LINE-1 hypomethylation, increased retrotransposition and tumor-specific insertion in upper gastrointestinal cancer

Abstract

The long interspersed nuclear element-1 (LINE-1) retrotransposons are a major family of mobile genetic elements, comprising approximately 17% of the human genome. The methylation state of LINE-1 is often used as an indicator of global DNA methylation levels and it regulates the retrotransposition and somatic insertion of the genetic element. We have previously reported the significant relationship between LINE-1 hypomethylation and poor prognosis in upper gastrointestinal (GI) cancers. However, the causal relationships between LINE-1 hypomethylation, retrotransposition, and tumor-specific insertion in upper GI cancers remain unknown. We used bisulfite-pyrosequencing and quantitative real-time PCR to verify LINE-1 methylation and copy number in tissue samples of 101 patients with esophageal and 103 patients with gastric cancer. Furthermore, we analyzed the LINE-1 retrotransposition profile with an originally developed L1Hs-seq. In tumor samples, LINE-1 methylation levels were significantly lower than non-tumor controls, while LINE-1 copy numbers were markedly increased. As such, there was a significant inverse correlation between the LINE-1 methylation level and copy number in tumor tissues, with lower LINE-1 methylation levels corresponding to higher LINE-1 copy numbers. Of particular importance is that somatic LINE-1 insertions were more numerous in tumor than normal tissues. Furthermore, we observed that LINE-1 was inserted evenly across all chromosomes, and most often within genomic regions associated with tumor-suppressive genes. LINE-1 hypomethylation in upper GI cancers is related to increased LINE-1 retrotransposition and tumor-specific insertion events, which may collectively contribute to the acquisition of aggressive tumor features through the inactivation of tumor-suppressive genes.

Abbreviations
• ESCC
• esophageal squamous cell carcinoma
• LINE-1
• long interspersed nuclear element-1
• RORA
• Retinoic acid receptor-related orphan receptor-α
• SD
• standard deviations
• upper GI
• upper gastrointestinal

INTRODUCTION

A large proportion of the human genome consists of interspersed repeat sequences, with a small subset of these being actively propagated as mobile genetic elements.^1,2 Long interspersed nuclear element-1 (LINE-1, L1) retrotransposons are one of the most active and abundant mobile DNA elements in the human genome, with LINE-1 sequences occupying approximately 17% of the genome.³ Retrotransposon insertions can deeply alter gene structures and expression levels, and have also been associated with several human pathological conditions.^4–7 The activity of LINE-1 is normally suppressed in most somatic cells by methylation of a CpG island in the internal LINE-1 promoter.^8,9 In contrast, LINE-1 is frequently hypomethylated in tumor cells, with the removal of this key inhibitory modification favoring retrotransposition.⁴ Although increased LINE-1 retrotransposition has been reported in a range of malignancies and linked to cancer progression,^10–13 its significance in upper gastrointestinal (GI) cancer, including esophageal and gastric cancer, has not been fully elucidated.

DNA methylation alterations in human cancers include global DNA hypomethylation and site-specific CpG island promoter hypermethylation.¹⁴ Global DNA hypomethylation is a major factor that promotes genomic instability and leads to cancer development.^15–17 As the LINE-1 retrotransposon constitutes a substantial portion of the human genome, the level of LINE-1 methylation is often used to gauge the global DNA methylation state.¹⁸ The epigenetic landscape regarding the methylation of LINE-1 has been shown to vary considerably between normal and tumor genomes in several cancer types.^19–21 LINE-1 hypomethylation specifically has been reported to exhibit a strong positive correlation with the poor clinical outcome of patients for certain cancer types.^22–26 We have previously demonstrated that LINE-1 hypomethylation in upper GI cancers is linked to poor patient prognosis,^27,28 suggesting that LINE-1 hypomethylation may be used as a biomarker to identify patients at greater risk of experiencing unfavorable outcomes. Given the potential relationship between LINE-1 hypomethylation and active LINE-1 source elements in cancer genomes,¹³ we posit that LINE-1 hypomethylation may promote LINE-1 copy number amplification and increase retrotransposition events in upper GI cancers, leading to tumor-specific insertions in cancer-related genes.

