Transposable elements in hematopoietic stem cells upon aging and myeloid malignancies

Abstract

Transposable elements (TEs) constitute nearly half of the human genome and profoundly influence hematopoietic stem cell (HSC) biology. In this review, we synthesize current evidence demonstrating that TEs exert dual and context-dependent roles in HSCs during steady-state hematopoiesis, stress responses, aging, and leukemogenesis. Under basal conditions, tightly controlled TE activity can be beneficial for HSC biology, through the induction of intrinsic type I interferon signaling and a fine-tuned control of gene expression. However, dysregulated TE activation upon stresses and aging can undermine HSC self-renewal, impair genomic integrity, and drive age-associated hematopoietic decline. TEs also play a dual role in leukemogenesis. Derepression of transcription factor motifs within TEs can activate oncogenic programs, while TE-derived nucleic acids can simultaneously elicit antiviral and DNA damage responses that trigger anti-tumoral p53- or interferon-dependent growth arrest or apoptosis. The balance between these pro- and anti-tumoral effects remains an open question, likely shaped by cellular context, TE subtypes, and the magnitude of TE expression. Finally, we discuss the potential to therapeutically modulate TE activity. Understanding TE dynamics in HSCs offers new opportunities for mechanistic insight and clinical innovation in myeloid malignancies.

Introduction

Transposable elements (TEs) are repeated sequences that constitute approximately 50% of both the human and mouse genomes. Through their capacity for mobilization and replication, they play a significant role in shaping genome architecture. TEs are broadly categorized into two major classes based on their transposition mechanisms [1–3]. Class I elements, or retrotransposons (RTEs), transpose via a “copy-and-paste” mechanism that involves transcription into an RNA intermediate, which is subsequently reverse-transcribed into complementary DNA (cDNA) for reintegration into the genome. Retrotransposons are further subdivided into long terminal repeat (LTR) and non-LTR elements. LTR retrotransposons include endogenous retroviruses (ERVs), which are remnants of ancient exogenous retroviral infections that became fixed in the germline. Non-LTR retrotransposons comprise three major families: long interspersed nuclear elements (LINEs), with LINE-1 (L1) being the most abundant; short interspersed nuclear elements (SINEs), such as Alu elements; and SINE–variable number tandem repeat–Alu (SVA) elements. Class II elements, or DNA transposons, mobilize using a DNA intermediate. These are subdivided into two main subclasses: terminal inverted repeat (TIR)-containing transposons, which are mobilized through double-stranded DNA cleavage and reintegration, and rolling-circle transposons (helitrons), which replicate via a single-stranded DNA intermediate through a rolling-circle mechanism.

Due to frequent incomplete reverse transcription, truncation, internal rearrangements and accumulation of mutations over time, only a very small fraction of TE copies is still able to mobilize. L1s are the only autonomously active TE in humans. However, only around 80–100 out of 500,000 copies present in the human genome are considered to be able to mobilize [4,5]. These functional L1 elements are among the most evolutionary recent subfamilies of L1 in the human genome, so-called L1-Hs for human-specific L1. They are full length and encode two open reading frames (ORFs): ORF1 encodes a nucleic acid chaperone protein, while ORF2 encodes endonuclease and reverse transcriptase required for retrotransposition. In contrast, SINEs are non-autonomous elements that lack the coding capacity for self-mobilization and hijack the L1-encoded enzymatic machinery for their propagation. Finally, only very few ERVs retain a complete structure made of key retroviral genes (gag, pol, and env) flanked by LTRs. Although ERVs account for about 8% of the human genome, nearly 90% are present as solitary LTRs because recombination between their LTRs after insertion commonly deletes the internal coding regions while preserving the LTRs that contain the original cis-regulatory elements [6].

When TEs are expressed, they can induce DNA damage and constitute a great source of genomic instability (Fig. 1A). Indeed, integration of new TE copies into the genome involves double strand breaks (DSBs). However, even an abortive retrotransposition, or the simple expression of the L1 endonuclease, is sufficient to induce DSBs [7]. TEs can also insert and disrupt tumor-suppressor genes [8].

Fig. 1.

