Switching Sides: How Endogenous Retroviruses Protect Us from Viral Infections

ABSTRACT

Long disregarded as junk DNA or genomic dark matter, endogenous retroviruses (ERVs) have turned out to represent important components of the antiviral immune response. These remnants of once-infectious retroviruses not only regulate cellular immune activation, but may even directly target invading viral pathogens. In this Gem, we summarize mechanisms by which retroviral fossils protect us from viral infections. One focus will be on recent advances in the role of ERVs as regulators of antiviral gene expression.

INTRODUCTION

Viruses do not have a good reputation. They are generally perceived as harmful pathogens causing acute or chronic diseases that need to be well controlled or, ideally, eradicated. Thus, it may at first glance seem worrisome that about 8% of the human genome consists of retroviral DNA sequences. These so-called endogenous retroviruses (ERVs) represent remnants of once infectious exogenous retroviruses that became fixed in our DNA and are now inherited in a Mendelian manner. While many of them are usually dormant, they can be reactivated by several stimuli, including viral infections. For example, dengue, herpes simplex, influenza, and human immunodeficiency viruses have all been shown to induce the transcription and translation of endogenous retroviral elements (1–5). Although human ERVs (HERVs) are not known to produce infectious viral particles in vivo, this resurgence of virus-derived RNAs and proteins in our cells may seem alarming. However, many ERVs are not detrimental and have even been coopted for important physiological functions in the host. Besides well-known examples, such as syncytins that regulate placental development (6, 7), ERVs have become integral parts of immune defense mechanisms and help to fight off invading viral pathogens.

To date, hundreds of thousands of HERVs have been identified, ranging from full-length proviruses to short gene fragments (8). Thus, it is not surprising that they support the antiviral immune response via numerous mechanisms. These include the enhancement of cellular sensing pathways, regulation of viral gene expression, blockade of entry receptors, and direct restriction of virion assembly. In this Gem, we summarize recent advances in the interplay of endogenous and exogenous retroviruses. We present important principles underlying the antiviral activities of ERVs and illustrate how the host cell has coopted fossils of possibly once harmful retroviruses to limit the spread of current viral pathogens.

ERV-DERIVED NUCLEIC ACIDS TRIGGER INNATE SENSING CASCADES

Host cells are equipped with a variety of pattern recognition receptors (PRRs) that sense virus-derived nucleic acids and induce antiviral gene expression upon infection. These include cytosolic sensors such as RIG-I or MDA5, as well as Toll-like receptors (TLRs) such as TLR3 or TLR9 that recognize double-stranded RNA (dsRNA), single-stranded RNA (ssRNA), and/or RNA-DNA hybrids (9–13). Usually, these sensors efficiently distinguish between self and nonself to prevent undesirable activation in the absence of infection (14). In a process of viral mimicry, however, the expression and subsequent sensing of endogenous retroviral nucleic acids may be beneficial and boost the induction of an antiviral state in the infected host (Fig. 1A). In particular, dsRNA seems to play an important role in ERV sensing. Upon ERV activation, dsRNA species may be formed by pairing of ERV RNA with cellular antisense transcripts or complementary exogenous viral RNA. In addition, intramolecular pairing of ERV transcripts can also activate PRRs that detect dsRNA.

FIG 1.

ERV-mediated mechanisms of antiviral immunity. (A) Enhancement of innate sensing by ERV-derived nucleic acids. (B) Regulation of antiviral immune responses by ERV-derived long noncoding RNAs (lncRNAs). (C) Enhancement of innate sensing by ERV-derived proteins. (D) Receptor interference by endogenous retroviral envelope (Env) proteins. (E) Dominant negative complementation of exogenous viral particles by endogenous retroviral proteins. (F) Interference with incoming viral capsids. (G) Increased gene plasticity due to ERV-mediated recombination events. (H) Regulation of antiviral gene expression by ERV-derived promoter and enhancer elements.