To test this hypothesis, we first examined the relationship between LINE-1 methylation level and copy number amplification in tissue samples of more than 200 patients with esophageal or gastric cancer. Next, we evaluated the profile of somatic LINE-1 insertion using an originally developed L1Hs-seq method.²⁹ Following this approach, we identified tumor-specific insertions within several tumor-suppressive genes. To the best of our knowledge, this is the first study to demonstrate the causal relationships between LINE-1 hypomethylation, copy number amplification, and increased retrotransposition in upper GI cancers. Our findings suggest that LINE-1 hypomethylation can contribute to the acquisition of aggressive tumor behavior through LINE-1 retrotransposition and tumor-specific insertion in tumor-suppressive genes.

MATERIALS AND METHODS Study patients

All samples used in this study were collected at the Kumamoto University Hospital between April 2008 and December 2018. The methylation levels and copy number of LINE-1 were quantified in 204 upper GI cancer tumors (101 esophageal cancer and 103 gastric cancer) and macroscopically matched healthy mucosa samples from the same individuals, for which freshly frozen specimens were available. The patient characteristics for these 204 individuals are shown in Table S1. Total RNA and genomic DNA were obtained from 10 esophageal and 10 gastric cancer tumors, matched with normal tissues from the same patients. The tumor samples selected for further analysis were based on the availability of sufficient amounts of freshly frozen tissues and RNA quality. Their clinical and pathological characteristics are shown in Table S2. Written informed consent was obtained from each patient, and the study procedures were approved by the institutional review board (Kumamoto university, #565).

Pyrosequencing to measure LINE-1 methylation

Genomic DNA was extracted from frozen esophageal cancer specimens using a QIAamp DNA Mini Kit (Qiagen) following the manufacturer's protocol. Genomic DNA was modified with sodium bisulfite using an EpiTect Bisulfite kit (Qiagen). PCR and subsequent pyrosequencing for LINE-1 were performed as previously described by Ogino et al. using the PyroMark kit (Qiagen).¹⁹ This assay amplifies a region of LINE-1 element (position 305 to 331, accession No. X58075), which includes four CpG sites (Figure 1A). Pyrosequencing reactions were performed using the PyroMark Q24 System (Qiagen). Bisulfite-pyrosequencing consists of three steps: bisulfite conversion, PCR amplification, and pyrosequencing analysis. Unmethylated cytosine (C) and methylated cytosine (^mC) are differentiated by bisulfite treatment followed by PCR. In the pyrosequencing step, the C:^mC ratio at each CpG site is measured as the ratio of T:C (where T represents thymine). The C content relative to the C plus T content at each CpG site is expressed as a percentage. We used the average relative C content at the four CpG sites as the overall LINE-1 methylation level in each given tumor sample.

FIGURE 1. Long interspersed nuclear element-1 (LINE-1) methylation level and copy number. (A) The LINE-1 structure and methylation site, together with copy number estimation sites. (B) LINE-1 methylation levels in the cancerous tissue were significantly lower than those in the normal tissue (left, total; middle, esophageal cancer; right, gastric cancer). (C) LINE-1 copy numbers in the cancerous tissue were significantly higher than those in the normal tissue (left, total; middle, esophageal cancer; right, gastric cancer). (D) Relationship between LINE-1 methylation level and copy numbers in upper GI cancers (left, total; middle, esophageal cancer; right, gastric cancer). The LINE-1 methylation level exhibited a significant degree of inverse correlation with copy number, with lower LINE-1 methylation levels corresponding to higher LINE-1 copy numbers.

LINE-1 copy number estimation

The LINE-1 copy number was estimated using 500 pg DNA, KOD SYBR® qPCR Mix (Toyobo), and the QuantStudio®5 real-time PCR system (Applied Biosystems) with minor modifications to the method described by Bundo et al.⁷ Primers were used after confirming that they can be used in SYBR-GREEN-based quantitative real-time PCR according to Coufal et al.⁸ Primers, probe location, and reaction chemistry are listed in Figure 1A and Table S3.

