Introduction

Molecular domestication of transposable elements (TEs) or TE-derived sequences give rise to novel genes and regulatory sequences in the genome contributing to both genetic and epigenetic variation in an organism (Jangam et al, 2017; Miller et al, 1992; Smit, 1999). TEs are key drivers of genome instability and thus are constantly exposed to silencing mechanisms by the host. Yet, co-option of TEs thorough evolution has created so-called domesticated genes beneficial to the gene network in a wide range of organisms (Modzelewski et al., 2022; Jangam et al., 2017). One example of such domesticated gene is LINE-1 type transposase containing domain 1 (L1TD1), which is the only domesticated protein-coding gene that originated from LINE-1 (McLaughlin et al, 2014). L1TD1 shares sequence resemblance with LINE-1 ORF1 protein (L1 ORF1p) that is required for retrotransposition (Martin, 2006) and L1TD1 was recently shown to facilitate the dissolution of stress granules (Jin et al, 2024). L1 ORF1p acts as nucleic acid chaperone with binding capacity to both RNA and DNA (Martin, 2006) and contributes to forming L1 ribonucleoproteins (L1-RNPs), key factors for retrotransposition of LINE-1 elements, but its co-factors are yet to be elucidated (Wallace et al, 2008).

L1TD1, also named Embryonic Stem Cell Associated Transcript 11 (ECAT11), is expressed predominantly only in early embryogenesis and germ cells and gets rapidly downregulated upon the exit from pluripotency (Iwabuchi et al, 2011; Mitsui et al, 2003). Although it is dispensable for mouse development (Iwabuchi et al., 2011), L1TD1 is essential for the maintenance of human pluripotency (Narva et al, 2012; Emani et al, 2015). In the context of human embryonic stem cells (hESCs), L1TD1 regulates translation of a distinct set of mRNAs (Jin et al., 2024), including core pluripotency factors (Emani et al., 2015). It has been proposed, that during mammalian evolution the L1TD1 gene was under positive selection due to the function in the maintenance of pluripotency in some species (McLaughlin et al., 2014). Apart from early development, L1TD1 is expressed in germ cell tumors and colorectal cancers. In this context, L1TD1 was identified as a possible prognostic marker for medulloblastoma (Santos et al, 2015) and colorectal tumors (Chakroborty et al, 2019), where its expression is highly correlative with malignancy progression (Urh et al, 2021).

DNA methyltransferase (DNMT) inhibitors are successfully used as anti-cancer drugs and two recent reports suggest that the lethal response of human tumor cells to these compounds is based to a large extent on the activation of an anti-viral immune response to endogenous retroviral transcripts (Chiappinelli et al, 2015; Roulois et al, 2015). DNMT inhibitor treatment of non-small cell lung carcinoma (NSCLC) patient-derived cells resulted in up-regulation of L1TD1 and reduced cell viability indicating an epigenetic control of L1TD1 (Altenberger et al, 2017). Accordingly, xenograft tumors with NSCLCs overexpressing L1TD1 showed decreased tumor growth suggesting a negative impact of L1TD1 expression on tumor viability. In contrast, L1TD1 was shown to be required for cell viability in medulloblastoma (Santos et al., 2015). Hence, the exact mechanistic function of this domesticated transposon protein in human tumor cells is still unrevealed.

Here, we aimed at elucidating the molecular function of L1TD1 and its potential role for the control of viability in human tumor cells. To this end, we activated L1TD1 expression by inducing DNA hypomethylation via deletion of the DNA maintenance methyltransferase DNMT1 in the nearly haploid human cancer cell line HAP1 (Carette et al, 2011). First, we found that L1TD1 expression and LINE-1 activation correlate with local DNA hypomethylation. Next, we identified L1TD1-associated RNAs by RIP-seq, which revealed transcripts and proteins with differential expression in the absence of L1TD1 by transcriptome and proteome analyses. We showed that L1TD1 protein binds to L1-RNPs. Importantly, L1TD1 facilitated LINE-1 retrotransposition in HAP1 cells. Our results demonstrated that upon DNA hypomethylation L1TD1 affects gene expression and cell viability and cooperates with its ancestor protein L1 ORF1p in the control of LINE-1 retrotransposition.

Materials and methods

Cell lines and cell culture

HAP1 is a near-haploid human cell line derived from the KBM-7 chronic myelogenous leukemia (CML) cell line (Andersson et al, 1987; Carette et al., 2011). HAP1 cells were cultured in Iscove’s Modified Dulbecco Medium (IMDM, Sigma Aldrich, I3390) supplement with 10% Fetal Bovine Serum (Sigma Aldrich, F7524) and 1% penicillin/streptomycin (Sigma Aldrich, P4333).

