human genome. This is because ERVs show a highly lineage-specific distribution, reflecting retroviral invasions unique to each species’ evolutionary history, and rendering animal models less effective than in other cases. A recent study in the placenta used a creative workaround, where a human bacterial artificial chromosome was genetically introduced into a mouse, testing the cis-regulatory connection between the enhancer and the gene in an animal model¹¹⁰.

Disease implications

There is mounting evidence that ERVs are crucial for the healthy function of mammalian immune systems, as well as other tissue systems, but their involvement in a variety of disease states is equally clear. ERV dysregulation has been tied to cancer, autoimmunity, multiple sclerosis, and neurodegeneration^51,111–113. While the focus of many recent studies have been on the pathological effects of ERV-derived transcripts and proteins, the fact that they show disease-specific reactivation implies that ERVs themselves are epigenetically dysregulated, which may have far-ranging regulatory consequences. However, studies of ERV behaviour in disease states have relied heavily on correlative observations. Though many studies have observed upregulated ERV expression in a number of human disease states, the potential causative roles for ERV activity in disease has long been subject to controversy.

The emerging roles for ERVs in healthy immune regulation imply that they may also be involved in dysregulation in autoimmunity. If ERVs regulate expression of immune genes, their dysregulation would likely lead to dysregulation of the immune system, causing chronic activation or silencing of specific genes. Transcriptional dysregulation of immune genes is associated with many autoimmune disorders, for example systemic lupus erythematosus (SLE), where global upregulation of ERV expression in B and T cells is thought to result from deficient DNA methylation^62,114,115.

Aberrant epigenetic activation of ERVs has manifold consequences. ERVs themselves contribute to spurious activation of both innate and adaptive immunity through a process coined “viral mimicry.” As with exogenous viruses, transcription of ERVs gives rise to double stranded RNA and cytoplasmic reverse transcribed DNA which are recognized by pattern recognition receptors of the innate immune system⁹⁹, activating innate immune responses. Translated ERV genes may even be recognized as antigens by immune cells and trigger an autoimmune response^116–118. In healthy cells, normal repression of ERVs is sufficient to prevent this mechanism, but disruptions in transposon silencing mechanisms could conceivably drive autoimmune responses to ERV-derived transcripts and proteins.

By triggering cellular autoimmune responses, ERVs have been implicated in SLE, multiple sclerosis (MS), neurological disorders, and neurodegeneration¹¹⁹. Studies have suggested an association between ERV transcription and the progression of MS^120–122, but initially it was unclear if this association was indicative of a role for ERVs in MS progression or if ERV derepression was simply a downstream effect of general disease state epigenetic dysregulation. Recently, however, several studies have pointed directly to ERV transcription activating innate immunity through pathogen sensing Toll-like receptor (TLR) pathways in both MS and neurodegeneration^123,124.

While inappropriate ERV activity may be a contributing factor to disease, immunostimulatory ERV reactivation has been harnessed in the application of epigenetic therapy for cancer treatment. It is now recognized that many cancer therapies impede cancer growth in part by causing transcriptional reactivation of ERVs and other TEs^99,125,126, which triggers the viral mimicry pathway. Therefore, the viral mimicry response can be seen as a double-edged sword, where the activation of innate immune responses by transcription of TEs could have either beneficial anti-tumor effects or pathological autoinflammatory consequences.

While there has been substantial recent interest in the relationship between ERV transcription and disease, most research has focused on the consequences of ERV-derived proteins or transcripts. The potential for ERVs to act as pathological regulatory elements is less well explored, but several examples exist. For example, the ERV-derived isoform of the complement component 4 (C4) is associated with progression of schizophrenia through overactive synaptic pruning¹²⁷, while low copy number of the same isoform has been associated with type 1 diabetes through an unknown mechanism¹²⁸. However, the extent to which ERV-derived enhancers contribute to epigenetic dysregulation in immune disorders is almost entirely unexplored. Given that ERVs have been shown to function as immune inducible enhancers, it stands to reason that aberrant activation of ERV-derived enhancers may be responsible for establishing transcriptional states associated with chronic inflammation.