Real time PCR

Total RNA was extracted from frozen esophageal cancer specimens using a miRNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. The cDNA was reverse transcribed from total RNA using ReverTra Ace® qPCR RT Master Mix with gDNA Remover (Toyobo), and mRNA expression was quantified using KOD SYBR® qPCR Mix. The reaction was performed in an Applied Biosystems QuantStudio 12 K Flex (ThermoFischer Scientific) with the following PCR conditions: 98°C for 2 min, followed by 40 cycles of 98°C for 10 s, 60°C for 10 s, and 68°C for 30 s.

L1Hs sequence

To determine the genomic loci flanked by the 3′ prime end of the somatic LINE-1 insertions, we performed LINE-1 amplicon sequencing using a Novel Elements Concentrated-sequence (NECO-seq).²⁹ A total of 250 ng DNA was extracted from the cancer (n = 20) and non-cancer tissue (n = 20), and used to prepare the sequencing libraries. Sequencing was performed using a HiSeq X Reagent Kit v2.5 (Illumina) with 30% PhiX control. We obtained approximately 11,429,668 read numbers on average from each library. The candidate somatic LINE-1 insertions were firstly selected according to the bioinformatic pipeline in NECO-seq, and then further selected with the following criteria: (i) 10 or more depth, and at least one read containing 10 or more T in length within the soft clipped region in reads 1, or (ii) 10 or more depth, and an average conservation rate of the 3′ prime end of L1Hs sequence (5′-AGTATAATAA-3′) of 80% or more in reads 2.

In silico analysis

To identify the genes and genomic regions where new L1 insertions were detected, genomic features were annotated using the R package “annotatr”³⁰ and TxDb.Hsapiens.UCSC.hg38.knownGene as the reference dataset. Gene set enrichment analysis was performed by the Enrichr server (https://maayanlab.cloud/Enrichr/)³¹ and g:Profiler (https://biit.cs.ut.ee/gprofiler/gost) to identify the biological processes and collective pathway protein complexes of the included genes. We identified genes that have been confirmed to be involved in carcinogenesis based on the Cancer Gene Census (CGC) data: Census_allTue_Apr26_03_39_28_2022(https://cancer.sanger.ac.uk/census).

Statistical methods

We used the JMP software (version 9; SAS Institute) for statistical analyses. Continuous variables were expressed as means and standard deviations (SD), and means were compared using the paired t-test. Pearson's correlation analysis was used for linear correlation analysis. Statistical differences were considered significant at a P value threshold <0.05. All reported P values are two-tailed.

RESULTS LINE-1 methylation level and copy number inversely correlate in upper GI cancers

We first examined the LINE-1 methylation level in 204 cancer tissues and paired normal mucosa samples (esophageal cancer, n = 101; gastric cancer, n = 103) using a bisulfite-PCR-pyrosequencing assay. The cancer tissues exhibited significantly lower levels of LINE-1 methylation (median 63.3, mean 60.9, SD 15.7, all in 0–100 scale) than the matched normal mucosa samples (median 76.2, mean 75.7, SD 3.4) (P < 0.0001 by the paired t-test) as shown in Figure 1B. Similar results were obtained when esophageal and gastric cancer samples were analyzed separately (Figure 1B). Next, we measured the LINE-1 copy number using quantitative real-time PCR and observed that the copy number of LINE-1 was higher in cancer tissues (median 7.5, mean 7.5, SD 0.68) compared to the normal mucosa samples (median 7.4, mean 7.3, SD 0.70) (P = 0.0005) (Figure 1C). Interestingly, the LINE-1 copy number remained significantly higher in cancerous areas than non-cancerous controls, even when esophageal and gastric cancers were analyzed separately (Figure 1C). We also evaluated the expression of LINE-1 mRNA (ORF1, ORF2) and observed higher expression of LINE-1 mRNA in cancerous than in non-cancerous regions (Figure S1). Finally, based on our analysis, the LINE-1 methylation level exhibited a significant degree of inverse correlation with copy number (P = 0.0010, R² = 0.052), with lower LINE-1 methylation levels corresponding to higher LINE-1 copy numbers (Figure 1D). No such similarly significant relationship for LINE-1 methylation level and copy number was returned from our analysis of healthy tissue samples (Figure S2). Collectively, these findings indicate that LINE-1 hypomethylation may be linked to increased LINE-1 copy numbers in upper GI cancers.