The human malignant papillary serous carcinoma cell line OV-90 (Gralewska et al, 2021) was cultured in MCDB 105 (Sigma Aldrich, 117) and Medium 199 (Sigma Aldrich, M4530) in a 1:1 ratio supplemented with 15% Fetal Bovine Serum (Sigma Aldrich, F7524) and 1% penicillin/streptomycin (Sigma Aldrich, P4333). The cells were kept in a humidified incubator at 37°C and 5% CO₂.

Gene editing

DNMT1 knockout (KO) HAP1 cells harboring a 20 bp deletion in the exon 4 of the DNMT1 gene were generated using CRISPR/Cas9 gene editing. DNMT1/L1TD1 double knockout (DKO) HAP1 cells were generated by CRISPR/Cas9 gene editing resulting in a 13 bp deletion in exon 4 of the L1TD1 gene. CRISPR/Cas9 gene editing was performed by Horizon Genomics (Horizon Discovery).

Viability and Apoptosis Assay

Both assays were performed as described previously (Hess et al, 2022). For the viability assay 2x10³ cells of each cell line were seeded in triplicates in solid white 96-well plates and allowed to attach to the plates overnight. Cell viability was measured using CellTiter-Glo Luminiscent Cell Viability Assay (Promega, G7571) and Glomax Discover plate reader (Promega). The measurement was performed every 24 hours starting from the day after seeding (24h, 48h, 72h). For the apoptosis assay 5x10⁶ cells were collected 48hours after seeding, fixed for 20 min 2% paraformaldehyde (Merck Life Science) and for 30 min in 75% ethanol. Cells were permeabilized with 0.1% TritonX100 (Merck Life Science) for 10 min and blocked in 10% donkey serum (Merck Life Science) for 30 min and then incubated with 1:50 diluted anti-cleaved caspase 3 antibody in phosphate-buffered saline (PBS) (Cell Signaling, 9661) for 1 hour. Cells were washed and incubated with 1:400 Alexa 488 anti-rabbit secondary antibody (Jackson Immuno Research, 751-545-152). Samples were acquired using FACSCelesta flow cytometer (BD Bioscience) and the data were analyzed using FlowJo software 10.6.1 (BD Bioscience)

Immunoprecipitation

HAP1 cells pellets harvested from 15-cm culture plates were lysed in Hunt Buffer (20mM Tris/HCl ph 8.0, 100mM NaCl, 1mM EDTA, 0.5% NP-40) supplemented with Complete Protease inhibitor cocktail (Roche, 11697498001), Complete Phosphatase inhibitor cocktail (Roche, 11836145001), 10mM sodium fluoride, 10mM β-glycerophosphate, 0.1mM sodium molybdate and 0.1mM PMSF. The lysis was achieved by repeating freeze-thaw three times followed by centrifugation 12.000g for 15 min at 4 °C. Cell lysate was collected and the protein concentration was measured using Bradford assay. 40ul of Dynabeads^TM Protein G beads (ThermoFisher Scientific, 10004D) was blocked with 10% BSA in a rotor at 4°C for 1 hour. 1mg of cell lysate was incubated with blocked Protein G beads and monoclonal L1 ORF1p antibody (EMD Millipore, MABC1152) at 12 rpm at 4°C overnight. The next day, the beads were washed 3 times with Hunt buffer.

Western Blot Analysis

Immunoprecipitated complex was eluted from the beads with 1x SDS loading dye followed by heat-incubation at 95 °C for 5 minutes for detection of proteins. For input samples, 20µg whole cell lysate was similarly denatured as IP samples. Proteins were separated in SDS-polyacrylamide gel and transferred onto nitrocellulose membrane (Amersham Protran, GE10600001, Sigma Aldrich) by wet transfer method. The membrane was blocked with the blocking solution (1x PBS, 1% polyvinylpyrrolidone, 3% non-fat dried milk, 0.1% tween-20, 0.01% sodium azide, pH 7.4) and incubated with the corresponding antibodies listed in Supplementary Table 6. To detect the protein of interest, ECL Western blotting detection reagents were used with a FUSION FX chemiluminescence imaging system.

