The MOV10 RNA helicase is a dosage-dependent host restriction factor for LINE1 retrotransposition in mice

Transposable elements constitute nearly half of the mammalian genome and play important roles in genome evolution. While a multitude of both transcriptional and post-transcriptional mechanisms exist to silence transposable elements, control of transposition in vivo remains poorly understood. MOV10, an RNA helicase, is an inhibitor of mobilization of retrotransposons and retroviruses in cell culture assays. Here we report that MOV10 restricts LINE1 retrotransposition in mice. Although MOV10 is broadly expressed, its loss causes only incomplete penetrance of embryonic lethality, and the surviving MOV10-deficient mice are healthy and fertile. Biochemically, MOV10 forms a complex with UPF1, a key component of the nonsense-mediated mRNA decay pathway, and primarily binds to the 3′ UTR of somatically expressed transcripts in testis. Consequently, loss of MOV10 results in an altered transcriptome in testis. Analyses using a LINE1 reporter transgene reveal that loss of MOV10 leads to increased LINE1 retrotransposition in somatic and reproductive tissues from both embryos and adult mice. Moreover, the degree of LINE1 retrotransposition inhibition is dependent on the Mov10 gene dosage. Furthermore, MOV10 deficiency reduces reproductive fitness over successive generations. Our findings demonstrate that MOV10 attenuates LINE1 retrotransposition in a dosage-dependent manner in mice.


Introduction
Transposable elements (TEs) constitute ~40% of the mammalian genome. Despite sometimes being called "junk" DNA, TEs play important roles in genome evolution, development, and diseases [1][2][3][4]. TEs have an enormous capacity to amplify in the host genome. On one hand, integration of TEs into new genomic sites can change the level or pattern of neighboring gene expression or generate new genes, resulting in 55 greater genetic diversity that could be beneficial to the host [2,3]. On the other hand, genomic insertion can disrupt gene function and cause immediate harm to the host cell/organism. Indeed, many sporadic genetic diseases in animal species and humans are caused by transposon insertion [1]. While TEs exploit the host cellular machinery to propagate, the host in turn has evolved multiple mechanisms to suppress their mobilization to protect genome integrity [5]. The outcome of the ongoing arms race between the 60 parasitic TEs and the host is particularly critical in the germline, where TE-induced genetic changes can impact fertility and the genetic integrity over subsequent generations.
Retrotransposons, including endogenous retroviruses, are the only known active TEs in mouse and human. While the vast majority of retrotransposons are truncated inactive copies, a subset of elements remains intact and are capable of transposition -the insertion of new copies at new genomic locations. LINE1 (long interspersed nuclear element-1; L1), SINEs (short interspersed nuclear elements), and LTR (Long terminal repeat) retrotransposons are active in mouse, while L1s, SINEs, and SVAs are active in human. L1 is the most abundant class of TEs in mammals, accounting for about 17% of the genome in mouse and human. It is estimated that up to 3000 copies of L1 in mouse [6] and ~100 copies of L1 in human are intact and active [7,8]. 70 L1 retrotransposons mobilize in the genome through a "copy and paste" mechanism using reverse transcription. The 6-kb full-length L1 element contains two open reading frames (ORF1 and ORF2). The L1 ORF1 protein is an RNA-binding protein that forms a trimer and possesses nucleic acid chaperone activity [9]. The ORF2 protein exhibits endonuclease and reverse transcriptase activities [10,11]. ORF1 and ORF2 proteins bind the L1 mRNA transcript to form L1 ribonucleoprotein particles that are imported 75 into the nucleus, resulting in possible integrations at new genomic locations. SINEs do not encode proteins and rely on L1-encoded ORF2 protein for retrotransposition [12,13]. Reverse transcription of L1 and SINE RNAs is primed by ORF2-nicked DNA in a process called target primed reverse transcription.
Subsequently host-encoded DNA repair enzymes complete the integration reaction.
piRNAs post-transcriptionally degrade L1 transcripts through Piwi proteins and transcriptionally silences active L1 transposons through de novo methylation of L1 promoters in the germline [20]. Inactivation of piRNA biogenesis factors such as MOV10L1 and Piwi proteins leads to de-silencing of retrotransposons 90 in germ cells, meiotic arrest, and male infertility in mice.
Genetic studies have identified a large number of genes that are important for retrotransposon silencing in the germline [5]. Most of these genes function in the piRNA and DNA methylation pathways.
These factors suppress transcription of retrotransposons and/or cause post-transcriptional degradation of retrotransposon transcripts. However, upregulation of retrotransposon transcripts does not necessarily lead 95 to a proportional increase in new retrotransposition, suggesting that additional host factors block retrotransposition. Notably, proteomic studies have identified several dozens of cellular proteins that are associated with L1 ribonucleoprotein particles [28,29], one of which is MOV10 (Moloney leukemia virus 10), a homologue of MOV10L1. While MOV10L1 is germ cell-specific, MOV10 is widely expressed.
MOV10 interacts with RNASEH2, a nuclear ribonuclease, which hydrolyzes the RNA strands of L1specific RNA-DNA hybrids in a MOV10-dependent manner [39]. MOV10 cooperates with two terminal uridyltransferases, TUT4 and TUT7, to promote uridylation of human L1 mRNA to destabilize it and 105 inhibit its reverse transcription [40]. While these studies have provided important insights into the role of MOV10 in preventing L1 retrotransposition in cultured cells, it has not been studied in intact host organisms and here we report its physiological function in mice. We find that loss of MOV10 causes partial embryonic lethality. Using a L1 reporter transgene, we demonstrate that MOV10 is a potent host restriction factor for L1 retrotransposition in both somatic cells and germ cells in mice.