The number of LINE-1 insertions is increased in the genomes of esophageal and gastric tumors

Although the LINE-1 copy number was found to be elevated in the tumor area, we analyzed tumor-specific LINE-1 insertion using the L1Hs-seq in 10 esophageal and 10 gastric cancers. We first performed L1Hs-seq on one sample of the tumor/non-tumor pair. The LINE-1 insertion site detected by L1Hs-seq was further corroborated by Sanger sequencing, which supported the accuracy of the L1Hs analysis pipeline. A total of 1012 unknown genomic insertions, as unknown non-reference germline insertions, were detected in 20 tumor and 20 paired non-tumor samples. This number excludes known references and known non-reference germline insertions, such as L1 base and previously reported L1 insertion sites.^32–36 The average number of somatic LINE-1 insertions detected by the L1HS-seq in 20 paired samples was 25.3 (range 4–149) (Figure 2A). We confirmed LINE-1 insertion by performing PCR using the 3′ prime end of LINE-1 and a specific primer for the new LINE-1 insertion site. The results of the validation of LINE-1 insertion sites on chromosomes 2, 11, and 17 are presented in Figure S3. The amplified PCR bands were additionally confirmed by Sanger sequencing. The number of somatic LINE-1 insertions was found to be higher in tumor (median 25.5, mean 37.9, SD 34.8) compared to normal tissues (median 10, mean 12.7, SD 8.1) (P = 0.0027; Figure 2B). When esophageal and gastric cancer samples were analyzed separately, similar results were observed (Figure 2B). It is important to note that the number of LINE-1 insertions was significantly higher in instances of LINE-1 hypomethylation (P = 0.018, R² = 0.27; Figure 2C). Such a relationship was not observed in normal tissues (P = 0.78, R² = 0.0043; Figure 2C). These results suggest that LINE-1 hypomethylation induces LINE-1 retrotransposition events with regards to esophageal and gastric malignant tumors.

FIGURE 2. The number of long interspersed nuclear element-1 (LINE-1) insertions in upper gastrointestinal (GI) cancers. (A) Number of LINE-1 insertions from 20 samples of patients with upper GI cancer. Deep blue indicates esophageal cancer and orange indicates gastric cancer. Each diagonal bar indicates the number of insertions in matched normal tissues. (B) The number of somatic LINE-1 insertions was higher in tumor compared to matched normal tissues (left, total; middle, esophageal cancer; right, gastric cancer). (C) Correlation between LINE-1 methylation level and number of LINE-1 insertions. The number of LINE-1 insertions was significantly higher in instances of LINE-1 hypomethylation in tumors (left). Such a relationship was not observed in normal tissues (right).

Chromosomal and genomic distribution of LINE-1 insertions

We next profiled the distribution and frequency of new somatic LINE-1 insertions across the genome. A list of genome positions where LINE-1 insertion was identified in 20 tumors is presented as Data S1. In addition, as an example, insertion into ARID1A is shown by using the Integrative Genomics Viewer (IGV) (https://igv.org) in Figure S4. According to a representative plot of chromosomes where somatic LINE-1 insertion was identified in 20 tumors, we observed that LINE-1 inserted randomly and evenly throughout autosomal and sex chromosomes (Figure 3A). Figure S5 depicts the mapping of L1 somatic insertion (coding region) and L1 somatic insertion (coding region, detected in two or more samples), indicating that LINE-1 is randomly and evenly inserted. Next, we focused on the genomic loci that contained L1 insertion sites. Our results revealed that LINE-1 insertions were predominantly distributed among intergenic (shown as black in Figure 3B) and intron (shown as gray in Figure 3B) regions, which accounted for 80%–85% of all LINE-1 insertions identified throughout the genome (Figure 3B). Taken together, our analysis of the genomic landscape in upper GI tumors indicates that associated LINE-1 insertions occur at a similar frequency across all chromosomes, but within distinct genomic loci that mainly consist of intergenic and intron regions.