Consequences of TE derepression. When TEs are expressed, they can: A propagate into the genome and induce double strand breaks, leading to TP53-mediated cell cycle arrest, senescence or apoptosis; B produce double strand (ds) RNA, RNA-DNA hybrids, or cytosolic cDNA that will be recognized as pathogens by RLR or cGAS/STING pathway and trigger the expression of type-I and III interferon (IFN) genes through activation of interferon response factor (IRF) or NFkB pathways. C In the absence of transcription, TEs can also rewire the 3D conformation of the genome, play the role of cis-regulatory elements by recruiting transcription factors (TFs), and provide alternative splice sites or alternative promoters. This image was created with Biorender.com and inspired by [15]

Expressed TEs also serve as a source of endogenous double-stranded (ds) RNA, RNA: DNA hybrids or cytosolic cDNA that are sensed as endogenous genomic parasites and detected by distinct innate immune pathways. The RIG-I–like receptors (RLR) family RIG-I and MDA5 recognize viral RNA species, particularly dsRNA, and signal through MAVS, whereas cGAS detects cytosolic DNA (and some RNA: DNA hybrids) and activates the adaptor STING [9–11]. Both pathways, trigger the production of intereron type-I (IFN-I) and type-III, and activation of innate antiviral immune pathways, in a process termed “viral mimicry” [11] (Fig. 1B).

In cancer cells, TEs can also serve as a source of neoantigens, either through peptides derived directly from canonical ERV ORFs [12] or from noncanonical splicing events between TEs located within introns and exons [13], leading to the activation of specific T cells. Additionally, neoantigens may arise from chimeric transcripts generated through TE promoter exaptation [14]. Indeed, even in the absence of transcription and mobilization, TEs are also recognized as major contributors of the three dimensional organization of the genome, and of gene regulatory networks [6, 15] (Fig. 1C). The loss of epigenetic control of pre-existing TE copies can qualitatively and quantitatively affect gene expression. TEs are an important source of promoters that may lead to the expression of oncogenes in different types of cancers [14, 16]. Often located in introns, TEs, particularly LINEs and Alu, represent a rich source of both acceptor and donor splice sites able to generate alternative splicing events. This can change protein localization and function. Recurrent TE-derived noncanonical alternatively spliced transcripts, notably at cancer-associated genes, are observed in different cancer tissues [17]. TE exonizing isoforms generated by mRNA splicing contribute to the production of stable proteins with functions distinct from canonical isoforms, expanding by this way the diversity of the human proteome [18]. Alternatively, transposable elements can regulate the abundance of the encoded protein by modulating the inclusion of nonsense mediated RNA decay switch exon [19].

When located in introns, TEs, particularly the most recent subfamilies of L1s, were also shown to attenuate the host gene expression by recruiting H3K9me3 via the human silencing hub (HUSH) complex [20]. Analysis of ENCODE data have revealed that TEs supply around 25% of human cis regulator elements, in a cell- and tissue-specific fashion [21, 22]. By reanalyzing ChIP-seq data for 26 transcription factors (TFs) in human-mouse pair of cell lines, Sundaram et al. further showed that up to about 20% of all TF binding sites originated from a TE [23]. Andrews et al. further showed that more than 85% of primate-specific TF binding sites are derived from TEs [24]. TE members of a same subfamily, scattered throughout the genome, and enriched for the same TF binding site motifs, could regulate different genes involved in the same pathway in a time and cell-specific manner depending on chromatin accessibility and transcription factors availability, eventually accelerating the evolutionary response to environmental challenges [6, 25].

TE expression is stringently controlled through epigenetic mechanisms. Heterochromatin, through DNA methylation and histone post-translational modifications such as di- and tri-methylation of histone H3 lysine 9 (H3K9me2/3), plays a central role in the repression of TEs [15, 26]. It has been proposed that these repressive epigenetic marks may have evolved to silence TEs during their expansion in early eukaryotic genomes [27, 28]. Additionally, members of the PRC2 complex are involved in H3K27me3 deposition at TEs in a backup silencing process in hypomethylated context [29]. Described for a long-time as junk DNA, TEs are gaining more and more interest, especially in aging and cancer that are characterized by a loss of heterochromatin (H3K9me3 and/or DNA methylation) at the global genome level [30, 31]. Overexpression of TEs upon aging is associated with increased senescence and low-grade inflammation (called inflammaging) in human fibroblast cell lines [32]. A recent report has shown that HERV-K, the most recently integrated human endogenous retrovirus, can reactivate and form retrovirus-like particles, inducing senescence and aging phenotypes in young cells—an effect that can be attenuated by reverse transcriptase inhibitors [33].

