Translational significance of the LINE-1 jumping gene in skeletal muscle

Abstract

Retrotransposons are gene segments that proliferate in the genome, and the Long INterspersed Element 1 (LINE-1, or L1) retrotransposon is active in humans. Although older mammals show enhanced skeletal muscle L1 expression, exercise generally reverses this trend. We hypothesize skeletal muscle L1 expression affects muscle physiology, and additional innovative investigations are needed to confirm this hypothesis.

SUMMARY

L1 is a mobile genetic element that is involved with muscle aging. However, exercise can downregulate this mechanism.

INTRODUCTION

Transposable elements (TEs), or “jumping genes”, were first discovered by Dr. Barbara McClintock at the Cold Spring Harbor Laboratory in the 1950s where she observed that two interacting genetic loci, activator (Ac) and dissociation (Ds), could “transpose”, or change positions in maize chromosomes (1). Thirty-three years following her discovery, McClintock received the Nobel Prize in Physiology, and her research inspired decades of genetic research seeking to characterize the biological mechanisms and consequences associated with TE activity.

In 1980, Adams et al. (2) discovered a six kilobase (kb) DNA sequence that flanked the human β-globin gene. Various wet laboratory techniques were used to estimate that there were approximately 5,000 copies of this sequence in the human genome and, importantly, this was the first study to identify the existence of the Long INterspersed Element 1 (LINE-1, or L1) element in humans. It was not until the introduction of advanced sequencing techniques, along with the implementation of more sophisticated in vitro experiments, that our knowledge of L1 was greatly enhanced. Unlike the TEs discovered by McClintock, which are now classified as class II TEs, L1 is classified as a class I TE. Class II TEs (or DNA transposons) can move from one genomic location to another via a “cut-and-paste” mechanism, and these sequences remain stable in copy number. On the other hand, class I transposable elements (or retrotransposons) are capable of amplification via a “copy-and-paste” mechanism termed retrotransposition, and they can increase in number. A distinguishing factor between the two types of transposable elements is that retrotransposons operate through an RNA intermediate to create a cDNA copy, and this critical characteristic will be discussed later in the review.

The current-day TE classification system maintains the two class system but implements more stringent sequence and enzymatic criteria to differentiate class, subclass, order, superfamilies, families and subfamilies of transposable elements. The purpose of this review is to highlight research examining L1, which belongs to the order of long interspersed nuclear elements. We will discuss: i) how aging affects skeletal muscle L1 mRNA expression, ii) the speculated consequences of increased skeletal muscle L1 expression, and iii) select evidence suggesting exercise can decrease L1 expression in skeletal muscle. Finally, our working hypothesis is that skeletal muscle L1 expression affects muscle physiology, and we posit novel investigations that can be pursued to test this hypothesis.

Mechanisms of L1 expression and retrotransposition

In stark contrast to the report by Adams et al. (2) suggesting there were approximately 5,000 copies of the L1 gene in the human genome, more recent sequencing experiments suggest that there are over 800,000 copies of L1, which in total, constitutes roughly 17% of the genetic code (3). It has also been estimated that approximately 80% of human genes contain one or more L1 fragments, thus illustrating the pervasive nature of this gene (4).

In humans, several L1 subfamilies exist based on the presence of sequence variants contained within the 5′ and 3′ untranslated regions (UTRs) including pre-Ta, Ta-0, Ta-1, Ta1-d, and Ta1-nd, and elements within each of these subfamilies are able to undergo retrotransposition (5). Moreover, L1 sequences evolve very rapidly, and primate-specific L1 elements exist in the human genome that have lost the ability to undergo retrotransposition due to the presence of mutations. Greater than 99.9% of L1 gene copies are “genomic fossils” of ancient retrotransposition events and cannot mobilize. However, upwards of 100 L1 copies in the human genome are “hot”, meaning that they are transcriptionally active and lead to the expression of full-length L1 mRNA transcripts (6, 7).

Full-length, retrotransposition-competent L1 elements are approximately 6–6.5 kb in length and consist of a 5’ UTR with a strong internal sense and weaker anti-sense promoter, two coding regions (ORF1 and ORF2), and a 3’ UTR that ends with an AATAAA polyadenylation signal and poly(A) tail (8) (see Figure 1). The protein encoded from the ORF2 segment of L1 is particularly critical retrotransposition, and this is discussed in greater detail below. Also notable are species comparisons of genomic L1 content. As mentioned prior, 17% of human DNA is made up of various L1 elements and estimates in rodents suggest approximately 18% of mice DNA and 23% of rat DNA are made up of L1 elements (9). Additional estimates suggest approximately 3,000 L1 copies in mice and 500 L1 copies in rats that are transcriptionally active, and these numbers dwarf the ~100 active L1 copies in humans (9).