One striking example of ERV sensing was recently described by the group of Benjamin Hale (15). The authors showed that influenza A virus (IAV) and influenza B virus (IBV) infection triggers a loss of SUMO-modified TRIM28/KAP1. RIG-I, MDA5, and alpha interferon (IFN-α) were not required for this loss, and the viral and/or cellular pathways triggering this SUMO switch remain to be discovered. Since TRIM28 acts as a restriction factor that limits retroviral transcription (16, 17), this derepression resulted in an increased expression of endogenous retroviruses. PRR knockout and inhibitor experiments suggested that the sensing of HERV-derived dsRNA via RIG-I triggers the expression of type I interferons (IFNs) and IFN-stimulated genes (ISGs), thereby limiting IAV replication (15). While the correlation of increased HERV expression with ISG induction is very suggestive, a causal link has not been proven in a definitive manner.

Another ERV-derived nucleic acid that is sensed and may contribute to antiviral immune responses has been described in birds. In 2019, Chen and colleagues demonstrated that an antisense long noncoding RNA (lncRNA) derived from an endogenous avian leucosis virus (ALVE1) activates the dsRNA sensor TLR3 (18). As a result, this lncRNA (termed lnc-ALVE1-AS1) induces the expression of beta interferon (IFN-β) and ISGs in chicken embryonic fibroblasts and suppresses replication of avian leukosis virus subgroup J (ALVJ) (Table 1). While expression of this lncRNA could be induced by the DNA methylation inhibitor 5-Aza-dC, its expression upon viral infection remains to be determined.

TABLE 1.

Examples for endogenous retroviruses involved in immunity

Mode of action	ERVb	Species (examples)	Molecular mechanisma	Target virus(es)	Reference(s)
Induction of sensing via nucleic acids	Multiple ERVs (e.g., ERV3-2, ERV4700)	Humans	Sensing of ERV-derived nucleic acids by MDA5, RIG-I, and TLR3	Most likely several viruses	15, 20–23
	ALVE1	Chickens	Induction of TLR3 signaling by ERV-derived lncRNA (lnc-ALVE1-AS1)	Avian leukosis virus subgroup J, possibly others	18
Regulation of immune signaling by lncRNAs	lnc-EPAV	Mice	lncRNA-mediated sequestration of SFPQ, a repressor of Rela expression	VSV, possibly others	25
	MER9a2, LTR5A, MLT2A1	Humans	Binding of SFPQ; induction of Rela expression (?)	Most likely several viruses	25
Modulation of immune activation by ERV proteins	HERV-W Env	Humans	Activation of TLR4/CD14; induction of IL-1β, IL-6, TNF-α, and IL-12p40	Most likely several viruses	26
	HERV-K(HML2) dUTPase	Humans	Activation of NF-κB via TLR2; induction of IL-1β, IL-6, IL-8, IL-10, IL-12p40, IL-23, IL-17, TNF-α, RANTES, IFN-γ, and CCL20	Most likely several viruses	28
Receptor interference	ev3, ev6, ev9	Chickens	Receptor blockade	Avian leukosis virus, subgroup E	40
	Fv-4r (Akvr-1)	Mice	Blockade of the receptor CAT-1	Ecotropic murine leukemia viruses	41, 42, 46
	Rcmf, Rcmf2	Mice	Blockade of the receptor XPR1	Polytropic murine leukemia viruses	47–51
	en-JSRV env loci	Sheep	Blockade of the receptor hyaluronidase 2 (HYAL2)	Jaagsiekte sheep retrovirus, enzootic nasal tumor virus (?)	52
	Refrex-1 (loci ERV-DC7 and –DC16)	Cats	Secreted truncated Env; extracellular receptor interference	Endogenous and exogenous feline retroviruses (ERV-DC genotype I, FeLV-D)	54
	FeLIX	Cats	Secreted truncated Env; extracellular receptor interference	FeLV-B	55
	HERV-T env locus	Primates	Depletion of MCT-1 from the cell surface	Extinct HERV-T ancestor	56
	Suppressyn	Primates	Blockade of the receptor ASCT2	RD114/mammalian type-D retroviruses	57
Dominant negative virion complementation	Fv-4r	Mice	Virion incorporation of Env with inactive fusion peptide	Ecotropic murine leukemia viruses	58
	HERV-K gag locus	Humans	Coassembly with exogenous viral Gag; generation of malformed and poorly infectious virions; reduced virion release	HIV-1	62, 63
	enJS56A1, enJSRV-20 (?)	Sheep	Dominant negative interference of a defective Gag protein with viral exit	Jaagsiekte sheep retrovirus	64–66
Interference with incoming viral particles	Fv1	Mice	Binding to incoming capsids	Murine leukemia viruses	67–70
Increased plasticity of immune genes	Multiple ERVs	Several species	Gene conversion and duplication events increasing the diversity of MHC II genes and potentially others	Most likely several viruses	85
Regulation of antiviral gene expression	X-MLV LTR	Mice	Enhancer of mA3; increase in the levels and antiviral activity of APOBEC3	MMTV, potentially others	91, 92
	HERVP71A LTR/enhancer L	Humans	Enhancer of HLA-G	Unclear	93, 94
	MER41	Humans	IFN-γ-inducible enhancer of AIM2 expression	Most likely several viruses	97
	MER41	Humans	IFN-γ-inducible enhancer of IFI6 expression	Several flaviviruses	97
	MER41	Humans	IFN-γ-inducible enhancer of SECTM1 expression	Most likely several viruses	97
	ERV9	Apes	Transcription start site for IRGM	Active against mycobacteria, antiviral activity unclear	105
	LTR12C	Humans	IFN-γ-inducible promoter of GBP2	HIV, measles virus, Zika virus, and other furin-dependent viral pathogens	107
	LTR12C	Humans	Promoter of GBP5	HIV, measles virus, Zika virus, and other furin-dependent viral pathogens	107
	LTR12C	Humans	Promoter of SEMA4D (?)	Multiple viruses?	107