Through their ability to modify the cell transcriptome and to induce genomic instability and an inflammatory context, TEs can exert an oncogenic activity. However, recent studies suggest that evolution has preserved certain active LINE-1 and ERV elements because their activation induces DNA damage and a viral mimicry response that triggers anti-tumor immunity. Tumor cells that develop mechanisms to prevent TE-induced viral mimicry, or DNA damage and TP53 activation may have a survival or proliferation advantage. There is indeed a positive correlation between LINE-1 expression and TP53 mutations [34] suggesting that loss of TP53 is required for proliferation and survival of L1 expressing colorectal carcinoma cells. Alternatively, the overexpression of the H3K9 methyltransferase SETDB1, is common in cancers and plays a key role in tumorigenesis through the epigenetic silencing of TEs [35, 36].(Fig. 2).

Fig. 2.

The dual role of TEs in oncogenesis. Depending on the nature of and/or their level of expression, TEs may induce both a pro- or an anti-tumoral response. Cells repressing TE expression or TE-induced anti-tumoral program may have a proliferative and/or survival advantage and participate to oncogenesis. This image was created with Biorender.com

Many TEs have been co-opted by the host genome and now function as regulatory elements [21]. This is particularly evident in the immune system, where TEs are thought to have played a significant role in facilitating adaptation to evolving infectious challenges [21,37–40], and TE activity in different mature immune cells have been reported [41]. Hematopoietic stem cells (HSCs), which sustain lifelong blood cell production, are central to immune cell generation. Importantly, HSCs can also serve as a reservoir for leukemic stem cells, which are capable of driving disease progression, therapy resistance, and relapse in leukemia. Therefore, dysregulation of TEs in HSCs may have profound and lasting consequences for hematopoiesis, potentially contributing to immune system aging, clonal expansion of mutated cells and the development of hematological malignancies.

In this review, we describe the mechanisms regulating TE expression in HSCs during aging and in response to stress. We also discuss the dual role of TEs in myeloid malignancies—as drivers of oncogenesis on one hand, and as enhancers of cancer cell immunogenicity on the other—and discuss the clinical implications of these findings.

Expression and roles of transposable elements in HSCs during aging and stress

Functional and molecular changes in aging HSCs

Aging significantly impairs HSC function, driving both quantitative and qualitative changes that contribute to hematopoietic and immune decline in the elderly. While adult HSCs remain largely quiescent in the bone marrow, this state is disrupted over time by stressors such as inflammation and environmental cues, promoting aging [42,43]. Paradoxically, the number of phenotypically defined HSCs increases with age, but their regenerative and self-renewal capacities diminish. Aged HSCs show impaired engraftment, defective differentiation, and reduced ability to sustain hematopoiesis, especially under stress or post-transplantation. This includes a shift toward myeloid-biased differentiation and diminished lymphoid output, leading to immunosenescence and chronic inflammation (“inflammaging”). Elevated inflammatory cytokines (e.g., IL-1β, TNF-α) further exacerbate myeloid skewing and HSC exhaustion. Additionally, during aging, HSCs accumulate DNA damage and somatic mutations, promoting clonal hematopoiesis (CH). CH is triggered by mutations in genes commonly mutated in myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPN), and acute myeloid leukemia (AML) and is associated with an increased risk of developing these hematopoietic malignancies, as well as cardiovascular and neurodegenerative diseases [44].

Perturbed chromatin as a driver of HSC aging

The aging phenotype of HSCs is reversible through a pluripotent intermediate, implicating epigenetic alterations as a key driver [45]. Different studies in mouse and human HSCs have characterized epigenetic changes associated with HSC aging. Changes in DNA methylation and histone modifications affect genes involved in HSC identity and differentiation, and genes frequently dysregulated in AML, supporting the hypothesis that age-related epigenetic remodeling predisposes to leukemogenesis [46–48].

Several studies suggested that global heterochromatin erosion contributes to aging by derepressing normally silenced genomic regions, including TEs, and generating aberrant transcriptomes [31]. An alternative model posits that aging involves not global loss but redistribution of chromatin modifiers, linked to the cumulative repair of DSBs arising from environmental stress or replication errors [31]. This chromatin relocalization process exemplifies antagonistic pleiotropy: efficient DSB repair preserves genomic stability and fitness during early life but promotes chromatin alterations that drive aging later.