Mass spectrometry analysis

Proteomes of DNMT1 KO and DNMT1/L1TD1 DKO cells were analyzed by quantitative TMT (Tandem Mass Tag) multiplex mass spectrometry analysis. Proteins were precipitated from cell extracts with acetone. After reduction and alkylation of Cys under denaturing conditions, proteins were digested with LysC/trypsin overnight. Peptides were cleaned-up using Oasis MCX (Waters) and labeled with TMT6plex (Thermo Fisher) according to manufacturer’s protocol. Equal amounts of each channel were mixed, desalted and fractionated by neutral pH reversed phase chromatography. Equal interval fractions were pooled and analysed by LC-MS3 on an Ultimate 3000 RSLCnano LC coupled to an Orbitrap Eclipse™ Tribrid™ mass spectrometer via a FAIMS Pro ion mobility interface (all Thermo Fisher). Data were analyzed with MaxQuant 1.6.7.0. (Tyanova et al, 2016). Differentially enriched proteins were determined with LIMMA (Ritchie et al, 2015). Workflow and analysis are described in detail in the Supplementary information.

RNA Immunoprecipitation

The native RIP protocol was followed as previously described in (Yang et al, 2017). In brief, cells were resuspended in 1ml Polysomal Lysis Buffer (100mM KCl, 5mM MgCl₂, 10mM HEPES (pH 7.0), 0.5% NP-40, 1mM DTT) supplemented with Protector RNase inhibitor (Sigma Aldrich, 3335399001) and protease inhibitor cocktail (Roche, 11697498001). Cell lysis was facilitated with 27G needle and syringe at least 7 times. The lysates were centrifuged at 12.000g for 15 min at 4 °C. 1% input was taken from the cell lysate for RIP-seq experiment. The supernatant was cleared via pre-incubation with Dynabeads^TM Protein G beads (ThermoFisher Scientific, 10004D) rotating for 1h at 4 °C. For immunoprecipitation 1mg of pre-cleaned cell lysate was incubated with 4µg mouse monoclonal L1TD1 antibody (R&D systems, MAB8317) at 4 °C at 20rpm on the rotor overnight. 40ul Dynabeads^TM Protein G beads per IP was blocked with 10 % bovine serum albumin (BSA) for 1h at 4 °C. Blocked beads were added to the lysate and antibody mixture rotating for 1h at 4 °C. Next, the beads were pelleted and washed 3 times with polysomal lysis buffer. 20% of the RIP was used for Western blot analysis to confirm immunoprecipitation of L1TD1. 1% of the RIP was saved as input. All experiments were performed in biological triplicates.

RNA isolation

To extract total RNA, immunoprecipitated RNA and 1% input RNA Monarch RNA Cleanup Kit (New England Biolabs, T2047L) was used according to the manufacturers’ instructions. In the case of RIP samples, the beads were first treated with DNase I (New England Biolabs, M0303S) at 37 °C for 15 minutes and 1ml TRIzol (Thermo Fischer Scientific, 15596018) was directly added on. Upon addition of 150 µl chloroform the mixture was vortexed vigorously for 15 sec and centrifuged for at 12.000g for 15 minutes at 4 C° . Aqueous phase was transferred to a new Eppendorf tube and mixed with 1 volume of EtOH (>95%). The mixture was loaded on the RNA columns and spun for 1 minute. The columns were washed 2 times with the washing buffer (supplemented with the kit) and eluted in 20ul nuclease free water. RNA concentration was measured using the Qubit RNA High Sensitivity kit (Thermo Fischer Scientific, Q32852)

cDNA library preparation

The cDNA libraries for the RIP-seq experiment were prepared as described in Smart-seq3 protocol by Hageman-Jensen et al., 2020

cDNA synthesis and qRT-PCR

1ug of total RNA was reverse transcribed using iScript cDNA synthesis Kit (Bio-Rad, 1708891). Reverse transcription reaction was performed as follows: 25 °C for 5 minutes, 46 °C for 30 minutes, 95 °C for 5 minutes. Resulting cDNA was diluted to 1:10 with nuclease free water. 5µl of the cDNA was used for SYBR Green qPCR Master mix (Bio-Rad 1725275) together with the primers used in this study which were listed in Supplementary Table 5.

Methylight Assay

Genomic DNA was extracted from DNMT1 KO, DNMT1/L1TD1 DKO and WT HAP1 cells using the Wizard Genomic DNA isolation kit (Promega) following the manufacturer’s protocol. Next, 1 µg of genomic DNA per sample was subjected to bisulfite conversion using the EZ DNA Methylation Kit (Zymo research Cat. no. D5001) following the manufacturer’s instructions. Bisulfite-treated DNA was eluted in ddH₂O to a final concentration of 10 ng/μL and stored at -20°C until further use. The MethyLight method was performed as previously described by Campan et al. (Campan et al, 2009) using primers for L1TD1 and LINE-1; ALU was used as a reference. The primer sequences are provided in Supplementary Table 5. The MethyLight reactions were done in technical triplicates. Each reaction contained 7.5 μL of 2X TaqMan Universal PCR Master Mix (Thermo Fisher Cat. No. 4324018), 300 nM of each primer and 100 nM of probe. For the L1TD1 reactions 50ng and for the LINE-1 reactions 20 ng of bisulfite-converted DNA were used in a total volume of 15 μL. The reactions were performed on a BioRad CFX96 thermocycler with an initial incubation at 95°C for 10 minutes, followed by 50 cycles of 95°C for 15 sec and 60°C for 1 minute. Each run included the individual samples, a serial dilution of fully methylated DNA, a non-methylated DNA control (Human Methylated & Non-Methylated (WGA) DNA Set D5013 Zymo research), and a non-template control. The PMR (percentage of methylation reference) of each individual sample was calculated using the following formula: 100 x [(GENE-X mean value) sample/ (ALU mean value) sample] / [(GENE-X mean value)100% methylated / (ALU mean value)100% methylated].

