A plan or pandemonium? The conundrum of retrotransposon activation in cancer

Abstract

Cancer arises from a convergence of genetic mutations and epigenetic alterations. Among the consequences of epigenetic deregulation is the aberrant activation of transposable elements, particularly retrotransposons, the most expansive and functionally impactful class. This review examines the paradoxical role of retrotransposons in cancer, emphasizing their dual capacity to support oncogenesis while simultaneously triggering innate and adaptive immune responses that may limit tumor growth. A central question is discussed: Is retrotransposon activation merely a stochastic byproduct of epigenetic instability, or could it represent an intrinsic layer of early tumor surveillance that inadvertently turns against the host as malignancy progresses?

Endogenous retroelements: a brief overview

Transposable elements are self-replicating DNA sequences that have propagated throughout genomes over evolutionary time and now comprise nearly 50% of the human genome [1]. Based on transposition mechanisms, they are classified into two major classes: DNA transposons, which move via a DNA intermediate, and the more abundant retrotransposons, also known as endogenous retroelements (EREs), which propagate through a copy-and-paste mechanism involving reverse transcription [2]. EREs are subdivided into LTR (long terminal repeat) and non-LTR elements. LTR retrotransposons, primarily represented by endogenous retroviruses (ERVs) characterized by genes encoding key viral proteins: gag (capsid), env (envelope), and pol (enzymes essential for replication), flanked by LTRs (Fig. 1). Non-LTR retrotransposons include long interspersed nuclear elements (LINEs) and short interspersed nuclear elements (SINEs). Full-length LINEs contain a 5′ untranslated region (UTR) with an internal promoter, a 3′ UTR with polyadenylation signals, and two open reading frames (ORF1 and ORF2) that encode proteins required for retrotransposition. In contrast, SINEs, such as the prominent Alu elements in the human genome, are short, non-coding sequences that depend on LINE-encoded proteins for their propagation [3].

Fig. 1.

Illustration of structural features of the major biotypes of endogenous retroelements in the human genome. An – poly(A) tail. For Alus: LM – left monomer; RM – right monomer; black box – A-rich linker

Over evolutionary time, most EREs have accumulated mutations that rendered them immobile [4]. Of the roughly 500,000 LINE-1 copies in the human genome, fewer than one hundred full-length elements, belonging to the evolutionarily youngest LINE-1 type, retain the structural integrity and the potential to retrotranspose [1, 5, 6]. Similarly, nearly 90% of ERVs lack all open reading frames and persist as solitary LTRs [7, 8]. In contrast to LINE-1s, even structurally intact proviruses are no longer capable of retrotransposition in the human genome.

The genomic fixation of EREs embodies a long-standing evolutionary negotiation between host defenses and parasitic elements seeking to survive. Current thinking holds that neutral or evolutionarily beneficial elements are retained, while potentially deleterious insertions are either eliminated by purifying selection or epigenetically silenced through DNA methylation and histone modifications [3, 9]. It has been proposed that the initial domestication of ERVs arose as a means to mitigate host-virus conflict, eventually providing a platform for key physiological innovations [10]. While full-length EREs are typically repressed in adult somatic tissues, ERE-derived transcripts detected in RNA-seq data [11] often arise from non-autonomous co-transcription of fragmented elements embedded within genes or other genomic regions. Despite structural damage, these remnants can retain intrinsic regulatory sequences, which allows their frequent co-option into regulation [12].

In cancer, the breakdown of the epigenetic silencing pathways leads to the reactivation of both autonomous retrotransposons and dormant ERE-derived regulatory elements. This reawakening has prompted significant interest in uncovering the mechanistic links between ERE activation and oncogenesis. Numerous comprehensive reviews have explored the diverse roles of transposable elements in cancer biology [13–19]. The present review synthesizes current insights into ERE activity in human cancer, highlights illustrative examples, and examines whether ERE activation is merely a byproduct of malignant transformation or a mechanism that has evolved to contribute to tumor suppression.