^aTLR, Toll-like receptor; IL, interleukin; TNF-α, tumor necrosis factor alpha; IFN-γ, gamma interferon; lncRNA, long noncoding RNA.

^bLTR, long terminal repeat.

It has also been proposed that more specific immune activation may be achieved by ERV-derived antisense transcripts that form dsRNA molecules upon binding to complementary RNA (cRNA) molecules derived from exogenous retroviruses (13, 19). In this pathway, the ERV RNA not only acts as a pathogen-associated molecular pattern (PAMP), but may also be considered a kind of PRR itself, since it recognizes complementary exogenous RNA and subsequently triggers the activation of downstream dsRNA-sensing cascades.

Additional evidence for a contribution of viral mimicry to cellular immune responses comes from cancer research. For example, several studies reported a beneficial effect of DNA demethylating drugs in tumor therapy that involves the induction and sensing of endogenous retroviruses and other transposable elements (20–22). In all cases, the induction of antitumor immune responses required the presence of dsRNA sensors, such as TLR3, MDA5, or RIG-I, and may result in increased T cell-mediated killing of tumor cells. Similarly, Lee and colleagues suggested that MDA5-mediated sensing of ERVs contributes to the beneficial effects of tumor radiotherapy (23). Intriguingly, additional knockout of the ERV-repressor TRIM28/KAP1 further boosted the beneficial effects of radiotherapy in two tumor models (23).

Further evidence for a role of ERVs in immune activation comes from congenital diseases, in which the accumulation of retroviral nucleic acids may result in a detrimental chronic IFN response. For example, patients suffering from Aicardi-Goutières syndrome (AGS) show aberrant immune activation and lack cellular factors such as SAMHD1 or TREX1 that reduce the levels of endogenous retroviral RNA/DNA species (24).

