Guardian of the Human Genome: Host Defense Mechanisms against LINE-1 Retrotransposition

Long interspersed element type 1 (LINE-1, L1) is a mobile genetic element comprising about 17% of the human genome, encoding a newly identified ORF0 with unknown function, ORF1p with RNA-binding activity and ORF2p with endonuclease and reverse transcriptase activities required for L1 retrotransposition. L1 utilizes an endonuclease (EN) to insert L1 cDNA into target DNA, which induces DNA double-strand breaks (DSBs). The ataxia-telangiectasia mutated (ATM) is activated by DSBs and subsequently the ATM-signaling pathway plays a role in regulating L1 retrotransposition. In addition, the host DNA repair machinery such as non-homologous end-joining (NHEJ) repair pathway is also involved in L1 retrotransposition. On the other hand, L1 is an insertional mutagenic agent, which contributes to genetic change, genomic instability, and tumorigenesis. Indeed, high-throughput sequencing-based approaches identified numerous tumor-specific somatic L1 insertions in variety of cancers, such as colon cancer, breast cancer, and hepatocellular carcinoma (HCC). In fact, L1 retrotransposition seems to be a potential factor to reduce the tumor suppressive property in HCC. Furthermore, recent study demonstrated that a specific viral-human chimeric transcript, HBx-L1, contributes to hepatitis B virus (HBV)-associated HCC. In contrast, host cells have evolved several defense mechanisms protecting cells against retrotransposition including epigenetic regulation through DNA methylation and host defense factors, such as APOBEC3, MOV10, and SAMHD1, which restrict L1 mobility as a guardian of the human genome. In this review, I focus on somatic L1 insertions into the human genome in cancers and host defense mechanisms against deleterious L1 insertions.

occurring during cancer development. 53% of the patients have at least one somatic L1 retrotransposition event, of which 24% were 3 ′ transductions, most frequently colorectal cancers (93%) and lung cancers (75%), suggesting that 3 ′ transduction are potentially mutagenic. Somatic L1 retrotranspositions tend to insert in intergenic or heterochromatin regions of the cancer genome (Tubio et al., 2014). Furthermore, somatic L1 insertions participate in the dynamics of many tumor genomes and lead to driver mutations. Surprisingly, L1 insertion was reported in colonic adenoma, a known cancer precursor, suggesting that widespread somatic L1 retrotransposition occurs early during development of gastrointestinal (GI) tumors, probably before dysplastic growth (Ewing et al., 2015). Similarly, a recent study demonstrated that L1 retrotransposition is active in esophageal adenocarcinoma and its precursor, Barrett's esophagus (BE), indicating that somatic L1 insertions occur early in BE and esophageal adenocarcinoma. Notably, two L1 insertions were detected in normal esophagus, indicating that some L1 insertions may occur in normal squamous epithelium cells (Doucet-O'Hare et al., 2015). In this regard, most of the new somatic insertions are truncated, and would not mobilize again. So mutations arising from insertions in the normal precursor esophageal or benign BE would be contributing to tumorigenesis. Otherwise, only a rare full-length somatic insertion has the potential to contribute to mutation during the various stages of transition to tumorigenesis. In addition, L1 insertions in pancreatic ductal adenocarcinoma (PDAC) were reported with discordant rate of retrotransposition between primary and metastatic sites, suggesting that L1 insertions in gastrointestinal neoplasms occur discontinuously. Thus, somatic L1 insertions contribute to genetic and phenotypic heterogeneity in PDAC . Interestingly, somatic insertions were identified in epithelial tumors but not in blood or brain cancers . However, we raise awareness regarding the following limitations of this study. For example, the sample size was small and the normal tissue was not from the same patient. In addition, in this study they only examined multiple myeloma and did not look at the entire spectrum of blood based cancers. In this regard, ten-eleventranslocation (TET) 2, a DNA demethylation-related protein, is frequently mutated in myeloid and lymphoid tumors (Ko et al., 2015). The TET family that oxidizes 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) in DNA, leads to the DNA demethylation. Since DNA methylation has a pivotal regulatory role in L1 silencing, TET2 may impact L1 mobility. Therefore, L1 insertions may be suppressed in such hematological cancers. Intriguingly, several somatic insertions occur in genes that are commonly mutated in cancers such as tumor suppressor gene. These insertions disrupt the expression of target genes, and are biased toward regions of cancer-specific DNA hypomethylation . Indeed, recent studies identified somatic L1 insertion in tumor suppressor genes, such as APC and PTEN (Miki et al., 1992;Helman et al., 2014). As well, the first case of familial retinoblastoma (Rb) caused by a de novo insertion of a full-length L1 into intron 14 of the Rb gene, resulting in the aberrant and non-canonical mRNA splicing of the Rb gene, was reported (Rodríguez-Martín et al., 2016). Furthermore, 18 retrotransposon insertions [14 Alu, 3 L1, and 1 poly(A)] were identified in neurofibromatosis type 1 (NF1) gene (Wimmer et al., 2011).