Figure 1.

Structure of the L1 gene

The full-length template strand of a retrotransposition-competent L1 gene is illustrated. The 5’ untranslated region (5’ UTR) contains promoters in the sense (SP) and anti-sense (ASP) directions, and CpG methylation sites are enriched in this region (indicated by lollipop-like structures. The L1 gene encodes for a bicistronic mRNA. ORF1 encodes for the ORF1p protein that has cis preference for binding to L1 mRNA, ORF2 encodes for ORF2p protein that has endonuclease (EN) and reverse transcriptase (RT) activities, and ORF2p also has cis preference to bind to L1 mRNA. The 3’ UTR contains a polyadenylation signal (AATAAA) as well as a poly(A) tail.

L1 expression begins with RNA polymerase II (pol II) binding to the 5’ UTR internal promoter and transcribing L1 mRNA in the sense direction. The recruitment of RNA pol II to the 5’ UTR promoter involves the binding of various transcription factors such as CEBPB, CTCF, E2F1, GABPA, JUND, YY1, RUNX3 and dozens of others (10). However, cell type likely influences which transcription factors regulate L1 expression. For instance, L1 mRNA is highly expressed in the MCF-7 (human breast cancer) and GM12878 (human lymphoblastoid) cell lines (10), and L1 expression is driven by the CTCF transcription factor in MCF-7 cells, whereas other transcription factors coordinate L1 mRNA expression in GM12878 cells. It is important to note that, in most non-cancerous tissues, the expression of L1 is typically low. However, along with L1 expression being prevalent in certain cancers, L1 expression increases across various cell types with aging; a phenomenon that will be discussed in later in the review.

Following L1 transcription, the bicistronic mRNA is exported out of the nucleus where ribosomes catalyze the formation of the ORF1p and ORF2p proteins. ORF1p is a 40 kilodalton (kDa) RNA binding protein that exhibits preferential binding to L1 mRNA in trimers and acts as a nucleic acid chaperone (11). ORF2p is a 150 kDa protein that possesses endonuclease and reverse transcriptase domains (12), and it too exhibits preferential binding to the L1 transcript towards the 3’ UTR. During the process of retrotransposition, resultant L1 mRNA/ORF1p/ORF2p ribonucleoprotein particles (L1 RNPs) translocate back into the nucleus where ORF2p is capable of nicking genomic DNA at consensus sites containing 5′-AA/TTTT-3′ sequences (13). ORF2p utilizes a free 3’ hydroxyl group to prime reverse transcription from the 3’ end of the L1 transcript (14); this process is termed target-primed reverse transcription (TPRT). TPRT is the critical step that leads to the formation of a new L1 insertion in the genome. However, reverse transcription frequently fails to proceed to the 5′ end, and this results in truncated de novo L1 insertions that are functionally inert. The process of L1 transcription, L1 RNP formation, and TPRT are illustrated in Figure 2. Notably, our fundamental understanding of this mechanism can be attributed to elegant in vitro experiments performed by Boeke, Moran, Kazazian and others (15, 16).

Figure 2.

The process of L1 transcription, L1 RNP formation, and TPRT

From upper left-hand portion of the figure, RNA polymerase II (Pol II) L1 mRNA transcription ensues with binding of cell-specific transcription factors. The bicistronic mRNA is exported from the nucleus and translated into the ORF1p and ORF2p proteins by ribosomes (Ribo). These proteins then associate with L1 mRNA forming the L1 ribonucleoprotein (L1 RNP); note, while not discussed, ORF1p and ORF2p can also associate with other select transposable elements such as Alu and SVA. The L1 RNP translocates back into the nucleus and ORF2p nicks DNA where 5′-AA/TTTT-3′ sites exist. ORF2p then uses its reverse transcriptase activity to promote first and second strand synthesis of a L1 cDNA copy. Ultimately, these events lead to the insertion of a new L1 copy.