Together, these observations highlight the potential of ERV-derived nucleic acids to enhance cellular immune responses. While translational approaches that take advantage of this sensing pathway are already pursued in tumor research, the induction of ERVs in the treatment of viral infections also deserves further attention.

ERV-DERIVED lncRNAs REGULATE ANTIVIRAL IMMUNE RESPONSES

ERV-derived nucleic acids may not only act as PAMPs triggering cellular sensing cascades, but also regulate immune signaling in a more direct manner (Fig. 1B). One interesting example is a murine ERV-derived lncRNA termed lnc-EPAV (for ERV-derived lncRNA positively regulating antiviral responses) that boosts antiviral gene expression by increasing the levels of the transcription factor NF-κB (25). More specifically, lnc-EPAV is expressed upon Sendai virus and vesicular stomatitis virus (VSV) infection and induces the expression of NF-κB/RelA by sequestering its repressor SFPQ (splicing factor, proline and glutamine rich) in the nucleus. As a result, the production of antiviral cytokines such as IFN-β or interleukin 6 (IL-6) is increased. Since lnc-EPAV itself is also NF-κB responsive, this ERV pathway may act as a positive feedback loop enhancing antiviral immune responses (25). Notably, the human ortholog of SFPQ also interacts with multiple HERV-derived RNAs (e.g., MER9a2, LTR5A, and MLT2A1), suggesting that a similar immune-regulatory mechanism exists in humans (25).

ENDOGENOUS RETROVIRAL PROTEINS MODULATE IMMUNE ACTIVATION

Besides endogenous retroviral nucleic acids, ERV-derived proteins may also be sensed and/or regulate immune activation (Fig. 1C). For example, plasma membrane-associated TLRs have been shown to detect different retroviral proteins. This includes sensing of the HERV-W envelope (Env) protein by TLR4 and its coreceptor CD14 (26). As a consequence, proinflammatory and antivirally active cytokines such as IL-1β, IL-6, TNF-α, and IL-12p40 are released (26). The TLR4-dependent induction of cytokines by HERV-W Env was confirmed in an independent study (27) that additionally described an increased expression of inducible nitric oxide synthase (iNOS) in the presence of this retroviral envelope protein. While the production of nitric oxide by iNOS is an important component of the innate immune response, the authors speculate that HERV-W Env-mediated iNOS induction may also contribute to brain damage in multiple sclerosis (27). Another TLR-stimulating retroviral protein is the dUTPase of HERV-K (HML2) that activates NF-κB and induces the release of cytokines (i.e., IL-1β, IL-6, IL-8, IL-10, IL-12p40, IL-23, IL-17, TNF-α, RANTES, IFN-γ, and CCL20) in a TLR2-dependent manner (28).

Exogenous retroviral Envs can also directly activate canonical NF-κB signaling, independently of PRR activation (29). This process involves recruitment of transforming growth factor beta (TGF-β)-activated kinase 1 (TAK1) to the cytoplasmic domain of Env. Thus, it is tempting to speculate that endogenous retroviral Envs with similar NF-κB-activating activity remain to be discovered. Notably, several HERV Envs with immunosuppressive activity have already been described. These include human and murine syncytins, which are expressed in placental cells and may allow maternal immune tolerance to the fetus and placenta (30, 31). Finally, endogenous retroviral Envs of HERV-K18 (32, 33) and possibly HERV-W (34) have been suggested to contribute to the pathogenesis of auto-immune disorders because they act as superantigens that induce excessive activation of T cells. However, this hypothesis remains controversial. While some studies observed an association of HERV-K18 polymorphisms with type I and II diabetes (33, 35), others found no evidence for a contribution of this HERV to the pathogenesis of autoimmune diseases (36, 37). Nevertheless, the above-mentioned examples illustrate that transcription and translation of ERVs need to be tightly regulated in order to prevent detrimental effects on the host.