Loss of MOV10 causes embryonic lethality with incomplete penetrance
In adult mice, the MOV10 protein is highly expressed in several tissues including testis, ovary, kidney, 115 spleen, and liver, present at a low level in lung, but not detected in brain, heart, and muscle ( Fig 1A). In HEK293T cells, MOV10 is associated with UPF1, a key component of the nonsense-mediated mRNA decay pathway [30]. Similar to MOV10, UPF1 is also abundantly expressed in multiple adult tissues but not heart and muscle (Fig 1A).
To identify MOV10-associated proteins in tissues, we performed immunoprecipitation of mouse 120 testicular extracts with anti-MOV10 antibody. Both the putative MOV10 band and extra bands in the immunoprecipitated proteins were subjected to mass spectrometry for protein identification (S1A Fig). In addition to the identification of MOV10, one associated protein was identified as UPF1 (S1A Fig). The association of MOV10 and UPF1 in testis was confirmed by co-immunoprecipitations and Western blot analyses (S1B-C Fig), showing that, as in human HEK293T cells, MOV10 forms a complex with UPF1 in 125 mouse tissues.
Because of the high abundance of MOV10 in testis, we examined its expression in developing testes and found that the MOV10 abundance was constant in testes from postnatal day 6 (P6) to P56  [41]. Amh (anti-Mullerian hormone)-Cre is specifically expressed in Sertoli cells in testis and granulosa cells in ovary [42]. Immunofluorescence analysis revealed that, as expected, MOV10 was We next used the ubiquitously expressed Actb-Cre to generate Mov10 global knockout mice. Actb-Cre is under the control of the strong -actin promoter [43]. Strikingly, intercross of Mov10 +/mice 150 produced viable Mov10 -/-(global knockout) offspring but at a reduced frequency, suggesting that Mov10 deficiency causes partial embryonic lethality (Fig 1D). At E16.5, only one live Mov10 -/embryo was observed, confirming the partial lethality phenotype (Fig 1D). Our Mov10 mutant mice are on a mixed genetic background (129 x C57BL/6 x FVB), possibly influencing penetrance. Histology of Mov10 -/testis and ovary did not reveal obvious defects. Western blot analysis showed that the MOV10 protein 155 was reduced in Mov10 +/testis and absent in Mov10 -/testis ( Fig 1C). In addition, MOV10 was absent in the Mov10 -/testis by immunofluorescence, demonstrating that the knockout allele is null ( S3C Fig). Strikingly, the surviving Mov10 -/mice were grossly normal and fertile. These results demonstrate that MOV10 is important for embryogenesis but that the penetrance of lethality caused by loss of MOV10 is incomplete.