Ere-mediated transcriptional regulation of oncogenic pathways

One of the most prominent examples of the regulatory co-option of ERE sequences is their ability to function as primary or alternative promoters of coding genes, leading to the production of chimeric transcripts and proteins [12, 20]. ERE-derived chimeric transcripts comprise a significant portion of the human transcriptome [21]. In cancer, cryptic activation of ERE promoters can induce the expression of oncogenes, a phenomenon known as onco-exaptation. This process has been observed across multiple cancer types [18, 21–28] and is increasingly recognized as a widespread mechanism of oncogene activation. A recent comprehensive study by Jang et al. identified over 100 onco-exaptation events involving 106 distinct oncogenes across thousands of tumors [22].

The most well-characterized cases involve LTR elements. In Hodgkin’s lymphoma, for example, a THE1 subfamily of MaLR LTRs becomes derepressed, with a single member acting as a promoter for myeloid-specific proto-oncogene CSF1R, driving its aberrant expression and ultimately promoting the survival of malignant B cells [23]. In B-cell lymphoma, a fusion between the LTR2 element and FABP7 generates a chimeric transcript, whose product promotes tumor cell proliferation [24]. In melanoma and sporadically in other cancers, an intronic LTR drives the expression of a truncated ALK isoform, capable of inducing transformation in vitro and accelerating tumor growth in vivo [25]. While LINE-1s and Alu elements are likely less frequently implicated, recent findings have identified an AluJb-derived promoter essential for driving LIN28B oncogene expression in lung cancer [22]. A prognostically relevant L1PA2-derived isoform of SYT1 that enhances tumor cell invasion in vitro has been identified in bladder cancer cells [28].

The overall frequency of onco-exaptation remains incompletely defined, but specific ERE-derived gene isoforms appear to be relatively prevalent. For instance, an LTR-driven oncogenic isoform of ALK and an L1PA2-driven isoform of SYT1 have been identified in approximately 11% of melanomas and bladder cancers from TCGA, respectively [25, 28].

Intronic EREs can be incorporated into novel transcript isoforms through a process known as exonization [29]. In many cases, this is facilitated by splicing signals embedded within EREs. For example, pan-cancer analyses have revealed that ERE-derived alternative splicing events are widespread in cancer [30, 31]. Although the functional consequences remain incompletely understood, emerging evidence suggests that these events can generate non-functional RNA and protein isoforms [32]. In parallel, promoter activity from EREs can trigger antisense transcription, collectively disrupting the expression of canonical isoforms [32]. Unexpectedly, these transcriptionally activated EREs compromise the function of numerous tumor-promoting genes, a phenomenon that may be potentially widespread [32].

At the same time, EREs can function as transcriptional enhancers, modulating oncogenic pathways via their regulatory sequences [33–35]. For example, in colorectal cancer, LTR10-derived enhancers drive cancer-specific gene expression programs [33]. In acute myeloid leukemia, ERV/LTR elements act as enhancers with oncogenic potential; deletion of a single element or epigenetic silencing of the entire family impairs leukemic cell growth by inducing apoptosis [35].

While regulatory co-option of EREs does not depend on retrotransposon integrity, full-length elements, particularly LINE-1s capable of autonomous mobilization, may contribute to oncogenesis through their transposition activity, as discussed in the following section.

Ere-induced genome instability in cancer

Replication-competent LINE-1 elements, typically silenced in adult somatic tissues through epigenetic regulation, can become aberrantly reactivated in cancer [14]. This reawakening is often marked by LINE-1 hypomethylation, a feature associated with poor prognosis in multiple cancers, including breast, hepatocellular, and esophageal carcinomas [36–38]. Once reactivated, these elements can regain mobility, triggering insertional mutagenesis, a potent source of genomic instability. LINE-1 retrotransposition events occur frequently in epithelial tumors [39–45]. Rodriguez-Martin et al. recently identified tens of thousands of somatically acquired LINE-1 insertions across thousands of cancer genomes [39]. The prevalence of LINE-1 activity varies widely by tumor type, with esophageal adenocarcinoma exhibiting the highest insertional burden, followed by head and neck, and colorectal cancers [39]. The rate of retrotransposition appears to intensify during tumor progression, as metastatic lesions of pancreatic cancer harbor significantly more LINE-1 insertions than their primary counterparts [46].