ENDOGENOUS RETROVIRAL ENVELOPE PROTEINS BLOCK VIRUS ENTRY RECEPTORS

Apart from rather broad and unspecific immune-modulatory activities, endogenous retroviral Env proteins can also directly inhibit viral infection via a process termed receptor interference. This mechanism involves direct binding of cell-associated or cell-free endogenous retroviral Env proteins to cellular entry receptors (Fig. 1D). As a result, these receptors are blocked and cannot mediate attachment of exogenous viruses using the same entry pathway. Furthermore, ERV-derived Envs may bind newly synthesized entry receptors intracellularly and prevent their transport to the cell surface. Since exogenous and endogenous retroviruses frequently share the same receptors, endogenous Envs offer a large repertoire for broadly active antiviral factors.

Early reports of receptor interference date back to the 1960s, when several groups demonstrated that Env proteins of avian leukosis viruses (ALV) induce resistance to exogenous viruses of the same subgroup (38, 39). In 1981, Robinson and colleagues identified three Env-expressing endogenous retroviruses in chickens (ev3, ev6, and ev9) that reduce the susceptibility to and entry of subgroup E ALVs (40). The authors already speculated that additional examples of receptor interference will be found and hypothesized that the previously identified antiviral genes Fv-4^r (41) and Akvr-1 (42) “represent defective endogenous viruses” in mice (40). Indeed, Fv-4^r and Akvr-1 subsequently turned out to represent a single transcriptionally active but truncated provirus (43–45). Experiments in transgenic mice revealed that the endogenous retroviral Env protein expressed from the Fv-4^r (Akvr-1) locus is responsible for the observed restriction of murine leukemia viruses (MLV) (46). Similarly, the murine loci Rmcf and Rmcf2 also express endogenous retroviral Env proteins blocking MLV entry (47–51). Apart from chickens and mice, protective retroviral env genes have also been identified in sheep (52, 53) and cats (54, 55). Notably, the antivirally active Env proteins Refrex-1 and FeLIX in cats are C-terminally truncated and released into the extracellular space, where they can block receptors and thus entry of feline leukemia viruses (FeLV) (54, 55).

More recently, Env-derived restriction factors have also been discovered in primates. Blanco-Melo et al. identified a retroviral envelope protein (HERV-T Env) in humans that fails to mediate infection, but abrogates susceptibility of human cells to infection with reconstituted HERV-T-like viral particles (56). This inhibitory effect was associated with decreased surface levels of the virus entry receptor monocarboxylate transporter-1 (MCT-1). Similarly, the Env-derived protein suppressyn may protect humans and other hominoids from retroviral infections, as it blocks the surface receptor ASCT2 that mediates attachment of RD114/mammalian type-D retroviruses (57). Members of this virus group circulate in nonhuman primates and other mammalian species. Thus, cooption of retroviral envelope genes enables efficient restriction of (zoonotic) transmission events and may have contributed to the extinction of the corresponding exogenous retroviruses.

ERV PROTEINS COMPLEMENT VIRIONS IN A DOMINANT NEGATIVE MANNER

Intriguingly, certain endogenous Env proteins suppress viral replication at several steps of the viral life cycle. For example, murine Fv-4 not only inhibits MLV entry via receptor interference, but also reduces virion infectivity (Fig. 1E). More specifically, Fv-4 is incorporated into budding virions, but fails to mediate infection as it harbors an inactivating mutation in its fusion peptide (58). Similarly, Env proteins of the evolutionarily young HERV-K family of human endogenous retroviruses can be incorporated into HIV-1 particles in vitro (59, 60). However, it remains to be determined whether HERV-K Env is also incorporated into HIV-1 virions and reduces particle infectivity in infected individuals. HERV-K108 Env has also been shown to negatively interfere with HIV-1 replication (61). However, the underlying mechanisms may be independent of virion incorporation, as this Env reduces HIV-1 protein levels and overall virion production (61).