MOV10 binds preferentially to somatic transcripts and 3 UTRs in testis
Because MOV10 is an RNA-binding protein [30], we sought to identify its RNA targets in wild type P21 testis. We performed high-throughput sequencing after cross-linking and immunoprecipitation (HITS-CLIP or CLIP-seq) (Fig 2A) [44]. Autoradiography after CLIP revealed specific MOV10-RNA protein 165 complexes in the testis (Fig 2B). We extracted RNA from the main radioactive signal band and constructed cDNA libraries for sequencing. The fold enrichment in CLIP-seq data was calculated by normalization using the P21 testis RNA-seq data. Our analysis showed that 2305 transcripts were highly bound by MOV10 (fold enrichment > 10), and 7174 transcripts moderately bound by MOV10 (1 < fold enrichment < 10, S1 Table). Quantifying the distribution of MOV10 CLIP tags showed that they were 170 highly enriched in the 3 UTRs relative to transcript coding regions ( Fig 2C). However, we noticed an inverse relationship between MOV10-binding and transcript expression level in testis ( Fig 2D), with highly bound transcripts expressed at low levels in testis, while those that were weakly bound by MOV10 were more highly expressed.
We next examined the expression of MOV10-bound transcripts in mouse somatic tissues by re-175 analyzing the RNA-seq data from these tissues ( Fig 2D) [45][46][47][48][49][50]. We found that the transcripts highly bound by MOV10 in testis were highly expressed in somatic tissues, particularly in brain (Fig 2D and   2E), while those weakly bound by MOV10 in testis were highly expressed in testis (Fig 2D and 2F).
These results raised the intriguing possibility that MOV10 specifically modulates the testis transcriptome by binding to and degrading somatically expressed transcripts in the testis.

Altered transcriptome in Mov10-deficient testis
To determine the effect of loss of MOV10 on the testicular transcriptome, we performed RNA-seq analysis of testes from P21 Mov10 +/+ and Mov10 -/mice. Using the cutoff (Fold change>2, FDR<0.05, RPM>1 in at least half of libraries), we found 926 down-regulated protein-coding genes and 60 185 upregulated protein-coding genes in Mov10 -/testes ( Fig 3A and S2 Table). Notably, the upregulated genes in Mov10 -/testes were highly expressed in somatic tissues, such as retina, hippocampus, and liver ( Fig 3B). However, the downregulated genes in Mov10 -/testes were strongly expressed in the testis in comparison with somatic tissues (Fig 3C). We then performed correlation analysis between MOV10binding (CLIP-seq) and the differential gene expression in Mov10 -/testis. Twenty transcripts were 190 strongly bound by MOV10 and significantly downregulated in Mov10 -/testis ( Fig 3D). The enrichment of the MOV10 CLIP tags in testis was positively correlated with the expression fold change (Mov10 -/-/ Mov10 +/+ ) in Mov10 -/testis ( Fig 3D). These results suggest that MOV10 binding leads to degradation of its target transcripts in testis.
MOV10 was reported to decrease the level of L1 transcripts in somatic cell lines [36,38]. 195 Therefore, we examined whether L1 transcripts were affected in Mov10-deficient mouse testis. Our analysis showed that although the overall abundance of L1 transcripts was not changed in Mov10deficient testis in comparison with wild type controls (Fig 3E), the expression of two families of L1 elements was upregulated in Mov10 -/testis: L1Md_F2 and L1Md_T ( Fig 3F). In contrast with the L1 elements, several SINE and ERV families showed reduced expression in Mov10 -/testis ( Fig 3F). These 200 results suggest that MOV10 regulates transcript abundance of a subset of L1, SINE, and LTR transposable elements in mouse testis.