LINE-1 mobilization is a powerful engine of genomic instability, capable of reshaping the cancer genome through chromosomal translocations, large-scale duplications, and megabase-scale deletions, representing the most prevalent form of somatic structural variation in specific cancer types [39]. Although LINE-1 retrotransposition events generally avoid protein-coding genes and lack recurrent hotspots [14, 39, 46], certain insertions, whether stochastic or selective, might promote oncogenesis. For instance, LINE-1 insertion into an intron of the proto-oncogene C-MYC has been linked to its amplification in breast cancer [47]. Likewise, insertion within the tumor suppressor APC disrupts gene function in colorectal cancer, potentially acting as tumor-initiating mutational events [48]. Less directly, LINE-1-induced megabase-scale deletions may lead to the loss of tumor suppressors such as CDKN2A or the amplification of oncogenes like CCND1 in cancers with elevated retrotransposition activity, thereby potentially promoting oncogenesis [39]. A specific outcome of LINE-1 activity is 3′-transduction, in which downstream sequences from the donor site are co-mobilized to new genomic loci [44]. These transductions can redistribute regions rich in open chromatin and transcription factor binding sites, potentially reshaping gene regulation [44], though the full functional impact remains to be elucidated.

Alu sequences can destabilize the genome through mechanisms independent of retrotransposition. Homologous recombination between Alu elements has been implicated in copy number alterations of the MYB oncogene in T-cell acute lymphoblastic leukemia, enabling clonal expansion via increased MYB dosage during leukemic transformation [49]. Similarly, Alu-mediated rearrangements within the BRCA1 locus result in pathogenic loss-of-function mutations of this critical tumor suppressor [50].

Genomic rearrangement is just one consequence of full-length ERE activation. ERE-encoded proteins can themselves be functionally linked to oncogenic processes.

Ere-encoded proteins in cancer

Proteins encoded by full-length ERE elements are frequently detected across a wide spectrum of cancers [13, 14, 19, 51–53]. Among them, LINE-1-encoded ORF1p, typically absent in normal somatic tissues, has emerged as a hallmark of neoplasia [14], with its circulating form proposed as a potential pan-cancer biomarker [54]. Approximately half of all cancers are immunoreactive for ORF1p [51], though expression levels vary within and across tumor types [13], likely reflecting its association with tumor progression. For instance, elevated ORF1p levels are observed in high-grade mesenchymal tumors, lymphomas, and pancreatic and astrocytic neoplasms compared to their low-grade counterparts [51] and correlate with the aggressiveness of pediatric malignant germ cell tumors [55]. Notably, subcellular localization may carry greater prognostic value than expression alone. In breast cancer, nuclear accumulation of ORF1p is linked to poorer survival and increased metastatic potential relative to cases with predominantly cytoplasmic expression [56]. Mechanistically, ORF1p expression has been causally linked to the activation of oncogenic signaling pathways in hepatocellular carcinoma [57]. In pancreatic cancer, ORF1p depletion reduces tumor growth in vitro, further supporting its functional role in tumor progression [46].

HERV-encoded proteins are also detected across a broad spectrum of human cancers [15, 19, 53, 58–63]. They are often detected in breast cancer [58, 64–69], in which HERV-K reverse transcriptase has been proposed as a prognostic marker [68, 69]. Rec and Np9, two proteins encoded from alternative reading frames of the human-specific HERV-K (HML-2) Env gene, are recognized as bona fide oncogenes. Rec has been shown to induce germ cell carcinoma in situ and to support tumor growth in vivo [70, 71]. Mechanistically, both Rec and Np9 disrupt transcriptional repression, particularly by inhibiting PLZF, leading to the derepression of the c-MYC proto-oncogene and subsequent promotion of cell proliferation and tumor progression [71, 72]. Rec also interacts with androgen receptor repressors, suggesting a potential role in hormone-sensitive cancers [73, 74]. Meanwhile, Np9 contributes to testicular tumorigenesis by interfering with proteasome pathways [75]. Beyond germ cell tumors, Rec and Np9 expression has been documented in melanoma, leukemia, and early-stage breast cancer, indicating their broader oncogenic potential [65, 76–78]. Experimental studies indicate that Rec facilitates the transition of melanoma cells to invasive stages [76], while Np9 promotes the expansion of leukemia stem cells [78].