Evolutionary implications for host immunity

The function of the immune system is to ward off an ever-changing and evolving barrage of pathogens. The constant threat of new and evolving pathogens is a source of genetic conflict that drives rapid evolution of host immune systems^129–131. As a result, protein-coding genes involved in host immunity show strong evidence of rapid divergence and evolution. Whether immune regulatory networks show similarly elevated patterns of evolution is not well understood, but cross-species transcriptional profiling has revealed extensive differences in innate immune responses and immune cell profiles^132,133. Thus, it is likely that the epigenetic landscape of immune cells and immune responses have undergone significant divergence throughout vertebrate evolution, driven by pathogens like viruses.

The finding that ERVs are co-opted for immune function implies that retroviruses directly facilitate host immune function by providing novel regulatory and protein sequences. ERV co-option may be especially well suited to the fast-paced evolution associated with genetic conflicts, such as the one between pathogens and the host immune system. ERVs provide ready-made proteins and regulatory elements with high copy numbers, and in some cases act as turncoats providing insights into the structure of the enemy. Endogenization of a retrovirus gives the host genome a head start at evolving products that will match and compete with their exogenous equivalents as new dominant-negative antiviral elements¹³⁴. Furthermore, the transposable nature of ERVs leads to high copy numbers which can mutate and drift rapidly, increasing the likelihood of a beneficial mutant.

The differential co-option of lineage-specific ERVs for immune regulation may be an underappreciated driver of species-specific differences in immune regulatory networks. Even well-conserved immune genes may exhibit distinct epigenetic and regulatory patterns, and thus transcriptional behavior, due to differences in ERV insertions giving rise to lineage-specific enhancers⁶⁴. For instance, a great ape-specific insertion of a primate-ubiquitous ERV family regulates pathogen responsive expression of antiviral proteins effective against both modern and ancient retroviruses¹⁰⁰. The ERV serving as the dominant promoter of human 3GAL-T5 is entirely absent in mice,¹³⁵. Even when species harbor ERVs of the same family, originating from the same ancestral infections, differences in insertion site and co-option events allows for differences in regulatory networks which might otherwise be much slower to evolve. While innate immune signaling responses such as the interferon response are generally assumed to be conserved in animal models, it is possible that lineage-specific ERVs have facilitated rewiring of immune transcriptional networks across species.

Even within species, ERVs give rise to genetic variation. In species where ERVs are replication-competent, such as mice, new ERV insertions could act as a source of regulatory variants²². Although ERVs are thought to be inactive in humans, insertional polymorphisms are still present in the population¹³⁶. As discussed in the previous section, these variations can be highly consequential, with proposed associations to schizophrenia and diabetes for one ERV-derived isoform. Other retrotransposons, including LINEs, SINEs, and SVA elements, remain an active source of new insertions in humans^137,138, and many of the epigenetic mechanisms discussed related to ERVs also apply to these elements.

It is clear that ERVs are a fast-evolving and dynamic source of regulatory activity integral to both healthy and pathogenic variance between species and between individuals. The need for competitive adaptation and counter-adaption at the host-pathogen interface necessitates rapid evolution. In a twist of irony, endogenized viruses give hosts a leg up in evolving restriction factors and reorganizing immune networks, but they come with the risk of serious pathological effects as well (Fig 3).

FIGURE 3.

Retroviruses both antagonize and facilitate rapid evolution of host immunity. Retroviruses drive evolution of host immunity both as pathogens and as a source of novel DNA. They occasionally endogenize and become co-opted for the benefit of the host. Cooption of ERVs is particularly prevalent in the immune system and allows for rapid reorganizing of the host immune regulatory networks in the ongoing arms race with exogenous viruses and other pathogens. Created with BioRender

Why are viruses repeatedly co-opted for immunity?