RNA-seq and RIP-seq Analyses

RNA sequencing data were analyzed using the RNA-Seq pipeline of IMP/IMBA Bioinfo core facility (ii-bioinfo@imp.ac.at). The pipeline is based on the nf-core/rnaseq pipeline [doi: 10.5281/zenodo.1400710] and is built with nextflow (Ewels et al, 2020).The RNA-seq analysis was performed as described in the following : Adapters were clipped with trimgalore (Martin, 2011). Abundant sequence fractions (rRNA) were removed using (bowtie2) (Langmead & Salzberg, 2012). Cleaned raw reads were mapped against the reference genome (hg38,Homo_sapiens.GRCh38.107.gtf) with STAR (reverse_stranded, STAR) (Dobin et al, 2013). Mapped reads were assigned to corresponding genes using featureCounts (Liao et al, 2014). Abundances were estimated through Kallisto (Bray et al, 2016) and Salmon (Patro et al, 2017). Analysis of differentially expressed genes was performed using DESeq2 (Love et al, 2014). TEtranscript software (Jin et al, 2015) was used to identify differentially expressed TE-derived sequences. In the case of RIP-seq, differential enrichment of transcripts relative to input as well as to the negative control (HAP1 DNMT1/L1TD1 DKO) was calculated using DEseq2. Only transcripts enriched (log₂FC>2, adjusted p-value < 0.05) in both assays (DNMT1 KO versus input and DNMT1 KO versus DNMT1/L1TD1 DKO) were considered as L1TD1 associated RNAs.

L1 Retrotransposition Assay

Transfection of HAP1 cells with pJJ101, pJJ105, EGFP, pLenti6.2, pCEP4 was performed using polyethyleneimine (PEI)/NaCl solution. Briefly, 0.2 x 10⁶ HAP1 cells/well were seeded into 6-well plates. 24 hours later and at 70% confluence, medium was replaced by fresh IMDM without penicillin/streptavidin followed by transfection. For each transfection, 100μl of 150mM NaCl, 7μl of PEI were mixed with DNA/NaCl solution (100μl of 150mM NaCl, 4μg of plasmid DNA), incubated at room temperature for 30 minutes, followed by adding the mixture in a drop-wise fashion onto cells. 18 hours post-transfection, the media was replaced with complete IMDM including 1% penicillin/streptavidin. Transfection efficiency was monitored 48 hours after transfection using fluorescent microscopy.

The retrotransposition assay was performed as described in (Kopera et al, 2016) with the following modification. To measure retrotransposition efficiency, 2x10⁵ cells per well were seeded in 6-well culture plates and cultured at 37°C overnight. On day 0, 4ug of pJJ101/L1.3, pJJ105/L1.3, pLenti6.2, pCEP4 vectors described in Kopera et al., 2016 were transfected separately using PEI/NaCl in DNMT1 KO and DNMT1/L1TD1 DKO cells. To correct for potential differences in transfection efficiency and susceptibility to blasticidin in DNMT1 KO and DNMT1/L1TD1 DKO cells, we additionally used blasticidin deaminase containing vector (pLenti6.2). The following day (d1) the media was replaced with IMDM medium containing 1% penicillin/streptavidin. On d3, 2x10⁵cells per well seeded into 6-well culture dishes for each genotype and condition. Blasticidin treatment 10ug/ml was started on d4 and the cells were cultured 37°C for 9 days without medium change. Blasticidin resistant colonies were fixed on d13 with 4% Methanol-free formaldehyde (Thermo Fischer Scientific, 28908) at room temperature for 20 minutes and stained with 0.1% bromophenol blue (w/v) at room temperature for 1 hour. The pictures of the wells containing the colonies were taken in FX chemiluminescence imaging system. The pictures were further processed inFiji Software following Analyze Particles function (Schindelin et al, 2012). The colony counts were obtained from three technical replicates per transfection. Mean colony counts were calculated and adjusted retrotransposition mean was calculated by adjusting for blasticidin resistant colonies in in blasticidin vector (pLenti6.2) transfected cells.