FIGURE 3. Locations of long interspersed nuclear element-1 (LINE-1) insertions. (A) The plot of chromosomes where somatic LINE-1 insertion was found in 20 tumors. LINE-1 inserted randomly and evenly throughout autosomal and sex chromosomes. Deep blue indicates esophageal cancer and orange indicates gastric cancer. The bar on the upper right shows the gene density of the human genome. (B) Genomic locations of LINE-1 insertion sites. LINE-1 insertions were predominantly distributed among intergenic (shown as black) and intron (shown as gray) regions, which accounted for 80%–85% of all LINE-1 insertions identified throughout the genome.

Genes and associated pathways affected by LINE-1 insertions

Lastly, we characterized the specific genes affected by LINE-1. In 20 samples of upper GI cancers, LINE-1 insertions were detected in 580 genes. Interestingly, by cross-referencing our returned target gene list of somatic LINE-1 insertions with COSMIC's CGC library, which contains genes with known roles in carcinogenesis, we identified 17 genes shared between the two datasets (Table 1). This evidence supports our finding that LINE-1 insertion may contribute to the development and progression of upper GI cancers.

TABLE 1 Long interspersed nuclear element-1 somatic inserted genes in the Cancer Gene Census of COSMIC.

Gene symbol	Gene name	Tier	Role in cancer (COSMIC)
ARID1A	AT rich interactive domain 1A (SWI-like)	1	TSG, fusion
PBX1	Pre-B-cell leukemia transcription factor 1	1	Oncogene, fusion
MDM4	Mdm4 p53 binding protein homolog	1	Oncogene
RGS7	Regulator of G protein signaling 7	2
CTNNA2	Catenin alpha 2	2	Oncogene
LRP1B	LDL receptor related protein 1B	1	TSG
EPHA3	EPH receptor A3	2
ESR1	Estrogen receptor 1	1	Oncogene, TSG, fusion
EGFR	Epidermal growth factor receptor (erythroblastic leukemia viral (v-erb-b) oncogene homolog, avian)	1	Oncogene
GRM3	Glutamate metabotropic receptor 3	2	Oncogene
CSMD3	CUB and Sushi multiple domains 3	2	TSG
FAM135B	Family with sequence similarity 135 member B	2
POU2AF1	POU domain, class 2, associating factor 1 (OBF1)	1	Oncogene, fusion
YWHAE	Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, epsilon polypeptide (14–3-3 epsilon)	1	TSG, fusion
VAV1	Vav guanine nucleotide exchange factor 1	2	Fusion
SMARCA4	SWI/SNF related, matrix associated, Actin dependent regulator of chromatin, subfamily a, member 4	1	TSG
EP300	300 kd E1A-Binding protein gene	1	TSG, fusion

Note: Tier1, list of genes known to have a high risk of carcinogenesis; Tier2, statistically reliable oncogene or tumor suppressor gene.

Abbreviation: TSG, tumor suppressor gene.

An additional important observation from our analysis was that several genes contained multiple somatic LINE-1 insertions. Table 2 shows the genes in which LINE-1 insertions were observed in two or more regions. Similarly, the differences regarding the number of LINE-1 somatic insertions in the same genes between cancerous and normal tissues are listed in Table S4. Next, we performed gene set enrichment analysis for the genes where multiple LINE-1 insertions were detected. Table S5 shows the pathways related to these genes (Enricher) and Table 3 shows protein complexes related to these genes (g:Profiler). As indicated by our results, LINE-1 retrotransposons inserted into tumor suppressive genes, such as those involved in retinoic acid receptor-related orphan receptor-α (RORA) activates gene expression and p300-CBP-p270-SWI/SNF complex.

TABLE 2 Long interspersed nuclear element-1 somatic inserted gene list in multiple cases (except for PXDNL).

Ensembl gene IDs	Gene symbol	Number of insertions	Insertion sites	Role in cancer (COSMIC)
ENSG00000135299	ANKRD6	2	EK-0101 (intron), MK-0437 (intron)	–
ENSG00000137074	APTX	2	NK-0437 (intron), MK-28T (exon)	–
ENSG00000168487	BMP1	2	EK-0095 (1–50 kb), EK-0101 (intron)	–
ENSG00000184408	KCND2	2	MK-0431 (intron), EK-0243 (intron)	–
–	LINCR-0002	2	EK-0101 (intron), MK-28T (1–5 kb)	–
–	LOC105374320	2	MK-0431 (intron), EK-0255 (intron)	–
ENSG00000168702	LRP1B	2	EK-0243 (intron), MK-28T (intron)	TSG
ENSG00000158458	NRG2	2	MK-0437 (intron), EK-0093 (1–5 kb)	–
ENSG00000184792	OSBP2	2	EK-0085 (intron), MK-0473 (intron)	–
ENSG00000144724	PTPRG	2	EK-0243 (intron), EK-0101 (intron)	–
ENSG00000147485	PXDNL	2	EK-0243 (intron & promoter)	–