Elevated L1 mRNA expression and de novo genomic L1 insertions occur in germline cells as well as during the early stages of embryogenesis and later stages of fetal development (17–19). The reprogramming of human dermal fibroblasts into pluripotent stem cells (iPSCs) also leads to enhanced L1 mRNA expression and an elevation in retrotransposition frequency (20). The relevance of increased L1 expression in germline cells or during developmental periods has been widely speculated. Some have hypothesized that L1 retrotransposition facilitates “genetic programming” events given the presence of endogenous regulatory sequences (i.e. promoters) on the gene that can drive the expression of other genes adjacent to de novo insertions. Others have maintained active L1 elements are “genomic parasites”, and retrotransposition leads to deleterious random insertion events that act to destabilize the genome (21). Interestingly, Percharde et al. (22) suggest L1 acts as a long non-coding RNA to help cells express distinct cell-specific genetic programs during embryogenesis. These authors also reported that L1 transcript knockdown results in the inability of cells to move into the next stage of embryogenesis. Hence, while TEs have been heavily researched in disease-related contexts (22–24), the co-option of L1 and other TEs seems to be a frequent theme in evolution.

Mechanisms of L1 inhibition with an emphasis on skeletal muscle

Given the implied involvement of L1 expression with several diseases, there has been a high level of enthusiasm in studying molecular pathways that down-regulate L1 mRNA expression or L1 RNP activity. In this regard, numerous studies over the past decade have reported several endogenous L1 inhibitors exist, and these proteins likely participate in host surveillance mechanisms. For instance, the P-element-induced wimpy testis (PIWI)-interacting RNA (piRNA) pathway has been shown to promote L1 mRNA degradation in Drosophila and germ cells from mice (25, 26). Wylie et al. (27) performed delicate experiments in Drosophila and Zebrafish to determine that the tumor suppressor protein, p53, transcriptionally represses L1. Sirtuin 6 (SIRT6) has been shown to bind to the 5’ UTR of L1 where it facilitates the formation of transcriptionally repressive heterochromatin in mouse and human fibroblasts (28), and this transcription factor will be discussed later in the context of skeletal muscle. The SAM and HD domain containing deoxynucleoside triphosphate triphosphohydrolase 1 (SAMHD1) protein has been shown to inhibit ORF2p-mediated L1 reverse transcription by degrading deoxynucleotide triphosphate pools (29), while also reducing ORF2p levels in L1 RNPs in human embryonic kidney (HEK) 293T cells (30). The three-prime repair exonuclease 1 (TREX1) cytoplasmic DNase reduces ORF1p levels and L1 RNP-mediated nicking of DNA in HEK293T and HeLa cells (31). Additionally, several reports suggest TREX1 degrades L1 cDNA (reviewed in (32)). Other studies using HeLa, HEK293T, and other cell lines have shown proteins of the apolipoprotein B mRNA editing complex polypeptide 1-like (APOBEC)-3 family down-regulate L1 activity by physically associating with and reducing the activity of ORF1p (33, 34). The Moloney leukemia virus 10 (MOV10) RNA helicase protein interacts with the L1 RNP to promote L1 mRNA degradation in HeLa and HEK293T cells (35). The p21^Waf1 and p27^Kip1 cell cycle regulators have been shown to reduce L1 activity by reducing ORF2p activity in HEK293T cells (36). MicroRNA-induced silencing of L1 also occurs; specifically, miR-128 reduces L1 mRNA levels and retrotransposition in HeLa cells (37).

L1 inhibitory mechanisms are likely cell- and context-specific, and many of these aforementioned processes may not carry over into skeletal muscle. For example, piRNA-mediated L1 inhibition occurs exclusively in germline cells (38), and our in-house RNA-sequencing data suggest Apobec3b, Mov10, Samhd1, and Trex1 are negligibly expressed in skeletal muscle from rats (i.e., FPKM values <1.0) (unpublished observations from (9)). However, there is ample evidence to suggest that miR-128 (39), p53 (40), p21^Waf1 (41), p27^Kip1 (41), and SIRT6 (42) are expressed in assayed skeletal muscle tissue. Thus, the potential for these genes to inhibit L1 expression in skeletal muscle exists and should be further explored, albeit muscle tissue contains a plethora of cell types (e.g., vascular endothelial cells, resident immune cells, resident satellite cells, fibroblasts, and pericytes) and this convolutes interpretations to a certain degree.

In addition to the aforementioned mechanisms, one highly conserved mechanism that inhibits L1 transcription across different cell types is the hypermethylation of cytosine residues in the 5’ UTR internal promoter region (21). DNA methylation is catalyzed by a family of DNA methyltransferases (DNMT1, DNMT3A, and DNMT3B) where a methyl group from S-adenyl methionine is transferred to the fifth carbon of a cytosine residue to form 5-methylcytosine. DNA methylation typically occurs on cytosine residues that precede a guanine nucleotide (i.e., CpG sites), and the hypermethylation of CpG sites in promoter regions recruits methyl-CpG-binding domain proteins as well as histone deacetylases and other chromatin remodeling proteins that can modify histones. These events catalyze the formation of tightly bound heterochromatin that is not accessible to transcriptional machinery. Zhi-meng et al. (19) demonstrated how L1 promoter methylation affects L1 mRNA expression. Specifically, these authors examined human placental specimens obtained from the first and third trimesters and noted that L1 promoter methylation was high and L1 mRNA expression was relatively low during the first trimester, whereas specimens in the third trimester exhibited ~50% reduction in L1 methylation and nearly a doubling of L1 mRNA expression.