Besides Env, endogenous retroviral Gag proteins can also restrict retroviral replication. This includes the Gag protein of HERV-K, which coassembles with HIV-1 Gag, thereby reducing the production of fully infectious HIV-1 particles (62). Subsequent mechanistic analyses revealed that the inhibitory activity of HERV-K Gag depends on its N-terminal capsid domain and results in the formation of malformed HIV-1 virions (63). Another interesting example comes from sheep that harbor the endogenous Jaagsiekte sheep retrovirus enJS56A1, which expresses a nonfunctional Gag (64). This defective protein directly associates with Gag proteins of exogenous JSRV. As a result, egress of JSRV is suppressed in a dominant negative fashion, and viral particles accumulate in intracellular compartments (64, 65). enJSRV-20, a related endogenous retrovirus, may restrict JSRV replication in a similar manner, as it harbors the same inactivating mutation (Trp21) in its Gag protein (66).

ERV PROTEINS INTERFERE WITH INCOMING VIRAL PARTICLES

The best-characterized Gag-derived restriction factor is probably Fv1 in mice, which inhibits MLV replication (67, 68) and interacts with exogenous retroviral capsid proteins (69). In contrast to the Gag proteins mentioned above, however, Fv1 acts in viral target cells and restricts MLV replication before integration by directly interacting with incoming retroviral capsids (69, 70) (Fig. 1F). Thus, the ability of endogenous viral Gag proteins to interact with their exogenous counterparts may confer blocks at several steps of the retroviral replication cycle. The importance of Fv1 in antiviral immune responses is further highlighted by signatures of positive selection in the Fv1 gene in the genus Mus that are indicative of a long-lasting coevolution with exogenous viral pathogens (71, 72).

Notably, the description of additional coopted gag genes such as wucaishi1 (wcs1) and wucaishi2 (wcs2) (73) suggests that more antiviral Gag proteins remain to be discovered. Moreover, other products of ERVs (e.g., RNA, polymerase, integrase, and reverse transcriptase) may in principle also be incorporated into exogenous viral particles and interfere with viral replication. Laderoute and colleagues reported an interesting association in this regard; they observed an increased number of genomic HERV-K102 pol copy numbers in highly exposed HIV-1-seronegative sex workers compared to those in HIV-1-infected individuals (74). However, the molecular mechanisms underlying this association have remained unclear. Notably, the incorporation of ERV-derived gene products into exogenous viral particles may not always negatively interfere with their assembly or infectivity. For example, certain HERV-derived protease (75, 76), integrase (77), or Env (78, 79) proteins may at least partially complement HIV-1 mutants with defects in the respective genes and sometimes even expanded the tropism to CD4-negative cells (78, 79). Nevertheless, the cooption of retroviral genes to negatively interfere with virion production represents a simple yet effective antiviral mechanism that has evolved several times independently during mammalian evolution.

REPETITIVE ERV ELEMENTS INCREASE PLASTICITY OF IMMUNITY GENES

All of the antiviral mechanisms described above depend on the transcription and/or translation of endogenous retroviral sequences. Nevertheless, transcriptionally silent ERVs can also provide a selection advantage to the host. For example, the presence of repetitive ERV elements may promote recombination processes that increase the number and/or allelic variety of host genes (80–82) (Fig. 1G). This mechanism of gene plasticity might usually be selected against in monogenic systems, since homologous recombination events can also result in the loss of genes or gene fragments (83, 84). In contrast, the risk-benefit ratio of ERV-dependent recombination is most likely lower in multigene families, where the insertion of repetitive ERV sequences may provide a fitness advantage and facilitate adaptation to viral pathogens. One notable example is the MHC locus in primates, where the integration of different ERVs may have promoted the diversity of alleles and thereby broadened the spectrum of viral peptides that can be presented to T helper cells (85, 86).