In contrast to their oncogenic isoforms, full-length HERV envelope proteins, though associated with cancer progression [79], generally lack intrinsic transforming capacity. However, in breast epithelial cells, HML2 Env expression has been shown to activate oncogenic signaling pathways that enhance cell proliferation, migration, and invasion [80, 81].

One hypothesis, still lacking robust evidence in the cancer context, suggests that syncytins, the envelope proteins of HERV-W and HERV-FRD, may promote malignancy by mediating cell-cell fusion between cancer and adjacent normal cells, potentially fostering tumor progression [82, 83]. Additionally, full-length HERV-H Env contains immunosuppressive domains within the transmembrane region [84], which may facilitate immune evasion by contributing to the expansion of pluripotent immunoregulatory CD271⁺ cells [85]. Yet paradoxically, HERVs and EREs, more broadly, can also trigger innate and adaptive immune responses that compromise tumor fitness.

Ere-induced innate immune response in cancer

ERE-derived oligonucleotides can activate innate immune pathways with anti-tumor potential, owing to their unique immunomodulatory properties that mimic those of exogenous viruses. In infected cells, RNA or reverse-transcribed DNA of the virus is usually detected by pattern recognition receptors (PRRs), initiating pro-inflammatory signaling cascades that drive viral clearance [86]. ERE-derived polynucleotide structures can mimic viral nucleic acids, triggering analogous immune responses, a phenomenon known as viral mimicry [87]. In the cancer context, cytoplasmic sensing of ERE-derived nucleic acids induces the expression of type I interferon (IFN-I) responsive genes, which exert anti-proliferative effects and impair tumor fitness [88]. Specifically, ERE-derived double-stranded RNA (dsRNA) in cancer is recognized by cytoplasmic RIG-I-like receptors, including RIG-I and MDA5, and Toll-like receptors, such as TLR3, on endosomal membranes. At the same time, reverse-transcribed cDNA is sensed by cyclic GMP-AMP synthase (cGAS) [89] (Fig. 2).

Fig. 2.

Activation of the innate immune system by ERE-derived polynucleotides in cancer. Cytoplasmic sensing of ERE-derived polynucleotides activates innate immune responses, including the induction of type I interferons (IFN-I) and pro-inflammatory signaling pathways. Distinct types of nucleic acids are detected by dedicated sensing machinery: cytoplasmic RIG-I–like receptors (RIG-I and MDA-5), cyclic GMP-AMP synthase (cGAS), and Toll-like receptors

To date, ERE sensing in cancer has primarily been demonstrated in contexts where the immunostimulatory elements are unleashed either pharmacologically or via genetic ablation of their suppressors. These include inhibition of DNA methyltransferases in colorectal and ovarian cancer cells [90, 91], EZH2 in prostate cancer cells [92], LSD1 in ovarian cancer cells [93], PRMT and DDX3X in breast cancer cells [94], IDH1 in brain cancer cells [95], WEE in ovarian cancer cells [96], and ZNF638 in glioblastoma cells [97]. In vitro studies using human cancer cell lines further demonstrated that sensing of ERE-derived polynucleotides can directly impair tumor cell proliferation by inducing anti-tumor immunity [93, 95, 98].

These studies have demonstrated that all major ERE types can trigger viral mimicry in cancer cells, with the specific ERE and mechanism depending on the cancer context and pathway disruptions. Among these, dsRNA generated by bidirectional transcription of ERVs is the most consistently identified driver of antiviral responses across a wide range of cancer types, including breast, ovarian, colorectal, melanoma, and lung cancers [90–94, 99, 100]. On the other hand, in colorectal cancer cells treated with DNA-hypomethylating agents, dsRNA generated from inverted Alu repeats emerges as the dominant immunogenic signal [98]. Similarly, in pancreatic cancer cells, Alu-derived dsRNA can activate type I IFN responses and exert anti-tumor effects in vitro [46, 98]. LINE-1 can also form immunogenic dsRNA, triggering IFN signaling in HEK293 cells upon reactivation via HUSH complex depletion [101]. In brain cancer spheroids, both reverse-transcribed full-length HERVs and LINE-1 elements can form dsDNA, activating type I IFN pathways [95].