In HSCs, loss of H3K9me3, for instance in mouse models knocked-out (KO) for Setdb1 or for Suv39h1 and Suv39h2 H3K9 methyltransferases, leads to a loss of identity, and an accelerated aging [49, 50]. We previously showed that decreased H3K9me2 and H3K9me3 in human and mouse HSCs occurs with age [51, 52]. The decrease in H3K9me3 in aged HSCs, due to increased miR-125b which suppresses Suv39h1 expression, is associated with loss of B cell differentiation [51]. Another study reported distinct subnuclear distribution patterns H3K9me3-labeled heterochromatin in hematopoietic stem and progenitor cells (HSPCs) upon aging, with no reduction in their global level of protein expression as assessed by immunofluorescence [53]. Aging may result from cumulative stress exposure. Upon stimuli such as cytokines, HSPCs undergo an initial phase of global, non-specific chromatin decompaction during differentiation onset [54]. Repeated chromatin remodeling throughout life may induce heterochromatin loss or redistribution, contributing to aging phenotypes. We showed that a single ionizing radiation exposure in mice leads to long-term H3K9me3 loss in HSCs and accelerates aging features [55]. We confirmed these data by showing a durable loss of H3K9me3 in HSCs from a mouse model of chronic inflammation mediated by serial injections of low doses of LPS, which has been shown to mimic the low-level chronic inflammation observed with age [56]. Likewise, replicative stress or Il1β administration, that induce HSC-premature aging, are associated with loss of DNA methylation [47,57].

TE expression in HSCs upon aging and stress-induced premature aging

TE expression in HSCs

Barbieri et al. used quantitative reverse-transcriptase PCR (qRT–PCR) on FACS-sorted HSCs or progenitor populations to assess the expression levels of TEs [58]. Primers were designed to detect various regions of the evolutionarily recent mouse L1 subfamilies—L1Md_A, Tf, and Gf. Remarkably, L1Md mRNA levels in HSCs were found to be comparable to those observed in embryonic stem cells and significantly elevated compared to multipotent progenitors and myeloid progenitors. Additionally, HSCs exhibited substantial expression of other TEs, including the SINE B1 family and endogenous retroviral LTRs such as IAP. Small but detectable amount of ORF1 protein (ORF1p) was also detected in HSCs at the basal level. A transgenic mouse model carrying an engineered human L1 element with a GFP-based retrotransposition reporter cassette was utilized to show that, in addition to their expression, L1 retrotransposition events could be detected in HSCs, although at relatively low frequencies under basal conditions. Retrotransposition activity was significantly higher in HSCs than in myeloid progenitor populations. Differences in H3K9 methylation and chromatin condensation between HSCs and their progenitors [59] may account for the reduced expression of TEs observed during HSC differentiation.

Using a curated reference of 14,968 human ERV (HERV) functional units, Alcazer et al. (2022) conducted a comprehensive analysis of HERV expression across human hematopoietic populations. Their study demonstrated that the HERV transcriptome effectively characterizes HSCs and distinguishes them from both progenitor and mature cell populations [12]. Chromatin accessibility, a key driver of cell differentiation, has been shown to capture cellular identity more accurately than gene expression in blood cells [60, 61]. Alcazer et al. (2022) further observed that transcriptionally active HERV regions located in distal regulatory elements enable the discrimination of hematopoietic cell populations. In a complementary study, Grillo et al. analyzed ATAC-seq profiles from six distinct HSC, progenitor, and mature cell populations isolated from human cord blood. They found that specific TE subfamilies—particularly endogenous retroviral LTRs—are consistently enriched in the open chromatin of HSCs/HSPCs. Notably, long-term and short-term repopulating HSCs exhibited highly similar accessible TE profiles, which closely resemble those observed in embryonic stem cells but differ markedly from mature hematopoietic cells [62].

Although TEs are highly expressed in HSCs, their expression further increases in both humans and mice with aging and under stress conditions [46,51,53,55,56,58,63]. Indeed, sublethal total body irradiation in mice results in long-lasting changes in the expression of recent active L1Md family members, as well as increased levels of ORF1 protein and L1 retrotransposition activity [55, 58]. Furthermore, TE expression in HSCs is induced by chemotherapeutic treatments [63], repeated injections of LPS [56], and cellular senescence [64]. This upregulation is linked to chromatin decondensation resulting from reduced heterochromatin integrity.

Dual roles of TEs in HSCs (Fig. 3)

Fig. 3.

The dual role of TEs in HSCs. Through induction of DNA damage, inflammation or transcriptomic changes, TEs are shown to play a role both in the physiology and the physiopathology of HSCs. As mentioned in Fig. 2, the nature of the derepressed TEs and/or their expression level at basal state or upon stresses may play a role in this balance. This image was created with Biorender.com and adaped from 79

At the basal state, TEs have been shown to play a beneficial role in HSCs/HSPCs. Like other stem cell types—and unlike more mature cells—HSCs exhibit robust, IFN-independent expression of IFN-stimulated genes (ISGs), under sterile condition. This intrinsic antiviral response allows immune surveillance of both TEs and exogenous pathogens [58,65].