Indirect Immunofluorescence Staining

The protocol with minor modification from Sharma et al. (Sharma et al, 2016) was performed for immunostaining. Briefly, 75x10³ HAP1 cells per well were plated on cover slips in 12 well plates. Next day, the cells were washed with 1x PBS and fixed with 4% methanol-free formaldehyde at room temperature for 10 minutes. The cells were washed with 1x PBS/Glycine pH 7.4 and permeabilized with 0.1% Triton-X in1x PBS/Glycine for 3 minutes. The samples were blocked with 1x PBS/Glycine/1% BSA at room temperature for 1 hour. After blocking, the cells were stained primary antibodies in blocking solution at 4 °C overnight. On the following day, the cells were washed three times with 1x PBS/Glycine/1% BSA for 5 minutes and stained with secondary antibody in blocking solution for 2 hours in dark. After a wash with PBS/Glycine, they were incubated with DAPI (Sigma Aldrich, D9542, 1:10000) in PBS/Glycine at room temperature for 10 minutes and washed three times with 1x PBS/Glycine pH 7.4. The slides were mounted with ProLong Gold Mountant (Thermo Fischer Scientific, P36930) and dried. The staining was monitored by Olympus Confocal Microscope.

Results

L1TD1 expression is activated through DNA hypomethylation in HAP1 cells

To address the function of L1TD1 in human tumor cells, we generated a defined cellular model by using the nearly haploid tumor cell line HAP1 cells. By CRISPR/Cas-9 mediated gene editing, we deleted DNMT1 in these cells resulting in HAP1 DNMT1 knock-out (KO) cells. As shown in Figure 1A, DNA methylation-specific PCR (MethyLight) analysis revealed significantly reduced DNA methylation at the L1TD1 promoter in DNMT1 KO cells. DNA hypomethylation at the L1TD1 gene correlated with a strong induction of L1TD1 mRNA and protein expression in DNMT1 KO cells (Figure 1B-D). To analyze the role of L1TD1 in a DNMT1 null background, we ablated L1TD1 in KO cells by gene editing resulting in HAP1 DNMT1/L1TD1 double knock-out (DKO) cells (Figure 1C). Knowing that L1TD1 is highly expressed in human germ cell tumors, we used the ovarian cancer cell line OV-90 as a positive control (Figure 1D-E). In accordance with previously published data with human embryonic stem cells (hESCs), indirect immunofluorescence analysis revealed that L1TD1 was preferentially localized in the cytosol of HAP1 DNMT1 KO cells and OV-90 cells (Figure 1E). L1TD1 was reported to localize in P-bodies in hESCs (Narva et al, 2012). We detected a similar granular localization of L1TD1 in HAP1 DNMT1 KO cells (Figure 1E). DNMT1 depletion did not affect cell proliferation but additional loss of L1TD1 in DNMT1/L1TD1 DKO cells led to a significant reduction in the number of viable cells, and increased apoptosis (cleaved caspase 3) and DNA damage (γ-H2AX) (Figure 1F-H). These results suggest that L1TD1 expression is regulated by DNA methylation and its depletion affects cell viability in DNMT1-deficient HAP1 cells.

Figure 1.

L1TD1 binds to LINE-1 transcripts and a subset of mRNAs

L1TD1 was previously shown to act as RNA-binding protein by binding to its own transcript (Narva et al., 2012) and selected group of mRNAs under L1TD1 translational control (Jin et al., 2024). To determine the L1TD1 binding repertoire in the context of L1TD1- and DNMT-dependent cell viability and proliferation control, we performed RNA immunoprecipitation (RIP) followed by sequencing (RIP-seq) (Figure 2A). L1TD1-containing RNP complexes (L1TD1-RNPs) were immunoprecipitated in DNMT1 KO cells by an L1TD1-specific antibody and DNMT1/L1TD1 DKO cells were used as a negative control. We identified 597 transcripts enriched (log₂FC>2, adjusted p-value < 0.05) in L1TD1-RNPs when compared to the input (Figure 2B and Suppl. Table S1). Upon stringent analysis excluding non-specific binders and artifacts (Suppl. Figure S1A and Suppl. Table S1), 228 transcripts were specifically detected in L1TD1-RNPs. Importantly, L1TD1 mRNA was detected as one of the top transcripts in L1TD1-RNPs (Figure 2B).

Figure 2.