MOV10 inhibits L1 retrotransposition in mice
MOV10 inhibits retrotransposition of L1, SINE, and IAP retrotransposons in cell culture-based 205 retrotransposition assays [35,36,38,39,51], suggesting that it may act as a host restriction factor for retrotransposons in vivo. As endogenous L1 elements are highly repetitive, it is difficult to distinguish new L1s from preexisting L1s. Therefore, we used an L1 reporter transgene, referred to as L1 tg [52]. This L1 transgene has several key features: 1) its transcription is under the control of an endogenous mouse L1 promoter, which is methylated and thus repressed in vivo; 2) its two ORFs are codon-optimized to 210 maximize translation; 3) it contains an intron in the disrupted GFP cassette in its 3 UTR that serves as an indicator of retrotransposition; and 4) the transgene is single copy (inserted in the first intron of the Tnr1 gene on Chr. 1) (Fig 4A) [52]. As the intron is expected to be spliced out following transcription, new L1 insertions should lack the intron and thus can be distinguished from the donor L1 transgene. The promoter of this L1 tg transgene is methylated and thus repressed in vivo [52]. 215 To trace new L1 insertions in the genome, we introduced this L1 tg reporter transgene into our Mov10 -/mice. Intron-flanking PCR of genomic DNA identified new L1 copies in a subset of tissues from E18.5 embryos and 8-week-old Mov10 -/-L1 tg/tg mice (Fig 4B). Sequencing of the PCR bands confirmed that the new L1 insertions lacked the intron and thus originated from retrotransposition.
To compare the L1 insertion frequency in different tissues, we examined a total of 80 tissues from 220 10 Mov10 -/-L1 tg/tg embryos (5 males and 5 females; 8 tissues per embryo) at E18.5 and detected new L1 copies in all tissues, with higher insertion frequencies in spleen, ovary, liver, and kidney, but relatively lower insertion frequencies in lung, brain, and heart ( Fig 4C). We then analyzed 224 tissues from 8-weekold Mov10 -/-L1 tg/tg mice (14 males and 14 females; 8 tissues per mouse) and found higher L1 insertion frequencies in spleen and kidney but lower frequencies in lung, brain, and heart ( Fig 4C). MOV10 protein 225 abundance was high in liver, kidney, and spleen ( Fig 1A). These results showed that the new L1 insertion frequency in embryonic and adult Mov10 -/tissues correlated with the MOV10 protein abundance in wild type tissues. In addition, the L1 insertion frequencies were similar between E18.5 tissues and adult tissues, indicating that the L1 insertion frequency in Mov10 -/tissues is likely to be age-independent ( Fig   4C).
To determine if L1 insertion is germline transmissible, we analyzed new L1 insertions in tail genomic DNA of 101 offspring from the mating of Mov10 -/-L1 tg/tg males with Mov10 -/-L1 tg/tg females, and detected new L1 insertions in 11 pups (Fig 4F). We then examined the offspring from matings of Mov10 -/-L1 tg/tg males or females with wild type mice to distinguish male from female germline 245 transmissions. Out of 63 offspring from the mating of Mov10 -/-L1 tg/tg males with Mov10 +/+ females, a new L1 insertion was detected in the tail of only one pup (Fig 4F). Out of 99 offspring from the mating of Mov10 -/-L1 tg/tg females with Mov10 +/+ males, new L1 insertions were detected in the tails of 7 pups ( Fig   4F). However, we examined seven tissues (lung, kidney, brain, spleen, heart, liver, and testis or ovary) from 9 pups with L1 insertion-positive tails, and only one pup (from the Mov10 -/-L1 tg/tg intercross) had L1 250 insertions in all tissues examined. These results suggest that most L1 insertions detected in tail were not germline-transmitted but rather reflected new L1 insertions that occurred only in the tail.