During embryonic hematopoiesis, TEs contribute to the formation of immune-stimulatory dsRNAs, which activate RIG-I-like receptors. Activation of these receptors triggers intrinsic inflammatory signaling that is essential for the emergence of HSCs and HSPCs [66–70]. Consistently, deficiency of RIG-I or MDA5 in zebrafish and mice impairs inflammatory signaling and disrupts HSPC development [70]. This function seems to be conserved in adult HSCs. Indeed, recently, Martinez et al. [71] demonstrated that the HSC intrinsic TE–ISG pathway is controlled by the transcription factor CUX1 and is required for the enhanced HSC self-renewal, an effect which can be abolished by blocking TE reverse transcription or STING signaling.

Derepression of ERVs and LINEs, leading to activation of the cGAS/STING pathway and ISGs also plays a key role in activating HSCs and erythropoiesis during pregnancy and after bleeding, a mechanism conserved in both mice and humans [72].

Finally, expression of TE-derived dsRNAs in HSPCs signals through TLR3 to prevent their phagocytosis by macrophages [73], while increased TE expression in senescent human HSPCs may contribute to their clearance via activation of immune pathways [64].

Even in the absence of expression, TEs can influence HSCs, notably by regulating gene expression. We recently showed that enrichment of H3K9me3 at intronic L1Mds containing NF-κB binding motifs is essential for proper HSC gene expression and self-renewal capacity [55]. Enrichment of H3K9me3 at these elements in gene bodies may prevent their expression and transcriptional interference. Even in the absence of L1 transcription, L1-mediated recruitment of H3K9me3 in gene bodies could slow-down RNA Polymerase II elongation rate eventually controlling alternative splicing, polyadenylation and transcript stability. This is particularly relevant in the case of genes bearing long introns such as genes from the HSC signature [55].

TEs may also play a role in shaping HSC cis-regulatory networks thereby regulating HSC fate. Compared to other tissues, HSCs and blood cells are particularly enriched in TEs exhibiting an epigenetic status characteristic of gene regulatory elements [21]. TEs enriched for binding sites motifs for TFs involved in the endothelial-to-hematopoietic transition are temporally accessible and may play the role of enhancers in the emergence of HSCs [69]. As an example, a mammalian interspersed repeat (MIR) element, a subfamily of SINEs, increases the activity of a previously characterized enhancer of the TF stem cell leukemia (SCL), also known as TAL1, involved in the specification of HSCs [74]. MIR elements are also involved in the three-dimensional organization of the genome in human HSPCs [75]. TE subfamilies such as LTR78, LTR67B, LTR41, and LTR55 are enriched in open chromatin regions of human HSCs/HSPCs and act as docking sites for key hematopoietic transcription factors, including LMO2, ERG, and RUNX1, as well as three-dimensional genome organizers, suggesting a role in regulating gene expression programs that sustain stem cell quiescence and self-renewal [62] .

Chimeric transposable elements, or composite transposons, could originate from DNA recombination, TE insertion into another TE transcription unit, RNA-RNA recombination, ligation or RNA template switching during the reverse transcription process [76–78]. These composite transposons, such as the SINE-VNTR-Alu (SVA) retrotransposons, were recently shown to harbor opposite activating and repressing histone marks, which then may function as bivalent enhancers in modulating HSC fate transition. Indeed, knockdown of SVA elements in human HSPCs skewed their differentiation towards the lymphoid lineage, demonstrating a role for SVA in myelopoiesis [78]. On another hand, TEs are enriched within T-cell enhancers, providing binding motifs for transcription factors involved in immune functions, and already poised in HSCs [38].

However, despite the beneficial roles of TEs, maintaining tight control over their epigenetic repression and/or transcriptomic levels is essential for proper HSC function. Increased expression of L1s following H3K9me3 loss in response to irradiation or chronic inflammation induces DNA damage in HSCs and decreased self-renewal capacity [55, 56, 58]. Activation of the TE/RLR pathways during chemotherapies or aging similarly leads to chronic activation of NF-κB/IFN pathways and the production of inflammatory cytokines triggering HSC cycling [53, 63]. Reducing TE-induced DNA damage or inflammation—by increasing antiviral signaling with HSC supporting cytokines such as thrombopoietin, or using reverse transcriptase inhibitors or LINE-1 shRNAs—has been associated with significant improvement in HSC clonogenic and reconstitution capacities [53, 56, 58, 63].