Endogenous dsRNAs derived from EREs, particularly Alu elements, are typically neutralized by the dsRNA-editing enzyme ADAR1, which catalyzes the deamination of adenosine to inosine (A-to-I RNA editing [102]. In many cancers, ADAR1 is dysregulated, resulting in the accumulation of unedited dsRNAs that are recognized by innate immune sensors [103]. However, in P53-mutant contexts, immunostimulation by dsRNA of Alu elements appear to be suppressed by LINE-1’s pORF1, mitigating anti-tumor immunity [46].

The effectiveness of ERE sensing in limiting the growth of established tumors remains uncertain. In some primary ovarian cancers, ERV expression correlates with the activation of antiviral defense pathways [91], suggesting a baseline level of viral mimicry may persist in select tumor types. However, to evade immune surveillance, cancer cells may suppress immunogenic EREs. When these restraints are lifted, as described above, ERE amplification can trigger antiviral signaling and elicit anti-tumor immune responses.

Pharmacological reactivation of tumor-specific EREs, alone or combined with immune checkpoint inhibitors, not only triggers dsRNA stress and antiviral signaling but also amplifies T cell-mediated immunity, driving increased infiltration of cytotoxic CD8⁺ T cells into the tumor microenvironment [91, 93, 95, 104, 105]. In melanoma patients receiving immune checkpoint blockade, elevated expression of viral defense genes correlates with durable clinical responses, likely due to the activation of cytotoxic T lymphocytes [91]. Although the interplay between innate and adaptive immunity against EREs in these contexts remains unclear, accumulating evidence indicates that activated EREs can directly stimulate adaptive immune responses.

Ere-induced adaptive immune responses in cancer

Translated ERE sequences have significant potential to encode highly immunogenic peptides, exhibiting key features of effective antigens that can elicit adaptive anti-tumor immune responses, mainly through T cell activation [106, 107]. For such responses to be initiated, cancer-associated antigens must be presented on the cell surface via major histocompatibility complex (MHC) molecules, specifically MHC class I (HLA-I in humans). These antigens are recognized by cytotoxic CD8⁺ T lymphocyte receptors (TCRs), leading to T cell activation and subsequent targeted elimination of tumor cells.

Recent proteogenomic approaches have helped to identify HLA-I-bound peptides translated from ERE sequences, which are uniquely expressed (neoantigens) or overexpressed (tumor-associated antigens) in both solid and hematological malignancies [11, 108–116], suggesting ERE-based anti-tumor CD8⁺ T cell surveillance. These antigens can arise from autonomously expressed, intact elements, as well as from co-transcription of ERE sequences with adjacent genomic regions, such as through exonization events (Fig. 3).

Fig. 3.

Activation of cytotoxic T lymphocytes by ERE-derived peptides. Peptides translated from autonomously transcribed full-length EREs, or from read-through EREs including chimeric transcripts incorporating ERE sequences and adjacent genomic regions (e.g., ERE-gene co-transcripts), can be presented on HLA-I molecules at the cell surface. These peptide–HLA-I complexes are recognized by TCRs on CD8⁺ cytotoxic T lymphocytes, triggering their activation. Once activated, CD8⁺ T cells can directly eliminate tumor cells that express ERE-derived antigens

T cells specific to HERV antigens have been detected in patients with a range of cancers, including myeloid malignancies, breast, kidney, colorectal cancer, seminoma, and melanoma [64, 108, 111, 114, 117–122]. Importantly, HERV-specific T cells have demonstrated potent cytolytic activity against tumor cells [64, 109, 114, 115, 121, 122]. Notably, HERV-K-specific cytotoxic T cells are able to lyse breast cancer cells expressing the HERV-K envelope protein [64]. Furthermore, engineered chimeric antigen receptor (CAR) T cells targeting HERV-K Env have shown potent cytotoxicity against breast cancer cell lines in vitro and effectively suppress tumor growth in vivo [80]. Clinically, sustained tumor regression in patients with advanced metastatic renal cell carcinoma following allogeneic stem cell transplantation has been associated with cytotoxic T cell responses against a peptide derived from a hypomethylated HERV-E element [118]. It has been shown that non-canonical splice junctions between coding exons and transposable elements can give rise to immunogenic peptides that elicit CD8⁺ T cell responses in patients with lung cancer [116]. Collectively, these findings underscore the promise of EREs as immunotherapeutic targets and their capacity to drive potent anti-tumor immune responses, even in advanced disease settings.