De novo DNA methylation events are catalyzed by DNMT3A and DNMT3B, and methylation patterns are maintained by the DNMT1 enzyme (43). Wong et al. (44) reported that DNMT1 and DMNT3A protein levels are highly expressed in mouse skeletal muscle. Our in-house RNA-sequencing data mentioned above suggest Dnmt3b is negligibly expressed in skeletal muscle from rats that are 3 and 24 months old (FPKM values ~0.1), although Dnmt3a is expressed at higher levels (FKPM ± standard deviation values: 3-mo rats = 3.03±0.92, 24-mo rats = 2.32±0.70; n=8 rats per group, p=0.041) (unpublished observations from (9)). These data collectively suggest de novo cytosine methylation events in skeletal muscle likely occur through the DNMT3A enzyme, and the maintenance of methylation occurs via DNMT1. In support of this hypothesis, Rajabi et al. (45) reported that disrupting DNMT1 expression in cancer cells reduced L1 methylation levels, and more recent evidence from Wang et al. (46) suggests DNMT3A methylates the L1 promoter. However, we recently determined that 5-Azacytidine, a DNMT inhibitor, does not affect L1 mRNA expression in rat-derived L6 myotubes (47). Thus, limited data in this area suggest a heightened complexity of L1 regulation in this tissue that likely extends beyond promoter methylation.

Skeletal muscle L1 expression during aging

Skeletal muscle aging in rodents and humans is largely conserved and there are a number of associated phenotypes including: i) a reduction in fiber size and number of type II fibers, ii) a loss in resident satellite cell proliferative capacity and number, iii) a dysfunction in translational machinery, iv) a decrease in contractile protein content, v) an increase in fibrotic tissue, and a vi) decrease in mitochondrial function (48). While these phenotypes are likely caused by multiple mechanisms, three muscle aging signatures exist and include transcriptome-wide changes in mRNA expression, genome-wide alterations in DNA methylation patterns that likely contribute to aforementioned transcriptome changes (49), and an alteration in mRNA splicing which leads to the formation of more transcript variants (50). Collectively, all three of these phenotypes possess an underlying theme of genomic instability.

It is reasonable to speculate that age-associated changes in nuclear genome stability likely contribute to the aging phenotypes described above from a top-down perspective; that is, a dysregulation in DNA chromatin state and resultant mRNA expression patterns lead to downstream disruptions in cellular homeostasis. Further, given the pervasiveness of L1 copies in the genome, age-associated changes in chromatin state likely increase skeletal muscle L1 expression. De Cecco et al. (51) provided compelling evidence to support this thesis by examining L1 markers in skeletal muscle specimens from mice that were 5 months old, 24 months old, and 36 months old. Skeletal muscle L1 mRNA expression was greater in 24- and 36-month-old mice versus 5-month-old mice (p<0.05), and the authors speculated that an age-related relaxation of heterochromatin in gene-poor/L1-rich genomic regions may lead to aberrant increases in tissue L1 mRNA expression. Our laboratory sought to replicate the findings of De Cecco and colleagues by examining L1 expression markers in the gastrocnemius muscle of rats that were 3 months old, 12 months old, and 24 months old (9). Two primer sets were used to interrogate L1 mRNA using real time RT-PCR-based methods. The first primer set (L1.3) was designed to probe for the most active L1 element based on the findings of Kirilyuk et al. (52). The second primer set (L1.Tot) was designed to encompass full-length L1 elements that contained a 5’ promoter, but did not have the ability to undergo retrotransposition based on mutations in the protein coding regions (52). Notably, both primer sets were designed to target the 5’UTR region. We discovered that skeletal muscle L1.3 mRNA expression increased in an age-dependent fashion (24-month-old rats > 12- and 3-month-old rats, p<0.05). While L1.3 DNA methylation numerically decreased with aging, it did not reach a level of statistical significance between age groups. However, we did observe that the amount of L1.3 DNA in accessible chromatin regions was higher in 24-month versus 3-month rats. Our study supported the findings of De Cecco and colleagues suggesting skeletal muscle L1 expression increases with aging. Further, our L1 chromatin accessibility data mentioned above expanded upon their findings by demonstrating this phenomenon is likely due to an age-associated relaxation of heterochromatin in gene-poor/L1-rich genomic regions.