Transposable elements in clonal hematopoiesis and age-related myeloid hemopathies

TEs in CHIP

As cells age, they acquire somatic mutations due to environmental exposure and replication errors. While most of these mutations will have no functional consequence, some may lead to a clonal advantage and subsequent expansion. In the hematopoietic tissue, this phenomenon is referred to as clonal hematopoiesis of indeterminate potential (CHIP). CHIP is defined as the presence of leukemia-associated somatic mutations with a variant allele frequency of ≥ 2% in the peripheral blood of individuals without evidence of hematologic disease. Up to 80% of CHIP are characterized by mutations in epigenetic factors such as TET2, DNMT3A and ASXL1. These mutations induce a competitive advantage of HSCs and imbalanced myeloid / lymphoid differentiation. CHIP cases are associated with a 10-fold increased risk of developing primary but also secondary myeloid leukemia following treatment of a primary solid cancer. Myeloid differentiated blood cells stemming from TET2 and DNMT3A mutated HSCs present enhanced and novel inflammatory properties [80, 81]. This has been linked to the higher risk of developing aging-related disorders, including cardiovascular diseases with CHIP [80].

The frequency of allelic variants at a given time depends both on the age at which the mutation appears (and the time the clone has had to expand) and on the clone’s ability to gain an advantage over non-mutated clones. Thus, the prevalence of CHIP increases with age. However, longitudinal studies have demonstrated that the expansion of identical mutations varies among individuals [82–84] and that environmental factors such as inflammation influence CHIP expansion [85].

The molecular mechanisms involved in the expansion of CHIP-mutated HSCs remain poorly addressed. One study has shown that Tet2-deficient HSPCs present an increased DNA methylation at binding sites for TFs involved in terminal differentiation as compared to WT, which is exacerbated by chronic exposure of IL-1β. This may account for the increased self-renewal capacity and clonal expansion of Tet2-deficient HSPCs [57].

Although TEs are regulated by epigenetic mechanisms, very few studies have interrogated the contribution of TEs in the clonal expansion of HSCs mutated for epigenetic factors. Mutations in DNMT3A are associated with increased ERV expression and ERV-induced activation of cGAS/STING pathway, increased inflammatory responses, enhanced self-renewal and skewed myeloid differentiation [86, 87]. In contrast, Hong and collaborators showed that Tet2-deficient HSPCs are resistant towards the spatial relocalization of H3K9me3-marked heterochromatin from the nuclear periphery to the nucleoplasm which is observed in WT HSPCs upon aging. This is associated with a less pronounced upregulation of LTR and ISG expression in Tet2-deficient vs. WT HSPCs with age [53]. Recently, we showed that Tet2-deficient HSCs resist chronic inflammation by epigenetically repressing L1s and their deleterious effects. In vitro and in vivo, Tet2-deficient HSCs present decreased competitivity when in competition with WT HSCs transduced with an sh-L1 (Fig. 3). These data suggest that the relative expansion of Tet2-deficient HSCs upon chronic inflammation is dependent on the presence of L1 transcripts and their deleterious effect on WT HSCs [56].

TEs in myeloid malignancies

CHIP is considered as a preleukemic state. MDS, MPN, and mixed syndromes such as chronic myelomonocytic leukemia (CMML) are clonal diseases of the HSC primarily affecting the elderly population. These neoplasms may evolve from CHIP. In addition, around 75% of AML patients have a CHIP mutation, years before diagnosis [88]. As previously reported in solid cancers [6, 79, 89], TE deregulation is associated with the induction of an oncogenic program in these myeloid malignancies [15, 90, 91].

MIR elements are enriched in topologically associating domains (TADs), and may participate in long distance gene regulation upon changes in DNA methylation in AML [75] (Fig. 3). By analyzing ATAC-seq data in blast samples from AML patient compared to normal HSPCs, Zeng and collaborators identified a cluster of TEs, particularly MIRs and L2, subfamilies of SINE and LINE elements respectively, in open chromatin regions in AML samples [92]. These TEs are enriched in motifs for AP-1 and CEBP TFs and can bind CEBPα in Kasumi-1 cell line. By combining H3K4me1, H3K4me3, H3K27ac CHIP-seq and RNA-seq data from AML samples, they further show that while some may function as promoters, the majority of these TEs may play the role of enhancers. They finally validated their enhancer activities through luciferase assays in vitro in leukemic cell lines. In another study, Deniz et al. showed that 6 families of LTR elements presented specific chromatin accessibility in AML cell lines and AML samples from patients, but not in differentiated cells. Their accessibility correlated with nearby gene overexpression. These LTRs also present characteristics of enhancers i.e. enrichment of H3K4me1 and/or H3K27ac, and are bound by TFs related to hematopoiesis and leukemogenesis. Interestingly, by deleting some of these LTR loci, they were able to validate their role in inducing the expression of their nearby genes. Finally, repressing multiple elements from a same subfamily of LTRs through dCAS9 recruitment of KRAB protein and apposition of H3K9me3 impaired the growth of the leukemic cell line, highlighting the role of these TEs as enhancers in inducing an oncogenic program in leukemic cells [93] (Fig. 3).