The mechanisms by which EREs stimulate adaptive immune responses in cancer remain incompletely understood. Although EREs are inherently foreign in origin, their expression during thymic development and early ontogeny suggests they may induce central tolerance. Indeed, Env peptides could be recognized as self-antigens, potentially shaping both T cell repertoires and humoral immunity [123]. One possibility is that immune tolerance is disrupted in the tumor microenvironment, activating ERE-specific responses. Alternatively, some ERE-derived antigens may evade central tolerance by being excluded from thymic expression or restricted to developmental niches. It is also plausible that tumor-reactive T cells, capable of recognizing ERE-derived peptides, escape negative selection and persist as a latent reservoir of anti-tumor immunity [124].

The discovery of the immunogenic potential of EREs in cancer has emerged as both unexpected and promising. As non-mutational antigens, EREs offer a unique opportunity to induce anti-tumor immune responses in cancers with low mutational burden, where immunity driven by mutation-derived antigens is limited. Although the clinical significance of natural ERE-specific immune responses remains to be fully elucidated, they may be especially effective in early tumor development, before immune evasion takes hold.

Eres in early vs late cancer progression: a hypothetical perspective

With growing insight into the role of EREs in cancer initiation and progression, a central question arises: Is their activation simply a stochastic consequence of global epigenetic dysregulation, or might it reflect an evolutionarily conserved mechanism to suppress malignancy? Unlike most genomic elements, transposons are of foreign origin, and their persistence in the genome likely reflects evolutionary trade-offs that favor the retention of either neutral or beneficial elements [3]. These ancient genomic negotiations may resurface during oncogenesis as an adaptive safeguard against malignant progression. It may be most informative to consider ERE activity in relation to the sequential stages of malignant transformation. Widespread ERE activation due to epigenetic disruption may induce genomic instability and activate immune responses that eliminate emerging pre-cancerous clones. However, if such a clone survives by evading both molecular and immunological surveillance, specific EREs may, stochastically or selectively, contribute to tumor progression (Fig. 4 summarizes reported cases across specific cancer types into a hypothetical model illustrating the host-protective versus tumor-promoting roles of EREs). Some EREs may play both roles from the outset, with the balance gradually shifting toward a pro-tumorigenic role as malignancy progresses.

Fig. 4.

Schematic representation of the potentially dual role of ERE sequences in cancer initiation and progression. In early malignant transformation, the activity of EREs may promote the elimination of pre-cancerous clones through LINE-1-induced genotoxic stress, sensing of ERE-derived oligonucleotides, engagement of adaptive immunity against ERE-encoded epitopes, and the generation of non-functional oncogenic isoforms via aberrant splicing or promoter activity. At later stages, however, EREs may shift from barriers to facilitators of tumor progression, contributing to malignancy through disruption of tumor suppressors, onco-exaptation and oncogenic enhancers, and, in specific contexts, through oncogenic viral proteins or immune evasion mediated by HERV-derived immunosuppressive domains

In early malignant transformation, the activity of intact EREs and their remnants may promote the elimination of pre-cancerous clones by inducing genotoxic stress through LINE-1 mobilization, triggering innate immune sensing of ERE-derived oligonucleotides, engaging adaptive immunity against ERE-encoded epitopes, and generating non-functional oncogenic isoforms through aberrant splicing or promoter activity. At later stages, however, EREs may transition from barriers to facilitators of tumor progression, contributing to malignancy through disruption of tumor suppressors, onco-exaptation and oncogenic enhancers, and, in specific contexts, through oncogenic viral proteins and immune evasion via HERV-derived immunosuppressive domains.