A follow-up study in our laboratory sought to replicate findings from both abovementioned rodent studies in humans (53). In this study, college-aged versus older (~60 years old) participants reported to the laboratory in an overnight-fasted condition and donated a skeletal muscle biopsy. The relative abundance of L1 DNA, as determined by real-time RT-PCR, did not differ between the younger and older participants. This supports the notion that skeletal muscle L1 retrotransposition does not appreciably increase with aging in humans. However, this may be due to skeletal muscle being post-mitotic, this is discussed in the next section of this review. We additionally reported that L1 mRNA was significantly greater and L1 promoter methylation was significantly lower in older versus younger participants. We speculate that the observed differences in promoter methylation between age cohorts increased L1 chromatin accessibility in the older participants, which in turn, increased L1 mRNA expression. Also important to note is that PCR-based methods were used for all of the assays, and primer sets were generally designed to target the 5’UTR of the L1 gene. These data collectively suggest skeletal muscle L1 expression increases with aging in mice, rats, and humans. While provocative, these studies did not elucidate the functional consequences of increased skeletal muscle L1 expression with aging, which warrants consideration to whether this phenomenon contributes to or merely coincides with muscle aging. Additionally, all our data was performed on muscle tissue, and our assays did not delineate if muscle fibers, versus other resident cell types, were the prominent source of the L1 signals. Finally, L1-based interrogations via PCR pose inherent methodological limitations that will be discussed later.

L1 expression post-mitotic myofibers likely drives inflammation rather than retrotransposition

A recent review by Gorbunova et al. (54) suggests L1 can operate in three distinct manners. First, L1 RNPs can affect the genetic and epigenetic landscapes via retrotransposition and insertional mutagenesis. Second, L1 RNP-mediated strand breakage with active or abortive retrotransposition can lead to DNA damage. Finally, newer evidence suggests cytoplasmic L1 cDNA can activate immune pathways in a variety of cell types leading to increased inflammatory signaling, and this will be discussed below. Several lines of indirect and direct evidence suggest enhanced L1 expression in post-mitotic cells (i.e., skeletal muscle) may not lead to retrotransposition and, instead, may trigger innate immune signaling. Boeke’s laboratory (55) demonstrated L1 retrotransposition primarily occurs at the DNA replication fork during the S phase of the cell cycle, and that the entry of L1 RNPs into nuclei requires nuclear membrane disaggregation. Given that these processes do not occur in the myonuclei of post-mitotic muscle fibers, age-related increases in skeletal muscle L1 expression likely do not result in retrotransposition. Moreover, the progeroid Sirt6-knockout model has been used to show heightened L1 mRNA expression leads to the accumulation of cytoplasmic L1 cDNA in various tissues (56). This suggests that L1 mRNA, cDNA, and RNPs are stalled in the cytoplasm of cells (rather than transported into the nucleus), and the authors showed that L1 cDNA subsequently activated the cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) pathway, which in turn, drove the expression of pro-inflammatory type I interferons. Similar results have been reported in human cell lines as well (57); specifically, the overexpression of an engineered L1 construct has been shown to activate interferon signaling. Given that low-grade inflammation has been associated with muscle aging (i.e., inflamm-aging), the notion that cytoplasmic L1 cDNA accumulation may initiate or be primarily responsible for age-associated muscle inflammation is a compelling hypothesis.

Skeletal muscle L1 activity and exercise

Only three studies to date, all from our laboratory, have sought to determine how exercise affects skeletal muscle L1 mRNA expression. The first report by Romero et al. (40) examined skeletal muscle obtained from college-aged males that participated in one of two exercise protocols. In the first protocol, 10 participants performed three consecutive days of lower-body resistance training. Prior to the exercise protocol, a skeletal muscle biopsy was obtained from the outer right thigh (termed “PRE”). A second muscle biopsy was obtained from the same leg position two hours following the first day of training (termed “Post1”), and a third muscle biopsy was obtained 72 hours following the third day of training (termed “Post2”). L1 mRNA, assessed using real-time RT-PCR and primers designed to amplify the ORF2P region, significantly decreased at Post1 and Post2 relative to PRE (p<0.05). Additionally, L1 5’ UTR methylation, assessed using MeDIP and downstream PCR-based techniques, also increased >4-fold at Post1 and Post2 relative to PRE, albeit neither time point reached statistical significance. The second exercise protocol in the Romero et al. study involved 12 weeks of full-body resistance training where 13 different participants donated a skeletal muscle biopsy from the outer right thigh prior to the protocol (termed “PRE”) and 72 hours following the last training bout during week 12 (termed “POST”). Similar themes emerged where L1 mRNA was downregulated PRE to POST (p=0.056), and L1 5’ UTR methylation increased >2-fold from PRE to POST (p<0.05). Our data collectively suggest resistance training down-regulates skeletal muscle L1 mRNA expression, and this is further supported by the observed increases in L1 promoter methylation. Critically, this was the first evidence to suggest that exercise downregulates skeletal muscle L1 expression in humans.