Impact of MOV10 deficiency on reproductive fitness over multiple generations
To test whether MOV10 deficiency affects reproductive fitness over multiple generations, we intercrossed 255 Mov10 +/mice (G0) to obtain Mov10 -/mice (G1), which were then intercrossed to produce G2 Mov10 -/mice. Successive intercrosses were made until the sixth generation (G6) (Fig 5A). Three males from each generation were analyzed for testis weight, sperm count, testis histology, and litter size. The data from Mov10 +/mice (G0) served as the control. G1 Mov10 -/mice displayed similar parameters such as testis weight and sperm count as Mov10 +/mice (G0) (Fig 5B-G). However, G2 Mov10 -/mice exhibited 260 significant reductions in testis weight, sperm count, and the percentage of abnormal seminiferous tubules (Fig 5B-D). The percentage of Sertoli cell-only tubules and litter size of G2 mice were similar with G0 mice (Fig 5E-F). G3 Mov10 -/mice showed further reduction in the testis weight and sperm count, and further increase in the percentage of abnormal tubules (Fig 5B-D). The litter size of the mice decreased from 5.8 at G0 to 3.8 at G3, but the decrease did not reach statistical significance (p = 0.10, Fig 5F). 265 These results showed that while the reproductive fitness of Mov10 -/mice decreased from G1 to G3, it did not continue to exacerbate in Mov10 -/mice beyond G3. Although the Mov10 -/mice from G4 to G6 still showed smaller testis, reduced sperm count, and defective spermatogenesis, the defects were less severe than in G3 mice (Fig 5B-G). These results show that reproductive fitness of Mov10 -/mice worsens progressively until the third generation but recovers to a limited extent in later generations.  6). Previous studies revealed that MOV10L1 is essential for piRNA biogenesis [22,23,[25][26][27]53]. piRNAs orchestrate both post-transcriptional Piwi protein-dependent degradation of L1 transcripts and 280 transcriptional silencing of active L1 transposons through de novo methylation of L1 promoters (Fig 6).
In the absence of MOV10L1, we have shown that L1 is highly upregulated in male germ cells [22] and a study using the L1 tg reporter assay has shown that loss of MOV10L1 leads to increased L1 retrotransposition in mouse germ cells [52].
In contrast to the germline-specific expression pattern of MOV10L1, MOV10 is expressed 285 ubiquitously, and in cultured somatic cells, MOV10 strongly inhibits retrotransposition [35][36][37][38]. Here, we have demonstrated that MOV10 is a potent inhibitor of L1 retrotransposition in mouse (in vivo). MOV10 regulates L1 retrotransposition through several mechanisms. First, MOV10 directly binds to a large number of transcripts including the L1 mRNA [30,54]. Our MOV10 CLIP-seq revealed that MOV10 preferentially binds to 3 UTRs of transcripts in testis (Fig 2). MOV10 forms a complex with UPF1, an 290 integral component of the NMD pathway, in HEK293T cells [30]. Our study further confirmed the interaction between MOV10 and UPF1 in mouse testis (S1 Fig). Therefore, MOV10 and UPF1 may target L1 mRNA for degradation. Loss of MOV10 caused upregulation of two L1 families but did not change the overall level of L1 transcripts in mouse testis. In addition, MOV10 sequesters L1 ribonucleoprotein particles in the cytosolic aggregates [55]. Second, MOV10 interacts with L1-encoded ORF2 and 295 recombinant MOV10 blocks reverse transcription in vitro [51]. In addition, MOV10 interacts with TUT7 uridyltransferase. Uridylation of L1 3 end inhibits initiation of reverse transcription [40]. Third, MOV10 interacts with RNASEH2 to degrade L1 mRNA in RNA-DNA hybrids. Knockdown of MOV10 or RNASEH2 in cells leads to accumulation of L1-specific RNA-DNA hybrids [39]. Lastly, since MOV10 is associated with L1 ribonucleic particles in cultured cells [28,29], it could inhibit L1 mobilization by 300 interaction with additional proteins present in the particles. Thus MOV10 likely inhibits LINE1 retrotransposition by multiple mechanisms.
In contrast with the presence of two related RNA helicases (MOV10 and MOV10L1) in vertebrates, Drosophila has only one homologous RNA helicase -Armitage, which functions in both the RNAi and piRNA pathways [56,57]. The RNA helicase activity is essential for retrotransposon control 305 by both MOV10 and MOV10L1 [25,26,36]. In vertebrates, Mov10 is likely to be the ancestral gene, because it is broadly expressed and inhibits retroviruses such as HIV-1, thus serving as an ancient innate anti-viral response. Mov10l1 likely arose through gene duplication of Mov10 and evolved to be germ cellspecific [21,58] to protect the genome integrity within the germ line from TEs throughout the vertebrate genome. The functional specialization of MOV10L1 with the piRNA pathway, which most likely 310 emerged later during evolution to deal with endogenous transposons, permitted generation of retrotransposon-specific piRNAs to mediate their targeted destruction, thus serving as a "de facto" adaptive anti-viral response to protect against retrotransposon mobilization. Thus, vertebrate MOV10 and MOV10L1 repress retrotransposons but by divergent mechanisms -inhibition of retrotransposition (MOV10) and production of piRNAs (MOV10L1) (Fig 6).  Despite increased retrotransposition, the surviving Mov10 -/mice were fertile. It is possible that 335 most retrotransposition events occur in non-coding genomic regions or disrupt genes in a heterozygous manner. However, as these retrotransposition-mediated mutations accumulate over multiple generations or become homozygous, defects are expected to develop in later generations. Indeed, Mov10 -/mice exhibited decreased testis weight, reduced sperm count, and increased defects in testicular histology over generations, with the most severe defects in the third generation. However, the fertility parameters did not 340 continue to worsen beyond the third generation, possibly because gametes with a high load of insertions/mutations were selected against over multiple generations. In addition, MOV10 may regulate biological processes in addition to retrotransposition given that MOV10 binds to a large number of protein-coding transcripts.
MOV10 inhibits retrotransposition of not just L1 but also SINE, and IAP retrotransposons in cell 345 culture-based assays [35][36][37][38], raising the possibility that it may act as a host restriction factor for all active TEs in vivo. The mouse genome harbors ~3000 intact L1 elements. The mouse genome also contains the intracisternal A-particle (IAP) element, an active rodent LTR retrotransposon that is absent in human.