To validate the RIP-seq results, individual transcripts selected from a list of top hits were amplified by gene-specific qRT-PCR analysis after repeating the RIP. Transcripts of L1TD1, ARMC1 and SYNJ2BP were specifically enriched in L1TD1-RNPs obtained from DNMT1 KO cells when compared to DNMT1/L1TD1 DKO cells. Of note, L1TD1 was also associated with the transcript of YY2, a retrotransposon-derived paralogue of yin yang 1 (YY1) (Tahmasebi et al, 2016), whereas the negative control GAPDH did not show such enrichment (Figure 2C). Gene Ontology analyses of 228 common hits indicated enriched groups in cell division and cell cycle and, in accordance with a recent study (Jin et al., 2024), enriched transcripts encoding zinc finger and in particular KRAB domain proteins which have been implicated in the restriction of endogenous retroviruses (Rosspopoff & Trono, 2023) (Suppl. Figure S1B).

Analysis of RIP-Seq using TEtranscript designed for the analysis of transposon-derived sequences (Jin et al., 2015), in the differentially expressed sets indicated an association of L1TD1 with LINE-1 transcripts (Figure 2D and Suppl. Table S2), which was also validated by qRT-PCR analysis of L1TD1 RIP samples (Figure 2E). In addition, we also identified ERV and AluY elements as L1TD1-RNP-associated transcripts (Suppl. Table S2). These combined results suggest that L1TD1 interacts with a set of transcripts including RNAs with functions in cell division. Furthermore, L1TD1 has kept the ability to associate with LINE-1 transcripts, and potentially regulating them similar to its ancestor protein L1 ORF1p (L1 ORF1p) (Wei et al, 2001).

Loss of L1TD1 modulates the proteome and transcriptome of HAP1 cells

To explore the cellular function of L1TD1, we analyzed the proteomes and transcriptomes of DNMT1 KO and DNMT1/L1TD1 DKO cells. Comparison of the DNMT1 KO and DNMT1/L1TD1 DKO proteomes revealed 98 upregulated and 131 downregulated proteins in DNMT1/L1TD1 DKO (log₂FC ≥ 1, adj. p-value ≤ 0.05) (Figure 3A and Suppl. Table S3). Gene Ontology Enrichment Analysis revealed that differentially expressed proteins were mainly associated with regulation of cell junction and cysteine-type endopeptidase activity involved in apoptotic process (Suppl. Fig. S2A). To investigate whether the corresponding protein abundance of mRNAs contained in L1TD1-RNPs is affected by L1TD1 expression, we compared proteomic data with the RIP-seq data. Comparison of differentially expressed proteins with transcripts associated with L1TD1 identified only L1TD1 (Suppl. Fig. S2B). The lack of overlap between RIP-seq and proteomics data (except L1TD1) suggested that modulation of the abundance of proteins encoded by L1TD1-associated transcripts is not the main function of L1TD1.

Figure 3.

Interestingly, we discovered that the L1 transposon protein ORF1p was up-regulated in the absence of L1TD1 (Suppl. Table S3). Western blot analysis confirmed the increased expression of L1 ORF1p in HAP1 DNMT1/L1TD1 DKO cells when compared to DNMT1 KO cells (Figure 3B), while LINE-1 transcript levels showed no significant change (Suppl. Figure S3A). This indicates that L1TD1 attenuates the expression of L1 ORF1p suggesting a potential regulatory crosstalk between L1TD1 with its ancestor protein.

To investigate a potential effect of L1TD1 on the expression levels of associated RNAs, comparative RNA-seq analysis was performed with DNMT1 KO and DNMT1/L1TD1 DKO cells. Upon loss of L1TD1, we identified 323 up-regulated and 321 down-regulated transcripts, respectively (log₂-fold change ≥ 1; adjusted p-value < 0.05) (Figure 3B and Suppl. Table S4). About 25% of the differentially expressed proteins (35% of upregulated and 21% of downregulated proteins) identified in the proteome analysis were also significantly deregulated at the RNA level (Suppl. Table S3). However, none of the deregulated transcripts, except L1TD1, was detected in L1TD1-RNPs (Figure 3B and Suppl. Tables S1 and S4). This finding was corroborated by qRT-PCR analysis of L1TD1-RNP-associated transcripts (Suppl. Figure S3B) suggesting that the cellular function of L1TD1 might be independent of a deregulation of its associated mRNAs.

Based on our observations implying that L1TD1 does not exert its cellular function as translational regulator while being associated with transposon RNAs (Figure 2D), we used TEtranscript (Jin et al., 2015) to identify differentially expressed TE-derived sequences transcriptome-wide. Loss of L1TD1 resulted in upregulation of specific SINEs such as AluY elements (Suppl. Figure S3C and Suppl. Table S4). Taken together, these data suggest that L1TD1 affects the abundance of transposon transcripts in DNMT1-deficient HAP1 cells.