A more recent study from our laboratory by Romero et al. (47) examined skeletal muscle L1 activity markers in two groups of female rats including those that performed voluntary wheel running from the ages of 5 weeks old to 7 months old (termed “EX rats”) versus aged-matched sedentary counterparts (termed “SED rats”). While this study lacked an aging component (i.e., older rats that exercised over weeks to months), the intent of this study was to examine if adopting long-term voluntary exercise at very young age affected skeletal muscle L1 expression. The aforementioned L1.3 and L1.Tot primer sets were used for molecular work. Skeletal muscle L1.3 and L1.Tot mRNAs were lower in EX versus SED rats (p<0.05). L1.3 DNA methylation was also greater in EX versus SED rats (p<0.05), and L1.3 chromatin accessibility assessed via PCR-based methods was greater in SED versus EX rats (p<0.05). To determine how exercise mechanistically affected skeletal muscle L1 mRNA transcription, we performed a series of in vitro experiments by treating rat-derived L6 myotubes with AICAR, which upregulates 5’ AMP-activated protein kinase (AMPK) activity like exercise (58). Interestingly, AICAR treatments reduced L1 mRNA expression levels in a dose-dependent fashion, although L1 DNA methylation was also reduced in a dose-dependent fashion. De Cecco et al. (51) reported that lifelong caloric restriction reduces skeletal muscle L1 mRNA in mice, and these data indirectly support our in vitro findings given that caloric restriction increases skeletal muscle AMPK activity (59). However, we find it interesting that AMPK activation reduced L1 mRNA in lieu of a reduction in L1 DNA methylation, and potential mechanisms are discussed below.

The third exercise study from our laboratory determined that one hour of cycling exercise in younger and older human participants decreased skeletal muscle L1 mRNA ~30% in both age groups (p<0.05) (53). Interestingly, this finding indirectly relates with our in vitro findings above given that a plethora of evidence suggests endurance exercise transiently increases skeletal muscle AMPK activity. Our limited findings establish a framework suggesting different forms of exercise down-regulate L1 mRNA expression through enhanced promoter methylation and (potentially) through increased AMPK activity. However, the latter finding needs more experimental evidence, and if AMPK downregulates L1 expression in vivo, we posit that this likely occurs through an unidentified mechanism unrelated to L1 DNA methylation. In fact, the rapid downregulation in L1 mRNA levels transiently following exercise may be attributed to an L1 mRNA destabilizing/degradation mechanism that occurs in the cytoplasm (i.e., outside of the nucleus) rather than reducing L1 transcription rates. In support of this hypotheses, Lai et al. (60) have shown that contractile activity greatly enhances RNA turnover in rat skeletal muscle, and this was due to the increased expression of RNA destabilizing enzymes (HuR and AUF1). Moreover, others have shown AMPK controls cellular localization and function of AUF1 (61). Hence, the potential ability of heightened AMPK activity with exercise reducing L1 mRNA levels through AUF1-mediated degradation remains an attractive hypothesis that needs to be further explored.

Interestingly, a report by Bagley et al. (62) showed that a single bout of resistance exercise leads to a slight but significant de-methylation of L1 4 hours following the bout. This finding counters many of our aforementioned findings. However, L1 demethylation only occurred in participants that had prior resistance training experience, whereas those with minimal training experience did not present this effect. The authors noted that this observation may have been more indicative of a global decrease in DNA methylation given the pervasiveness of the L1 sequence, and this was likely transient in response to the training stimulus. Notwithstanding, these findings are intriguing given that they strengthen the link between exercise and L1 regulation.