Although still debated, cell division seems to be required for efficient L1 retrotransposition (Shi et al., 2007;Xie et al., 2013). In fact, retrotransposition was strongly inhibited in the cells arrested in the G 1 , S, G 2 , or M phase of cell cycle. The reduction in L1 transcript abundance limits retrotransposition in nondividing cells, suggesting that inhibition of retrotransposition in non-dividing cells protects somatic cells from accumulation of deleterious mutations caused by L1 insertions (Shi et al., 2007). In contrast, there is an opposite report that L1 retrotransposition was detected in non-dividing and primary human somatic cells using adenovirus-L1 hybrid vector, even though they detected L1 retrotransposition in G 1 /S-but not in G 0-arrested cells (Kubo et al., 2006). In addition, retrotransposition was also inhibited during cellular senescence in primary human fibroblasts. So far, several biomarkers of cellular senescence have been identified such as senescence-associated β-galactosidase (SA-β-Gal), p53/p21, p16 INK4a , senescence-associated heterochromatin foci (SAHF), senescence-associated secretory phenotype (SASP), autophagy, telomere-induced foci/DNA damage response (DDR), and cell cycle arrest (Kuilman et al., 2010) and the reduction in L1 retrotransposition may be a biomarker of cellular senescence. Thus, cell cycle may affect L1 retrotransposition.
L1 protein expression is a common feature of many types of high-grade malignant tumor, yet is rarely detected in early stage of tumorigenesis (Rodić et al., 2014). L1 promoter is normally silenced by methylation in normal somatic cells (Woodcock et al., 1997;Schulz et al., 2006). In contrast, L1 promoter is hypomethylated (Baba et al., 2014), and expression of L1 is elevated in many tumors. In fact, L1 expression was detected in human breast carcinomas and testicular cancers (Bratthauer and Fanning, 1992;Bratthauer et al., 1994;Nangia-Makker et al., 1998). L1 ORF1p protein is detected in a variety of tumor cells including breast cancer, colon cancer, pancreatic ductal adenocarcinoma, and HCC but not in normal somatic cells (Bratthauer et al., 1994;Asch et al., 1996;Rodić et al., 2014). Thus, L1 ORF1p expression seems to be a hallmark of many human cancers as a highly specific tumor marker.
In addition to expression of L1, a hallmark of tumor cells is an activated telomere maintenance mechanism that allows prolonged survival of the malignant tumor cells. In more than 80% of tumors, telomeres are typically maintained by telomerase. Notably, the reduced length of telomeres was reported in the L1 knockdown cells, indicating that L1 is involved in telomere maintenance in telomerase positive tumor cells (Aschacher et al., 2016). Accordingly, L1 involves in a transcriptional regulation of hTERT and upregulation of its transcription factors c-Myc and KLF-4 (Aschacher et al., 2016). Thus, L1 may contribute to the development of cancers. However, these studies were not done in alternative lengthening of telomeres (ALT)-positive tumors or telomerase negative tumors. Consequently, it is uncertain if L1 is directly contributing to telomere maintenance or if the reduction in telomere length is contributed to the reduction in telomerase levels. Indeed, the stoichiometry of telomerase is important for maintaining telomere length (Armanios et al., 2005;Goldman et al., 2005).
Chronic infection with hepatitis B virus (HBV) is a major risk for the development of HCC. HBV integration into the human genome was found in most HBV-related HCC and it has been implicated in the development of HCC. An initial study proposed that HBV integration occurs randomly without preferred integration site (Matsubara and Tokino, 1990). However, high-throughput sequencing-based approaches identified recurrent integration sites in HCC (Ding et al., 2012). HBV integration favored chromosome 17 and preferentially integrated into human transcript units. At least, telomerase reverse transcriptase (TERT) and fibronectin 1 (FN1) genes were identified as the recurrent HBV integration sites. Furthermore, seven integrations were found in the repeat regions including L1, LTR/ERV1, and SINE/Alu (Ding et al., 2012). Similarly, a recent transcriptome sequencing study of HBV-positive HCC cell lines discovered that HBV integrates into L1 (Lau et al., 2014). Insertion of the gene encoding hepatitis B virus x (HBx) into L1 on chromosome 8p11 produces an oncogenic HBx-LINE1 chimeric RNA transcript (Lau et al., 2014; Figure 2). The HBx-LINE1 RNA transcript was detected in 23.3% of HCC, suggesting that HBx-LINE1 is selected for in HCC oncogenesis. The long non-coding RNA (lncRNA)-like HBx-LINE1 transcript confers cancer-promoting properties through activation of Wnt/β-catenin signaling pathway (Lau et al., 2014).