L1TD1 interacts with the LINE-1 ORF1p protein

An interaction of L1TD1 with LINE-1 RNA (Figure 2D) and increased L1 ORF1p levels in L1TD1-depleted cells (Figure 3B) imply a direct association of L1TD1 with L1 ORF1p. We thus hypothesized that L1TD1 associates with L1-RNPs via L1 ORF1p. Therefore, we immunoprecipitated L1 ORF1p from HAP1 DNMT1KO and OV-90 cells and tested the IPs for the presence of endogenous L1TD1. Indeed, L1TD1 was co-precipitated with L1 ORF1p from DNMT1 KO cells (Figure 3D). Next, we showed that the association of L1TD1 with L1 ORF1p was RNA-independent (Figures S4A and S4B).

In agreement with their association in protein complexes, indirect immunofluorescence experiments revealed a partial co-localization of L1TD1 and L1 ORF1p in both cell lines (Figure 3E and Suppl. Figure S4C). In summary, our results suggest L1TD1 associates with its ancestor ORF1p and this association does not depend on RNA intermediates.

L1TD1 has a positive impact on LINE-1 retrotransposition

Since L1TD1 associates with LINE-1 RNA and modulates L1 ORF1p expression, we next asked whether the domesticated transposon protein can impact LINE-1 retrotransposition in HAP1 cells. To this end, we carried out plasmid-based retrotransposition assays (Kopera et al., 2016) in DNMT1 KO and DNMT1/L1TD1 DKO cells (Figure 4A-B). Following transfection and blasticidin selection, the rate of retrotransposition of the reporter construct was assessed through counting blasticidin-resistant colonies, transfected either with a retrotransposition-competent reporter construct or a retrotransposition-deficient control. An average of 506 colonies were observed in DNMT1 KO cells in three independent experiments whereas DNMT1/L1TD1 DKO cells yielded over 5-fold fewer colonies when transfected with the retrotransposition-competent reporter construct (86 colonies on average) (Figures 4C-D). No blasticidin resistant clones were obtained for the negative controls, but comparable retrotransposition rates were observed for retrotransposition-competent reporter constructs in three independent retrotransposition assays (Suppl. Figures S5A-B). Combined, these data suggest that L1TD1 enhances LINE-1 retrotransposition in DNMT1-deficient HAP1 cells and this effect is mediated through the association with ORF1p and LINE-1 RNA.

Figure 4.

Discussion

In this study we show that the domesticated transposon protein L1TD1 promotes LINE-1 retrotransposition. Numerous cellular factors have been shown in the past to affect retrotransposition (Ariumi, 2016; Goodier, 2016; Pizarro & Cristofari, 2016; Protasova et al, 2021). However, many of these proteins are part of host defense mechanisms and restrict retrotransposition, and it was speculated that L1TD1 also limits L1 retrotransposition (McLaughlin et al., 2014). Unexpectedly, we identified L1TD1 as one of the cellular factors that has a positive impact on L1 retrotransposition, most likely by interacting with L1 RNPs and acting as RNA chaperone. L1TD1 deletion resulted in reduced L1 retrotransposition despite upregulation of ORF1p indicating that the upregulation of the transposon protein cannot fully compensate the loss of L1TD1. Interestingly, a similar effect has been previously described for the nonsense-mediated decay factor UPF1 (Taylor et al, 2013). UPF1 knockdown increased the amount of L1 mRNA and proteins but simultaneously reduced the effectiveness of the retrotransposition. A positive effect of L1TD1 on retrotransposition was recently observed upon L1TD1 overexpression in HeLa cells (Jin et al., 2024). However, the authors considered this as artefact of ectopic L1TD1 expression. L1TD1 was lost or pseudogenized multiple times during mammalian evolution and has evolved under positive selection during primate and mouse evolution suggesting an important function during development (McLaughlin et al., 2014). L1 elements can retrotranspose in the germline, in embryonic stem cells and in the early embryo (Ariumi, 2016; Goodier, 2016; Pizarro & Cristofari, 2016; Protasova et al., 2021). Mouse embryos with repressed LINE-1 expression show impaired embryonic development, suggesting that LINE-1 is part of the developmental program of early embryos (Jachowicz et al, 2017). The simultaneous L1TD1 expression during early embryogenesis might ensure activation of L1 retrotransposition at this developmental stage (Iwabuchi et al., 2011).