Emerging evidence suggests that EREs act as early sensors of disrupted cellular homeostasis. Activation of EREs, particularly LINE-1 elements, is likely an early event of malignant transformation. LINE-1 ORF1p has been detected in a range of precancerous lesions, including serous tubal intraepithelial carcinomas (precursors to high-grade serous ovarian cancer) [125], in neoplasia that give rise to pancreatic ductal adenocarcinoma [51], in epithelial precursors of ovarian carcinoma [52], and in adenomas that precede gastrointestinal cancers [126, 127]. Although full-length LINE-1 expression is generally tolerated in established tumors, its broad genotoxic effects may create exploitable vulnerabilities during the early stages of malignant transformation [39, 128, 129]. Activated LINE-1s can disrupt genomic stability by creating complex rearrangements and the loss of centromeres and telomeres [39]. To counteract this threat, the tumor suppressor p53 acts as a barrier, restricting the growth of cells with aberrant LINE-1 activity [130]. Upon LINE-1 activation, p53 induces growth arrest and activates interferon signaling, a mechanism that may selectively eliminate damaged clones [129]. The frequent expression of LINE-1 ORF1p in TP53-deficient tumors [51] suggests that the loss of this tumor suppressor enables emerging clones to tolerate LINE-1-induced stress better, facilitating their survival and progression toward malignancy.

Additional barriers to transformation may be imposed by ERE-derived promoters and ERE-mediated splicing events. While specific ERE promoters are subjects of onco-exaptation, others may support latent tumor suppression, preserving the cell’s functional integrity. For example, LTR9-ERV elements can drive the expression of gene isoforms with pro-apoptotic functions under physiological conditions [131]. A notable example includes an isoform of TP63, transcribed from an upstream LTR9-ERV promoter in normal human testis [131]. Reactivation of this promoter induces apoptosis in testicular cancer cells [132]. In established tumors, EREs can generate non-functional transcripts from pro-tumorigenic genes, a process that may confer a selective disadvantage to pre-cancerous clones [32].

From an immunological perspective, ERE-driven innate immune responses may represent an early, evolutionarily conserved alarm system designed to eliminate pre-malignant cells. In support of this, studies in blind mole rats have shown that reverse-transcribed ERVs form RNA:DNA hybrids in pre-malignant cells activate the cGAS–STING pathway and trigger IFN-I-dependent necrotic cell death [133]. Moreover, ERE-mediated antitumor adaptive immune responses can be remarkably potent and may remain active even in established human tumors. Notably, high levels of ERE expression are often associated with increased immune cell infiltration and improved clinical outcomes [113].

During tumor progression, cells must balance the oncogenic gain of ERE activation with its immunogenic cost. These trade-offs, however, remain poorly understood. For instance, it remains unclear to what extent ERE exonization contributes to oncogenesis, what immunogenic burden it imposes, and how HERVs balance antigenicity with immunosuppressive potential. Tumors have the capacity to evade immune detection, for example, by losing HLA-I molecules, dampening T-cell activity through inhibitory checkpoint receptors [134], and suppressing immunogenic EREs [46]. Some EREs may be pleiotropic, with their net impact becoming more pro-tumorigenic as cancer progresses. In kidney cancer, for example, HERVs may play both agonistic and antagonistic roles in innate sensor immunity, with the latter linked to an immunosuppressive microenvironment [108]. Under such conditions, immunogenic EREs may lose their immune-stimulating potential, while their pro-tumorigenic functions remain intact.

Paradoxically, it is the evolutionarily younger EREs, such as higher primate-specific LINE-1s, Alus, and HERVs, that show the strongest association with innate immune activation in human cancers [113]. Moreover, ERE-mediated switching to dysfunctional isoforms of pro-tumorigenic genes frequently involves Alu and LINE-1 subfamilies [32]. This raises the intriguing possibility that these elements were evolutionarily co-opted in an accelerated manner to promote the elimination of precancerous clones. This may have helped to enhance population-level fitness in large anthropoid primates, which face the unique challenges of rapid phenotypic evolution.