Does mere poetic justice drive the evolutionary domestication of ERVs for antiviral defense, or is there a more logical explanation at play? Current observations suggest that ERV co-option is particularly prevalent in gene networks that establish immune specific cell types and signaling pathways. We speculate that one potential reason is that many retroviral LTRs have regulatory features that predispose them to immune function, having already evolved to hijack host immune regulatory networks. Indeed, ERV LTRs often contain immune-related transcription factor binding motifs, particularly for those specific to immune signaling pathways, encoded in ERV LTRs^51,63,66. It is likely that in many cases, immune-related transcription factor binding motifs were already present in the LTRs of the originating ancient retroviruses. Consistent with this idea, modern exogenous retroviruses including HIV-1 harbor immune-inducible transcription factor binding motifs in their LTR sequences. The dependence of HIV-1 on immune related transcription factors is well documented^42,44,45,47, and sequence analysis of many other retroviruses shows a similar pattern of motif abundance. The repeated integration of retroviral LTRs would therefore provide a recurrent source of ready-made immune regulatory elements.

There are clear selective benefits that might drive exogenous retroviruses to incorporate these host factor binding motifs. HIV-1, as with many other retroviruses, is regulated by many transcription factors specific to the host immune system. Considering the fact that HIV’s primary target cell type are professional immune cells this is not surprising. Once integrated as a provirus, HIV-1 benefits from being transcriptionally driven by those signaling pathways most reliably active in its host. However, there may be other benefits to the virus of being transcriptionally upregulated by the host antiviral response. Even in other retroviruses which do not specifically target immune cell types, the antiviral response is a signaling pathway that coincides with infection, and encoding regulatory motifs that allow viral transcription to be upregulated by this pathway may be a way for the virus to hijack cellular antiviral signaling and turn it into an advantage. In addition, HIV-1, along with a number of other retroviruses, persists long term in an infected host organism through a program of transcriptionally silent latency, punctuated by transcriptionally active reactivation events^50,139,140. It is known that these reactivation events can be triggered by external activation of the immune response, including secondary infection, cellular stress or injury, or inflammation. The sporadic nature of immune activations like these is well suited to the punctuated latency that allows a retrovirus like HIV to persist as a chronic infection.

Once endogenized, ERVs that exhibit inducible activity might be further favored for retention and eventual co-option in the host genome. Hijacking of host defensive signaling pathways may allow the ERV to avoid being epigenetically silenced by the cell, and result in further replication. Their inducible activity potentially makes them less disruptive as a class of insertions, as the regulatory effects of new insertions would only be exposed during infection. Such ERVs may be favored to accumulate in host genomes, and potentially co-opted for functional regulatory roles.

Although we have focused on inducible regulatory activity, such as stimulation by interferon, it is possible that ERV-derived enhancers have additional unique regulatory features. The promoter behavior integral to retroviral LTRs is often retained by endogenized LTRs. The stochastic, or “bursty” transcriptional activity of exogenous proviruses is well characterized^141,142. The source of this noise is unclear, though studies have suggested that the functional activity of transcription factors and cofactors can be affected by both sequence and epigenetic landscape around the core binding sequence^143,144. Transcriptional noise is also observed in native cellular genes, often in a cell type dependent manner^145–147, and is apparently critical in allowing for important response diversity and differential fate decisions in otherwise isogenic populations of cells^148,149.

It is possible that the inherent stochasticity of proviral regulation could be exploited in regulating the host immune system. Stochastic or variable gene expression has been widely observed in the context of immunity where it is proposed to be important for converting signals that can be thought of as binary, on or off cellular responses, often pulsatile into persistent, throttled responses at the tissue level^150,151. Beyond this regulation of intrinsic transcriptional noise, ERVs are also known to exhibit high epigenetic variability^152,153, which may contribute to a higher order patterns of variability, such as the bimodal responsiveness of single cells to interferon signaling or differences across individuals. Like the lower order intrinsic noise that smooths signal responses through temporal variability, variability in interferon responsiveness is also thought to be functional, allowing for a flexible and tunable immune response¹⁵⁴. The balance of the immune response between restraint and proper activity is crucial to organismal health and difficult to achieve through a global all-or-nothing signal. Variability of signal responsiveness is likely a major safeguard against an overactive or underactive interferon response, but the mechanistic origin of this variability is not well characterized. Whether epigenetically variable ERV-derived regulators play a role in this system remains to be seen.