Clustering of samples based on chromatin accessibility at HERVs and HERV expression enable classification of AML from normal cells, in accordance with their cell of origin, and delineates AML subtypes with different prognoses. HERV expression correlates with the expression of neighboring cancer-associated genes [12]. Importantly a HERV leukemic stem cell (LSC) signature could be established that mirrored the performance of the stemness LSC17 gene signature on overall patient survival. Finally, Grillo et al. showed that chromatin accessibility at TEs can discriminate self-renewing LSCs (LSC⁺) from non-self-renewing leukemic cells (LSC⁻) in AML. These differentially accessible TEs represent a signature of stemness in AML associated with a poor outcome. TE accessibility specifically in LSCs compared to normal HSPCs defines docking sites for TFs essential in AML [62]. Repressing LTR12C, one of the LTR found more accessible in Deniz et al.’s study [93], through CRISPR/dCas9-KRAB recruitment of H3K9me3 in OCI-AML22 model, reduced the percentage of LSC⁺ fraction [62].

All these studies suggest that TE derepression may play a role in leukemogenesis. However, decreased expression of TEs was also observed in AML and MDS compared to normal samples, in LSCs vs. blasts, and in high-risk vs. low-risk cases of MDS [94–96]. Decreased repeat-to-gene ratio, and particularly decreased L1 expression, are associated with a poor prognosis in AML [95, 96]. Repression of TE expression may thus play a role in leukemogenesis (Fig. 2). Epigenetic repression of TEs through H3K9me3 apposition by the HUSH complex and SETDB1 was indeed shown to be necessary for AML proliferation in vitro and development in vivo [35, 95]. Gu and collaborators showed that L1 overexpression upon the loss of MPP8, a member of the HUSH complex, induces DNA damage accumulation and p53-dependent cell cycle exit in AML cell lines. Reactivation of L1 expression through CRISPR-dCas9 mediated activation of L1 promoter resulted in impaired leukemia cell growth in vitro and in vivo [95]. In another study, Cuellar et al. showed that loss of SETDB1 induced LTR, LINE-1 and satellite repeat expression and TE-induced production of dsRNAs, activation of an innate immune response and IFN-induced apoptosis [35]. In patients with CMML, 60% of whom present with biallelic mutations of TET2, we recently showed an increase of the repressive H3K9me2 mark at TEs and a repression of immune associated transcripts. This could allow the resistance and further expansion of mutated cells as compared to their WT counterparts upon aging [52].

Manipulating TEs for therapeutic interventions in myeloid malignancies

Response to epigenetic drugs

Hypomethylating agents (HMAs) such as 5-azacytidine (5-AZA) and decitabine (DAC) are standard therapies for high-grade MDS, CMML, AML in older patients. Beyond their cytotoxic and epigenetic effects, HMAs can trigger antiviral immune responses through the derepression of ERVs. Pioneering studies first demonstrated that ERV reactivation by HMAs suppresses tumor-initiating cells and enhances immune checkpoint blockade in solid cancers, prompting investigation of similar mechanisms in myeloid malignancies [97, 98]. Evidence, however, remains mixed. 5-AZA upregulates ERVs in MDS CD34⁺ cells [99] and sensitizes DNMT3A-mutant AML cells via viral mimicry [86], yet patient studies yield inconsistent immune signatures. Kazachenka et al. (2019) observed TE upregulation in MDS and CMML HSPCs without IFN activation, whereas Ohtani et al. (2020) reported innate immune signaling linked to young TE expression in responders [100, 101].