To gain insight into the regulatory function of L1TD1 in tumor cells, we performed RIP-seq, transcriptome and proteome analyses in HAP1 cells. The RIP-seq approach identified, in addition to the known associated L1TD1 mRNA (Narva et al., 2012), a defined set of transcripts comprising mRNAs, lncRNAs and transposable elements including LINE-1 transcripts. These findings indicate that L1TD1 has not only inherited its function as RNA-binding protein but also its affinity for transposon transcripts. Recently, Jin et al. published a study on the impact of L1TD1 on the translation in hESCs and identified L1TD1-bound RNAs by CLIP-seq (Jin et al., 2024). Interestingly, the majority of L1TD1-associated transcripts in HAP1 cells (69%) identified in our study were also reported as L1TD1 targets in hESCs suggesting a conserved binding affinity of this domesticated transposon protein across different cell types.

The same study (Jin et al., 2024) showed that L1TD1 localizes to high-density RNP condensates and enhances the translation of a subset of mRNAs enriched in the condensates. Our mass spectrometry data showed that L1TD1 ablation leads to a significantly changed proteome. However, none of the proteins encoded by the mRNAs associated with L1TD1 showed significantly changed abundance in the absence of L1TD1 suggesting that this is not a direct effect of L1TD1 association with the respective mRNAs. The different outcomes might be due to different cell types and methods used in the two studies. Similarly, and in accordance with the hESC study (Jin et al., 2024), we did not find an overlap between RNAs associated with L1TD1 and their deregulation at the RNA level. This suggests that in HAP1 cells L1TD1 does not have a global effect on RNA stability and translation of its associated RNAs. This is consistent with a recent study on L1 bodies, where RNA processing association of L1 body components did not correlate with RNA stability or translational control of the RNA targets of L1 bodies (De Luca et al, 2023).

Of note, L1TD1 associated with SINE transcripts such as AluY and L1TD1 ablation positively affected AluY transposon expression. This relatively young subfamily of SINE elements is still retrotransposition-competent and their mobilization depends on the activity of LINE proteins (Roy-Engel, 2012; Styles & Brookfield, 2009). These observations suggest that L1TD1 has not only inherited its function as an RNA-binding protein but also the affinity to LINE-1 and SINE transcripts. L1TD1 association with AluY transcripts could potentially be connected to modulation of LINE-1 and SINE-1 transposon activities.

LINE-1 is the only protein-coding transposon that is active in humans and LINE-1 overexpression and retrotransposition are hallmarks of cancers (Kazazian & Moran, 2017; Mendez-Dorantes & Burns, 2023). Similarly to the epigenetic control of the ancestral LINE-1 elements (Sanchez-Luque et al, 2019; Zhou et al, 2020), L1TD1 expression in HAP1 cells is restricted by DNA methylation and DNA hypomethylation induced by DNMT1 ablation results in robust L1TD1 RNA and protein expression. This observation is in line with the findings in the NSCLC cancer model, where pharmacological DNMT inhibition resulted in L1TD1 induction (Altenberger et al., 2017). In contrast to the negative effect of L1TD1 on cell viability and tumor growth in the NSCLC xenograft model, we report that L1TD1 deletion in DNMT1-deficient cells led to reduced cell viability and increased apoptosis in HAP1 cancer cells. This is in accordance with the observation that L1TD1 has a positive impact on cell viability in medullablastoma (Santos et al., 2015) and suggests a (cancer) cell type-specific effect of L1TD1 that might be related to the DNA methylation state of the tumors.

Taken together we present here a novel cellular function for a domesticated transposon protein by showing that L1TD1 associates with L1-RNPs and promotes LINE-1 retrotransposition. This function might have a beneficial effect during early development but could also impact on tumorigenesis.

Acknowledgements

We would like to thank Wolfgang Sommergruber for the screen of human tumor cells for L1TD1 expression and Jernej Ule for help with the analysis of RNPs. We also want to thank Luisa Seufert, Stephanie Schneider, Christina Maria Schuh and Marlene Müller for help with the characterization of transgenic HAP1 cells, Brigitte Gundacker, Urska Janjos and Milena Mijovic for professional technical support and Dorothea Anrather for help with data analysis. We are also grateful to Wolfgang Miller, Heinz Fischer and Matthias Schaefer for numerous helpful discussions.

GL and TM were supported by the Austrian Science Foundation (FWF, doc.funds grant DOC59), CS was supported by the Austrian Science Foundation (P34998, DOC32) and the Austrian Research Promotion Agency (FFG): Bridge Early Phase project 5722451. GK were PhD students of the doc.funds program (DOC32) supported by the Austrian Science Foundation (FWF). TPC was supported by a Student Fellowship from the Ernst Mach Grant - ASEA-UNINET through the Austrian Academic Exchange (OeAD).

Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al, 2022) partner repository with the dataset identifier PXD047402. RNA-seq and RIP-seq data were submitted to GEO under accession numbers GSE254459 and GSE254460.