ERV-DERIVED PROMOTERS AND ENHANCERS REGULATE ANTIVIRAL GENE EXPRESSION

Retroviral elements may also contribute to the regulation of host gene expression (Fig. 1H). One key characteristic of retroviruses is the presence of promoters, enhancers, transcription start sites, and other gene regulatory elements is their long terminal repeats (LTRs). Exogenous retroviruses harbor multiple binding sites for cellular transcription factors in their LTRs and exploit them to regulate the expression of their own genes. This frequently includes immune-regulatory transcription factors such as NF-κB, interferon regulatory factors (IRFs), or STATs, since they are activated in response to viral infection. Upon endogenization, the ability of LTRs to recruit transcription factors can be exploited by the host to regulate the expression of cellular genes. This cooption of regulatory elements is not a rare phenomenon, and it has been estimated that about 20% of all transcription factor binding sites in humans are found in HERVs and other transposable elements (87). In line with this, a meta-analysis of chromatin immunoprecipitation sequencing (ChIP-Seq) data sets identified about 800,000 transcription factor binding sites within HERVs (88).

Intriguingly, almost 90% of all HERVs represent so-called solo LTRs (89). These HERVs lost the prototypical retroviral genes gag, pol, and env due to homologous recombination of their flanking LTR sequences, leaving single LTR promoters in the genome. Due to their activation upon immune stimulation, ERV LTRs have already been termed “landing strips for inflammatory transcription factors” (90), and evidence for their role in regulating cellular immune responses is growing. One of the earlier reports comes from mice, where the insertion of a retroviral LTR increases the transcription of APOBEC3 (91). This deaminase introduces hypermutations in retroviral genomes, thereby restricting replication of mouse mammary tumor virus (MMTV) and potentially that of other viruses (92). The endogenized LTR is derived from a xenotropic mouse gammaretrovirus and harbors typical regulatory elements, including CAT and TATA boxes and an enhancer region. Mice harboring the LTR insertion show about 4- to 20-fold higher APOBEC3 mRNA levels than in those lacking it (91). These findings strongly suggest that the fixation of a retroviral LTR in the resistance gene APOBEC3 confers a selection advantage in virally infected mice, although direct in vivo evidence is missing.

Another example of an endogenous retroviral LTR modulating the expression of an immunity gene is the HERVP71A LTR (termed enhancer L), which acts as a distal enhancer for HLA-G in trophoblast cells (93, 94). This nonclassical MHC protein has been suggested to primarily prevent killing of placenta cells by NK cells, but it may also play a role in antiviral immune responses, as it is able to present intracellular peptides (95, 96).

In 2016, Chuong and colleagues performed a seminal study, deciphering the role of ERVs in IFN-mediated immune responses in a global and unbiased manner (97). Taking advantage of available ChIP-Seq data, they searched the human genome for transposable elements that show an increase in STAT1 and/or IRF1 binding upon IFN-γ stimulation and which may therefore contribute to the expression of ISGs upon viral infection. This approach identified 20 HERV families that are enriched for IRF1/STAT1 binding peaks upon IFN-γ stimulation. One notable example is the MER41 family of endogenous gammaretroviruses, whose members frequently harbor a so-called gamma-activated site (GAS) that is bound by STAT1. Notably, STAT1-binding MER41 elements were found in the vicinity of immunity genes such as AIM2, APOL1, IFI6, and SECTM1 (97). While AIM2 triggers an inflammasome-mediated immune response upon sensing of cytosolic dsDNA (98), APOL1 lyses intracellular parasites (99), IFI6 inhibits replication of several flaviviruses (100–103), and SECTM1 contributes to T-cell activation (104). In agreement with an important immune-regulatory activity of MER41 elements in vivo, their CRISPR/Cas9-mediated deletion reduced or abrogated the responsiveness of the adjacent immunity genes to IFN-γ (97).