Skeletal muscle L1 expression with aging and exercise, and the mechanistic tie-in with SIRT6

The aforementioned Sirt6-knockout mice exhibit a severe premature aging phenotype driven through enhanced L1 expression. SIRT6 is a nuclear-localized sirtuin and is best characterized for its nicotinamide adenine dinucleotide (NAD⁺)-dependent deacetylation of histone lysine residues H3K9 and H3K56 (63). Also notable, and as mentioned above, SIRT6 inhibits L1 expression by binding to the 5’UTR of L1 where it facilitates the formation of transcriptionally repressive heterochromatin (28). Simon et al. (56) recently reported that hind limb muscle masses as well as quadriceps muscle fiber diameters are significantly lower in these animals compared to wild-type mice. In line with the notion that enhanced L1 expression primarily operates through stimulating the pro-inflammatory cGAS-STING pathway, these authors demonstrated that nucleoside reverse transcriptase inhibitors (NRTIs) reduced cytoplasmic L1 cDNA accumulation and IFN-α and IFN-β1 expression the in these mice. Critically, there is indirect evidence to suggest SIRT6 may be a conserved mechanistic target responsible for both the age-associated increases in skeletal muscle L1 expression and an exercise-associated reversal in this trend. First, skeletal muscle NAD⁺ concentrations are substantially lower in older versus younger human subjects (64), albeit 10 weeks of resistance training nearly doubled muscle NAD⁺ concentrations in the former group, thus restoring them to “youth-like” levels. Also notable was the significant increase in global sirtuin activity in skeletal muscle observed in older individuals. While the activities of individual sirtuins were not assayed, we speculate that these events may have coincided with enhanced SIRT6 activity. However, whether all the above coincided with a down-regulation in skeletal muscle L1 expression remains to be determined. In lieu of these knowledge gaps, we posit future work using muscle-specific floxed Sirt6-knockout mice or muscle-specific L1-knockout mice will be useful in determining the role that increased L1 expression has on skeletal muscle physiology. Moreover, performing acute and chronic exercise training studies with these animals may reveal novel exercise-inducible factors that inhibit L1 expression and/or the downstream effects of reduced cytoplasmic L1 cDNA accumulation.

L1 expression in other cell types with exercise?

While skeletal muscle was emphasized herein, it is notable that exercise may affect L1 markers in other cell types. For instance, others have suggested that higher levels of physical activity downregulate leukocyte L1 expression (65, 66). Conversely, Muotri et al. (67) reported that mice engaged in two weeks of voluntary wheel running experience increased L1 DNA copy number in hippocampal neurons. The authors noted that the relationship between retrotransposons and their hosts is probably not entirely antagonistic. Additionally, it was speculated that an increase in L1 expression with exercise might act to coordinate chromatin remodeling during cell proliferation, which occurs in the hippocampus in response to exercise training. This hypothesis is attractive given the association between increased L1 expression and an upregulation in rRNA synthesis to support rapid cell proliferation in mouse embryonic stem cells (22). However, one notable difference between our data in skeletal muscle versus the findings of Muotri and colleagues is that skeletal muscle tissue largely consists of post-mitotic muscle fibers. Thus, it is plausible that an exercise-induced up-regulation L1 expression is needed for chromatin remodeling events, which in turn, permits successful cellular proliferation during hippocampal neurogenesis.

While these limited data in leukocytes, hippocampal neurons, and skeletal muscle have been insightful, it remains to be determined how exercise training affects L1 markers in other tissues. In this regard, future research examining how exercise training affects L1 expression across various exercise-responsive tissues (e.g., liver, kidneys, heart, and blood vessels) will yield insightful results. Moreover, given muscle tissue contains a variety of cell types as mentioned above, innovative single-cell sequencing interrogations from cells derived from muscle tissue extracts will yield fruitful data regarding how L1 expression is expressed in each of these cell types.

Future directions

Investigating the translational significance of L1 in skeletal muscle is ripe for investigation given that limited evidence from our laboratory is the crux of knowledge to date in this field. As mentioned prior, certain mouse models (e.g., muscle-specific floxed Sirt6-knockout mice, or muscle-specific tet-ORFeus mice) can be generated an examined with exercise studies to see how muscle phenotypes are altered in lieu of heightened L1 expression. Moreover, while RNA from repetitive elements have been largely ignored with RNA-sequencing studies, there are RNA-seq interrogation methods that have been used to quantify repetitive element enrichment (68). Hence, applying these methods to exercise training studies where muscle biopsies are obtained may lead to insightful findings. Other mouse models exist (e.g., H2B-GFP mice (69)), which allow for flow cytometry-based myonuclear capture and downstream sequencing analyses. Hence, exercise interventions with these animals where elegant myonuclear capture and back-end epigenetic or RNA-seq is performed would enable researchers to examine muscle-specific L1 DNA methylation and RNA expression events without the confounding influence of other cell types that exist in skeletal muscle. ATAC-seq (Assay for Transposase-Accessible Chromatin using sequencing) allows for the genome-wide interrogation of chromatin-accessible regions, and this may also be a useful endeavor to examine how aging and exercise affects transcriptionally-active L1 loci in the genome.