Histological and immunofluorescence analyses
For histological analysis, testes were fixed in Bouin's solution at room temperature overnight, embedded 395 with paraffin and then sectioned at 5 µm. Sections were stained with hematoxylin and eosin. As for immunofluorescence analysis, testes were fixed in 4% paraformaldehyde (in 1PBS) for 6 hours at 4°C, dehydrated in 30% sucrose (in 1PBS) overnight and sectioned at 5 µm. The primary and secondary antibodies used for immunofluorescence analyses were listed in S3 Table. 400

Immunoprecipitation, mass spectrometry, and Western blotting analyses
For immunoprecipitation (IP), 100 mg P21 testes were lysed in 1 ml RIPA buffer (10 mM Tris, pH 8.0, 140 mM NaCl, 1% Trion X-100, 0.1% sodium deoxycholate, 0.1% SDS, 1 mM EDTA) supplemented with 1 mM PMSF. Cell lysates were centrifuged by 16,000 g for 30 min at 4C, and 1.5% of the supernatant was set aside as input. The remaining lysates were pre-cleared with 30 μl protein G  For MOV10 mass spectrometry, the gel was stained with Coomassie Blue dye, and the bands that were present in the MOV10 lane but absent in the IgG lane were sent for mass spectrometry at Wistar Institute Proteomics Core. For western blot analysis, the resolved proteins were transferred onto a nitrocellulose membrane using iBlot (Invitrogen) and immunoblotted with primary and secondary antibodies (S3 Table).