The presence of immune-related binding motifs in ERVs may also be the result of random chance. Each species contains hundreds of families of lineage-specific repeats, and binding motifs for transcription factors are often short and degenerate. While the biology of these ancient retroviruses will always remain a matter of speculation, their impact on our evolution is beginning to come into focus.

Conclusion

The advent of epigenomic technologies have brought to light the fact that ERVs and other transposons are a major source of regulatory activity in mammalian cells, particularly in the context of immunity. They provide ready-made modules of transcription factor binding sites and promoter elements with the potential to be moved, duplicated, and dispersed throughout the genome. While most insertions are likely to be either detrimental or of little biological consequence, the occasional insertions may confer a beneficial function and become retained and conserved.

Given the fact that ERVs and transposons represent a major source of lineage-specific DNA, it is possible that they represent a widespread mechanism of regulatory evolution. While there are a growing number of examples demonstrating the co-option of ERV-derived enhancers for immunity, it remains to be seen whether these represent rare events or the tip of the iceberg. Genomic studies continue to uncover many ERVs showing canonical markers of promoter, enhancer, or insulating activity in a variety of cell types and contexts, but the bottleneck of experimental validation means that only a small subset of these putative regulatory elements have been tested for functional importance. Application of single cell chromatin profiling technologies in immune cells^155–157 promise to reveal even more putative regulatory ERVs that may be active only in rare or transient cell populations. The continued development of CRISPR-based screening technologies for regulatory elements^108,109 will eventually allow researchers to more thoroughly dissect the regulatory and functional consequences of ERVs.

Despite their emerging importance in the evolution of immune systems, we know from analysis of mammalian genomes that ERVs are highly specific to different lineages and species, reflecting different histories of ancient retroviral infections. However, thus far almost all investigations of ERV regulatory activity have been done in either human or mouse cells, due to the wealth of epigenomic resources for these species. More broad functional studies are needed to explore whether co-option of ERVs and other transposons as regulatory elements is a general mechanism of organismal evolution. Already there is evidence of immune co-option of non-ERV transposons in plants, where they were first discovered^158,159. It is likely that as the immune responses of more diverse species are investigated using epigenomic assays, ERVs and other transposons will continue to be implicated in the evolutionary diversification of species immune responses.

Further study is warranted to understand the role of regulatory ERVs in human disease states such as autoimmunity and cancer. ERVs and other transposons are unique in that they exhibit a complex duality of beneficial and pathogenic potential, which is reflected in their context-dependent recruitment of activating or repressive epigenetic marks. However, studying their effects in human disease will be challenged by the added difficulty presented by lineage-specific differences in transposon landscapes, which preclude most animal models. Experimental study of human ERVs in the genomes of human patients is nearly impossible, therefore technically difficult approaches using patient-derived cells and/or humanized animal models will be necessary to fully characterize their potential immune functions.

In addition, the focus of many of the studies has been on “older” elements that are fixed in all humans. Practically, this is because older elements have accumulated more random mutations, and are easier to map using short-read technology. Polymorphic elements will also often be absent from the species reference genome, requiring specialized software to analyze¹⁶⁰. Detecting transposon-derived structural variants can be technologically challenging, but the adoption of long-read sequencing technologies will significantly facilitate their identification^161,162. However, to truly incorporate newly identified transposon polymorphisms into genomic studies, adoption of graph-based “pan-genome” reference genomes that include structural variation will be necessary¹⁶³.

Viral pandemics are established to have been significant selective forces that have driven coevolutionary adaptation of host immune systems for eons¹³¹. As discussed in this review, studies on ERVs are beginning to reveal a new, underappreciated route through which viruses have shaped the evolution of host immunity. By endogenizing and becoming co-opted as immune regulatory elements, ancient retroviruses have become domesticated to help thwart infections by modern infectious viruses and other pathogens. Further elucidation of this tangled evolutionary pathway, as well as its implications on immune epigenetic landscapes, will be key to fully understanding the molecular basis of immune regulation and dysregulation in infection and autoimmunity.