The endogenous LTRs mentioned above all act as enhancers and may increase antiviral gene expression in response to viral infection and/or IFN stimulation. In some cases, however, ERVs have also been coopted as the primary promoters and/or transcription start sites (TSS) for immunity genes. One example is an ERV9 element harboring the TSS for the immunity-related GTPase family M protein (IRGM) (105). This large IFN-inducible GTPase eliminates mycobacteria by inducing autophagy (106). In this case, the insertion of an Alu retrotransposon initially disrupted the open reading frame of IRGM in the common anthropoid ancestor, but was subsequently resurrected by the insertion of the ERV9 transcription start site (TSS) (105). Interestingly, expression of two additional members of the family of large IFN-inducible GTPases is also regulated by ERV9 promoters. Using an RNA sequencing approach, we recently found that guanylate-binding proteins 2 and 5 (GBP2 and GBP5) are expressed under the control of ERV9 solo-LTRs, termed LTR12C, and can be further increased by IFN-γ stimulation and HIV-1 infection (107, 108). These two GTPases exert broad antiviral activity by inhibiting the virus dependency factor furin (109). Intriguingly, numerous LTR12C and related LTR12D elements are activated upon HIV-1 infection (107, 110), suggesting that they may be part of a larger network regulating antiviral immune responses. In line with this, expression of an LTR12C-SEMA4D fusion transcript is also increased in HIV-1-infected cells (107). SEMA4D (also called CD100) is a regulator of B-cell activation and may contribute to T-cell senescence in HIV-1-infected individuals (111, 112).

In the context of the current COVID-19 pandemic, the report of a retroelement (MIRb) that acts as promoter for an unusual isoform of ACE2 has received some attention (113). Since ACE2 is the primary receptor of SARS-CoV-2, its expression pattern and inducibility are important determinants of the tissue tropism and pathogenicity of this pandemic pathogen. However, the MIRb-ACE2 transcript turned out to be translated into a truncated and unstable form of ACE2 that most likely fails to mediate viral entry, as it lacks domains required for SARS-CoV-2 binding (113). In contrast to the canonical ACE2 isoform, the unusual MIRb-ACE2 transcript is strongly IFN inducible. However, its exact role in viral infections remains unclear.

In summary, retroviral LTRs have been coopted independently several times to regulate the expression of antiviral host factors, and it is very likely that additional examples remain to be discovered.

IMPORTANCE AND FUTURE DIRECTIONS

Our knowledge about endogenous retroviral elements involved in cellular immune responses is constantly increasing. Recent key findings include the discovery of retroviral Envs restricting retroviral infection in primates (56, 57), the identification of a SUMO-dependent pathway of antiviral immunity that involves sensing of ERV nucleic acids (15), and the description of additional retroviral promoter and enhancer elements regulating antiviral gene expression (93, 97, 107). These reports not only provide important insights into innate antiviral defense mechanisms, but also help to assess the role of ERV activation that is observed in a variety of infectious and noninfectious diseases. Furthermore, they strongly suggest that the host accepts possible risks associated with the cooption and activation of endogenous retroviruses (e.g., detrimental immune activation), and that the positive effects of ERVs usually prevail.

Nevertheless, several open questions remain. Although many studies report interesting associations of increased ERV activity with immune activation, causality is frequently not shown, and the relative contribution of ERVs to antiviral immune responses still remains unclear. Furthermore, several studies characterize antiviral mechanisms of individual ERVs in vitro, but do not investigate whether these antivirally active ERVs are activated upon infection in vivo.

Another important aspect of future studies will be to decipher the pathways that trigger the activation of specific ERV repeats or families upon viral infection. Are ERVs primarily activated by the cytokine storm in response to viral infection (97)? Are they derepressed because exogenous viruses counteract host restriction factors that suppress retroviral gene expression, such as TRIM28 (15, 23)? Why do viruses such as HIV-1 primarily activate specific families of ERVs (107, 110)? Answering these questions will shed light on the ERV immune network that is activated in response to infection and may uncover strategies that viruses have evolved to evade ERV-mediated immunity. Finally, cancer research has already demonstrated that artificial induction of ERV expression can boost antitumor immune responses, and it will be important to investigate whether similar beneficial effects can be achieved for the therapy of viral diseases.