Because DNA methylation cooperates with repressive histone marks to enforce TE silencing, dual inhibition of these epigenetic pathways may be required to trigger a full antiviral-like response (Fig. 4). Indeed, increased H3K9me2 levels have been shown to prevent type I IFN activation in AML cell lines, and inhibition of the G9a/GLP H3K9me2 methyltransferases synergizes with HMAs to induce leukemic cell death [102, 103]. Combined treatment with HMAs and G9a/GLP inhibitors induced the overexpression of LTR and ISGs in OCI-AML3 cell line, and leukemia reduction in OCI-AML3 xenograft models, as well as decreased viability and clonogenicity of primary samples from DNMT3A^mut NPM1^mut AML patients [104]. Finally, we recently demonstrated that ERV reactivation and the induction of ERV-mediated immune responses in CMML HSPCs can only be achieved through combined treatment with HMAs and G9a/GLP inhibitors [52]. While HMAs alone are insufficient to eradicate the malignant CMML clone [105], the combination therapy selectively reduced the fitness of mutated cells while sparing residual non-mutated patient cells [52]. This observation is consistent with previous reports showing that HMAs preferentially target granulomonocytic progenitors in advanced MDS and AML [106, 107].

Fig. 4.

Dual inhibition of DNA and histone methylation/deacetylation is necessary to allow TE expression and TE-induced eradication of AML or CMML malignant clones through viral mimicry. This image was created with Biorender.com

Finally, inhibition of additional pathways known to repress TE expression or TE-induced viral mimicry and frequently deregulated in LSCs as well as in high-risk MDS and AML—such as dsRNA editing by ADAR1 [108–110], autophagy [94] or SUMOylation [111]—has also demonstrated promising anti-leukemic effects, both as monotherapy and in combination with DAC, in vitro and in preclinical models. These findings further support the rationale for targeting TE-suppressive mechanisms to enhance the therapeutic response in myeloid malignancies.

TE-derived neoantigens

Hematological malignancies have a low mutational burden. Approaches to uncover tumor neoantigens derived from HERV-encoded peptides, or TE-exon chimeric transcripts induced by non-canonical splicing events or promoter exaptation that would allow the development of chimeric antigen receptor-T (CAR-T) cell therapies adapted to these diseases are an intense field of research (Fig. 4).

Goyal et al. demonstrated that combined HMAs and HDAC inhibitor treatment induces MHC-I presentation of neoantigens from chimeric TE-exon transcripts in AML samples [112]. Using a proteogenomic approach combining RNA-seq and mass spectrometry in AML patient samples, Ehx et al. showed that many tumor antigens originated from TE loci were shared across patients and were immunogenic in vitro [113]. HERV-derived epitope candidates specifically expressed in patient bone marrow and HLA-I-restricted CD8⁺ T-cell clones reactive to these epitopes were found in patients with MDS, CMML or AML at diagnosis [12,114]. Future research will determine whether these TE-derived neoantigens are present in patients at diagnosis or relapse, and whether they are effectively immunogenic and presented on the surface of LSCs.

Conclusion

Altogether, these studies highlight the dual role of TEs in HSCs during steady-state hematopoiesis, stress responses, and leukemogenesis. While low levels of TE activity—by enabling a controlled low level intrinsic type I IFN signaling or providing TF binding sites within promoters and enhancers—may facilitate adaptation to infectious challenges, enhance HSC resilience during transplantation, and fine-tune HSC gene expression, excessive TE activation, particularly of the evolutionarily recent LINE-1 elements with DNA-damaging potential, can have deleterious consequences. During aging, chemotherapy, irradiation, or inflammatory stress, such overactivation leads to the loss of HSC self-renewal capacity and contributes to the clonal expansion of CHIP-mutated cells, in which LINE-1 expression is repressed.

During leukemogenesis, derepression of transcription factor motifs within certain TEs, mainly LTR elements, induces oncogenic programs, while their expression can also trigger anti-tumoral responses through DNA damage and antiviral signaling, ultimately leading to p53- or IFN-dependent cell cycle arrest or apoptosis. It remains to be determined whether the pro- versus anti-leukemogenic effects of TEs depend on their intrinsic nature or on their level of expression. Monitoring TE expression dynamics in parallel with mutational and transcriptional changes during clonal evolution [115] could provide valuable insights into when TE activation shifts from adaptive to pathogenic. Defining these transition points would help identify therapeutic windows for releasing TE expression to enhance anti-leukemic immune responses. Such questions could be experimentally addressed through precise modulation of TE activity using CRISPR-dCas9–based approaches.

Authors’ contributions

EEM and FP all contributed to the writing of the manuscript, and have all read and approved the final manuscript.

Funding

This work was supported by grants from the French National Agency for Research (ANR-25-CE14-1668-01 to EEM and ANR-23-CE14-0017-01 to FP).

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.