MOV10 HITS-CLIP
MOV10 HITS-CLIP was performed as described previously [60]. Testes from P21 mice were collected, de-tunicated, dissociated by mild pipetting in ice-cold HBSS, and followed by UV crosslinking three times at 400 mJ/cm 2 , with 30-s intervals for cooling. Testicular cells were pelleted at 1200 g for 10 min at 4°C, washed with 1PBS, then the cell pellet was snap-frozen in liquid nitrogen and kept at −80°C if not 420 used immediately. UV light-treated cells (from two testes) were lysed in 700 μl of 1×RIPA buffer (10 mM Tris-HCl, pH 8.0, 140 mM NaCl, 1% Trion X-100, 0.1% sodium deoxycholate, 0.1% SDS, 1 mM EDTA) with 1 mM PMSF at 4°C for 1 hour. After that, lysates were treated with 10 μl DNase (Promega) and 2 μl RNase T1 at 37°C for 5 min. The lysates were centrifuged at 90,000g for 30 min at 4°C.
For each immunoprecipitation, 3 μg of rabbit anti-MOV10 polyclonal antibody (A301-571A, 425 Bethyl lab) was bound on protein A Dynabeads in the antibody-binding buffer (0.1 M Na₃PO₄, pH 8, 0.1% IGEPAL CA-630, 5% Glycerol) at 4°C for 3 hours, and then antibody-bound beads were washed three times with 1×PBS. Antibody-bound beads were incubated with lysates at 4°C for 3 hours. Ligation of the 32 P labeled RL3 RNA adapter was described before [60]. Immunoprecipitation beads were eluted for 10 min at 70°C using 30 μl 2×SDS sample buffer. The eluted samples were separated by 10% precast 430 gels (Biorad, 4561033). Cross-linked RNA-protein complexes were transferred onto the nitrocellulose membrane (Invitrogen, LC2001), and then the membrane was exposed to film overnight. Membrane regions containing the main radioactive signal and up to 15 kDa higher were cut. RNA extraction, 5 linker ligation (RL5), reverse transcription and two rounds of PCR were performed as described previously [60]. The sequences of the primers for the first PCR (DP3 and DP5) and the second PCR 435 (DSFP3 and DSFP5) were available in S4 Table. The DNA products were resolved on 3% agarose gels and extracted with QIAquick gel extraction kit and submitted for deep sequencing. Four libraries with different indexes (S4 Table) were sequenced on the Illumina HiSeq 2500 platform at 100 cycles. MOV10 HITS-CLIP data were analyzed as previously described [25]. The MOV10 HITS-CLIP-seq data are available under the GEO accession no: GSE217336.

440
The expression of MOV10-bound transcripts in mouse somatic tissues was obtained by reanalyzing the RNA-seq data from these mouse tissues: liver, bone marrow, bone, retina, kidney, prefrontal cortex, and brain stem [45][46][47][48][49][50]. Adapter sequences were trimmed and the low-quality reads were removed. The clean reads were mapped to the mouse genome (mm10) using STAR with default parameters. The number of reads mapped to each gene was counted by htseq-count (http://www-huber.embl.de/users/anders/HTSeq/) based on the 455 annotation from ENSEMBL (http://uswest.ensembl.org/) mouse gene annotation v99. The expression of transposable elements was analyzed using TEtranscripts [61]. Identification of differentially expressed genes was performed by edgeR. Differential expression was defined as a fold change greater than 2 and false discovery rate (FDR) < 0.05. FDR was calculated based on Benjamini and Hochberg multiple testing correction.

Statistics
Statistical analysis was performed with Student's t-test, if not otherwise described.

Acknowledgments
We thank Wenfeng An for the L1 tg mice, Hsin-Yao Tang and Thomas Beer for mass spectrometry, Jonathan Schug for Next-gen sequencing, Shantan Reddy, Aoife Roche, and Frederic Bushman for advice on the retrotransposition analysis. We thank Frederic Bushman and Leslie King for critical reading of the manuscript.

Competing interests:
The authors declare that they have no competing interests. Data and materials availability: The CLIP-seq and RNA-seq data that support the findings of this study are publicly available from NCBI under the GEO accession no: GSE217336.  is the first in vivo host restriction factor of L1 retrotransposition. L1 transcript is normally present at a very low level, due to transcriptional silencing of L1 by methylation of the CpG dinucleotides in the L1 770 promoter. New L1 insertions are short (<1 kb) and often 5 truncated due to incomplete reverse transcription.