Finally, maternal exercise affects the epigenetic landscape of offspring, and this may be a fruitful avenue for L1 research. Marshall et al. (70) published human data suggesting maternal physical activity patterns did not affect L1 DNA methylation in the whole blood of offspring. While this tempers enthusiasm, this area is relatively understudied given that the publication by Marshall and colleagues is the only publication to our knowledge that has reported these outcomes. Hence, rodent and human studies that adopt similar approaches will provide more information as to how maternal physical activity habits affect L1 regulation.

Conclusions

The purpose of this review was to provide insight on L1 for those in the exercise science community, and how aging as well as exercise affect L1 markers. Figure 3 provides a hypothetical summary of the literature performed to date on this topic.

Figure 3.

Hypothetical model of how aging and exercise affect skeletal muscle L1 expression

Panel a illustrates how aging affects skeletal muscle L1 mRNA expression. In short, aging is associated with chromatin remodeling (i.e., loosening) that exposes more transcriptionally-capable L1 elements. Other mechanisms, such as reduced cellular NAD⁺ concentrations that lead to a reduction of SIRT6 at L1 loci, may also increase the number of transcriptionally-active L1 elements. These age-associated alterations in myonuclei lead to the cytoplasmic accumulation of L1 mRNA and L1 cDNA. Heightened L1 cDNA levels trigger the cyclic GMP–AMP (cGAS) and the cyclic GMP–AMP receptor stimulator of interferon genes (STING) pathway, and this leads to a heightened expression of pro-inflammatory genes. What remains to be determined is whether this L1 mechanism is a major participant in age-related muscle atrophy and dysfunction. Panel b illustrates how exercise may affect L1 expression. Several studies by our laboratory suggest exercise increases L1 methylation at the 5’UTR, and this likely results in reduced L1 mRNA expression. Whether or not exercise effects are primarily due to increases in cellular NAD⁺ concentrations and SIRT6 activity remains to be determined. We have additionally shown that AMP-activated protein kinase (AMPK) can downregulate L1 mRNA in vitro in a methylation-independent manner. Although this has not been fully elucidated from a mechanistic perspective, other literature suggests AMPK can modulate RNA stability-related enzymes which may target L1 for degradation.

Since the discovery of transposable elements by Dr. McClintock in the 1950s, advanced interrogations of mobile genetic elements have emerged in various scientific disciplines. L1 has gained a great deal of notoriety given that these elements constitute nearly one-fifth of the human genome, and several studies have shown the presence of L1 expression markers in a variety of tissues. Evidence in skeletal muscle is limited, albeit it appears that aging is associated with an up-regulation L1 expression, and our laboratory has established that acute as well as chronic exercise has the remarkable ability to reverse this process. While compelling, all our laboratory’s L1 DNA methylation, DNA accessibility, and mRNA data are limited to PCR-based methods. This is critical to note given that numerous sequence iterations of the L1 gene are present in the genome, and several versions of the L1 transcript likely exist. Moreover, L1 fragments can be present in various mRNAs, and PCR primers are incapable of discriminating full-length L1 transcripts versus mRNAs that harbor L1 fragments. However, several viable monoclonal antibodies exist to detect the ORF1p protein, and we have assayed this marker in tandem with L1 mRNA. Notwithstanding, and despite meticulous care taken during primer design phases of our projects, we submit that PCR (versus advanced sequencing methods) is likely a relatively “blunt” analytical tool to make profound discoveries in this area. Despite this limitation, and based on the arguments posed herein, it is certainly plausible that an exercise-induced decrease in skeletal muscle L1 expression over a lifetime could be a chief mechanism responsible for promoting muscle health. Critically, this area of research is in its infancy and is ripe for further investigation.

KEY POINTS:

• Long INterspersed Element 1 (L1) is a highly repetitive gene segment classified as a retrotransposon, and dozens of copies per nucleus are capable of retrotransposition (i.e., forming a cDNA copy via an RNA intermediary and inserting the newly formed copy back into the genome).

• Although cancer biologists have extensively examined the “mechanics” of L1 (e.g., L1 inhibitors, genomic L1 proliferation, etc.), little exercise science research on L1 exists.

• Interestingly, somatic L1 expression increases in various cell types during aging, and limited evidence suggests various forms of exercise can transiently and chronically downregulate L1 mRNA in muscle tissue.

• Several outstanding questions remain including: i) How does exercise affect L1 transcription or L1 mRNA stability? ii) What cell types in muscle tissue prominently express L1? and iii) Are age-associated increases in muscle L1 expression coincident with or causally linked to muscle aging?