Future directions

Looking ahead, several key questions merit focused investigation to advance both basic understanding and translational applications:

• (i)Are there recurrent, functionally relevant patterns of ERE activation across distinct tumor types, molecular subtypes, and patients?
• (ii)When is locus-level resolution necessary, and when is a biotype-level approach sufficient?
• (iii)How does the immunogenicity-oncogenicity trade-off evolves during tumor progression?

To date, no consensus exists, as ERE activity remains highly variable both across and within tumor types [113, 135, 136]. Focusing on shared, recurrent events could help distinguish stochastic activation from selected, functional deregulation. Furthermore, in the context of innate immune sensing, locus specificity may be less critical, as recognition likely depends on conserved sequences or structures. As discussed above, viral mimicry induced by global ERE derepression through epigenetic therapy can elicit anti-tumor responses. Equally, immunogenic peptides produced by evolutionarily young ERE species are likely redundant and may not require locus-specific resolution.

In some contexts, targeting EREs at the family level may be sufficient to impair tumor growth: for instance, silencing LTR2Bs, which are enriched in oncogenic enhancers [35], or activating tumor-suppressive L1H elements [128] using CRISPR-mediated approaches in leukemic cells. However, locus specificity becomes essential when considering splicing events and, more generally, ERE-overlapping chimeric transcripts. In the case of LTRs, such transcripts show locus specificity independently of broader family activation [137]. However, specific subfamilies of Alus, MIRs, and LINE-1s are recurrently incorporated into ERE-containing transcripts across diverse cancer types [32].

To explore these questions in greater depth, high-resolution ERE mapping using long-read sequencing and improved genome assemblies will be essential. Single-cell and spatial transcriptomics could uncover context-specific ERE activation linked to immune infiltration and genomic instability. Advancing computational methods to track ERE-derived epitopes and TCR clones may reveal conserved features of immunogenicity. Moreover, characterizing ERE dynamics across tumor progression could clarify how the trade-off between immunogenicity and oncogenic potential evolves over time. This will likely require profiling tumors from pre-malignant to metastatic stages, alongside matched normal tissues, as early fitness-limiting events may be undetectable in advanced disease. While much remains to be learned, ERE-focused research may significantly contribute to the collective effort to identify cancer vulnerabilities.

Conclusions

Although growing evidence implicates EREs in cancer progression, an unresolved question remains: Could they also help prevent malignant transformation by activating immune responses and counteracting oncogenic pathways? From this perspective, rather than being mere genetic hitchhikers of a global perturbation, their involvement in the early stages of oncogenesis may represent an adaptive mechanism, one that has been evolutionarily selected to antagonize malignancy. The reported pro-tumoral consequences may reflect a slip from the razor’s edge, where the delicate balance between high-benefit and high-risk strategies is crossed. Unraveling the context-dependent and biotype-specific behavior of EREs in cancer is key to understanding the evolutionary negotiation between these genomic parasites and their hosts. But one truth about EREs emerges ever clearer: the wolf may lose its teeth, but never its nature.

Abbreviations

LINE

long interspersed repeats

SINE

short interspersed repeats

LTR

long terminal repeats

ERV

endogenous retroviruses

Alu

a short (approximately 280 bp) element

HERV

endogenous retroviruses in human

RIG-like receptors (RLRs)

proteins that detect viral RNA in the cytoplasm and trigger innate immune responses

TLR

toll-like receptor

TCR (T cell receptor)

a protein on the surface of T cells that recognizes specific antigens presented by other cells, triggering an immune response

MHC-I

a major histocompatibility complex, a protein complex found on the surface of nearly all human cells. It presents small fragments of proteins (peptides) from inside the cell to the immune system, allowing CD8 +  T cells to detect and destroy infected or abnormal cells, such as cancer cells

HLA-I

(Human Leukocyte Antigen class I) is MHC-I in human

Cytotoxic CD8⁺ T cells

are immune cells that recognize and kill infected or cancerous cells by detecting antigens presented on HLA-I molecules

Author contributions

EZ wrote the manuscript and prepared the illustrative figures.

Funding

EZ receives funding from « Agence Nationale de la Recherche » (grant ANR-23-CE14-0017-02).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.