Furthermore, endogenous L1-mediated retrotransposition was identified in the germline and somatic cells of HCC patients (Shukla et al., 2013). The germline L1 insertion in the tumor Frontiers in Chemistry | www.frontiersin.org suppressor mutated in colorectal cancers (MCC) was detected in 21.1% of HCC, resulting in the aberrant expression of MCC. Moreover, suppression of tumorigenicity 18 (ST18) was activated by a tumor-specific somatic L1 insertion (Shukla et al., 2013). Thus, L1-mediated retrotransposition seems to be a potential etiological factor in HCC.
Small RNAs have been implicated in the regulation of L1 mobility. Piwi proteins and Piwi-interacting RNAs (piRNA) silence L1 during genome reprogramming in the embryonic male germ line (De Fazio et al., 2011;Marchetto et al., 2013). Notably, Hamdorf et al. uncovered a new mechanism in which microRNAs restrict L1 mobilization and L1-associated mutations in cancer cells, cancer-initiating cells and iPS cells (Hamdorf et al., 2015). Indeed, miR-128 represses L1 retrotransposition by binding directly to L1 RNA, suggesting a new function of microRNAs in mediating genomic stability by suppressing the mobility of endogenous retrotransposons.
Tumor suppressor p53 mutations occur in most of human cancers, however, precisely how p53 functions to mediate tumor suppression is not well understood. In this regard, p53 was reported to restrict L1 mobility and suggested that p53 restricts oncogenesis in part by restricting transposon mobility (Wylie et al., 2016). Although normal human p53 suppressed transposons, mutant p53 from cancer patients could not. In contrast, L1 activity was elevated in p53 negative human cancers. Thus, ancestral function of p53 may be associated with transposon control as a guardian of human genome.

CONCLUSION
L1 has successfully propagated and composed 17% of the human genome, resulting in evolutionary force. Activation of the normally silent L1 is associated with a high level of cancer-associated DNA damage and genomic instability. Indeed, L1 insertions into the human genome may cause cancers through insertional mutagenesis. In fact, recent high-throughput sequencing-based approaches could identify numerous somatic tumor-specific L1 insertions in a variety of cancers (Iskow et al., 2010;Lee et al., 2012;Solyom et al., 2012;Shukla et al., 2013;Helman et al., 2014), however there is no sufficient evidence. Therefore, it should clarify the role of L1-mediated retrotransposition in human cancers. Indeed, the implication of L1 insertion events as either passenger or driver mutations with a causative role in tumorigenesis still remains to be clarified (Rodić and Burns, 2013). Intriguingly, somatic insertions were only identified in epithelial tumors . Accordingly, epithelial cells can be transformed to cancer stem cells (Wang et al., 2013) and metastasis is more prevalent in epithelial tumors (Gotzmann et al., 2004). Thus, epithelial cells seem to be plastic (Carreira et al., 2014). Cancer stem cells are defined as rare cells with indefinite potential for self-renewal that drive tumorigenesis (Reya et al., 2001). However, it remains to be clarified the role of L1 mobility in cancer stem cells. Recent studies focus on the relationship among L1 mobility, reprogramming, and differentiation. Indeed, reprogramming somatic cells into iPS cells activates L1 mobility (Wissing et al., 2012;Friedli et al., 2014;Klawitter et al., 2016). On the other hand, L1 mobility is enhanced in tumor cells. In this regard, the elevation of L1 protein or RNA expression levels may be useful as a diagnostic hallmark of many human cancers and as a tumor specific marker, metastasis, and prognosis. Furthermore, recent advances in single cell analysis will be useful for comparison of the L1 mobility and the integration site of L1 at a single cell level in human cancers.
Finally, tumor suppressor proteins may be associated with transposon control to restrict deleterious retrotransposition as a guardian of the human genome. Wild-type p53 suppresses transposon mobility in normal cells, while mutant p53 in cancer cells could not, resulting in the activation of L1 mobility in cancer cells (Wylie et al., 2016). Furthermore, recent studies identified somatic L1 insertion in tumor suppressor genes, such as APC, PTEN, NF1, and Rb (Miki et al., 1992;Wimmer et al., 2011;Helman et al., 2014;Rodríguez-Martín et al., 2016). Thus, L1 insertions in the tumor suppressor genes may disrupt their functions and be associated with tumorigenesis. Altogether, host cells have evolved several defense mechanisms protecting cells against retrotransposition.

AUTHOR CONTRIBUTIONS
The author confirms being the sole contributor of this work and approved it for publication. ACKNOWLEDGMENTS I thank Ms. Kazumi Tsuruhara for secretarial assistance and her kind encouragement.