TABLE 3 Reactome and CORUM related with long interspersed nuclear element-1 somatic inserted genes in multiple cases: g:Profiler.

Source	Term name	Term id	Adjusted P value
REAC	RORA activates gene expression	REAC:R-HSA-1368082	0.001630502
REAC	Signaling by ERBB2 TMD/JMD mutants	REAC:R-HSA-9665686	0.041800296
CORUM	p300-CBP-p270-SWI/SNF complex	CORUM:570	0.026580578
CORUM	Kv4.2-DPP10 channel complex	CORUM:595	0.049851859
CORUM	ESR1-CEBPB complex	CORUM:6068	0.049851859
CORUM	ESR1-GRIP1 complex	CORUM:6069	0.049851859
CORUM	ESR1-MDM4 complex	CORUM:7250	0.049851859

Note: CORUM: the comprehensive resource of mammalian protein complexes.

DISCUSSION

To the best of our knowledge, this is the first study to demonstrate the relationship between LINE-1 hypomethylation, increased retrotransposition, and tumor-specific insertion using clinical specimens from more than 200 patients with upper GI cancers. We have previously reported that LINE-1 hypomethylation is associated with poor prognosis in patients with upper GI cancers.^27,28 In this study, we explored this relationship further and observed that LINE-1 copy number was higher when distinct global hypomethylation of LINE-1 sites was detected in tissue samples of patients. This was also mirrored by a concomitant increase in somatic LINE-1 insertions across the genome of such patients. The adjusted L1Hs-seq also identified somatic LINE-1 insertions into tumor-suppressive genes, such as those involved in the RORA activates gene expression and p300-CBP-p270-SWI/SNF complex pathways. These L1 insertions may at least partially contribute to the inactivation of those genes and subsequently facilitate the acquisition of aggressive tumor characteristics.

The LINE-1 family represents a major group of retrotransposons, with L1 repetitive elements accounting for about 17% of the human genome,³ therefore the level of LINE-1 methylation is often used as a readout of global DNA methylation.¹⁸ Tumor-related LINE-1 hypomethylation has been associated with poor prognosis not only in upper GI cancers but also in a range of different human tumor types, including glioma, colon, ovarian, and lung cancers.^22–28 As such, here we aimed to elucidate the mechanism by which tumoral LINE-1 hypomethylation correlated with malignant phenotype or patient prognosis in human cancers. In earlier research, we reported that LINE-1-hypomethylated tumors showed increasingly frequent genomic gains at various loci containing candidate oncogenes such as CDK6.³⁷ Given that LINE-1 is the only autonomous retrotransposon that encodes the proteins necessary for mobilization and reinsertion into the genome, we speculated that LINE-1 hypomethylation could lead to increased LINE-1 mobilization, which may in turn promote acquisition of malignant behavior. Pan-cancer analysis of whole genomes identified driver region rearrangements mediated by LINE-1 retrotransposition, with potential implications for the development of human tumors.³⁸ However, similar studies in upper GI cancers remain scarce. Doucet-O'Hare et al. have shown that somatic retrotranspositions occur early in many patients with Barrett's esophagus and esophageal adenocarcinoma, and suggest that early retrotransposition events even in histologically normal esophageal cells may be clonally expanded and ultimately promote esophageal adenocarcinogenesis.³⁹ Yamaguchi et al. found that somatic LINE-1 retrotransposition showed significant interpatient and intercellular variability in patient cohorts with the same type of upper GI cancer and among different cells within a particular cancer type, respectively.⁴⁰ The relationship between immune signatures and LINE-1 retrotransposition in upper GI cancers has also been examined.⁴¹ In the present study, we underscored the relationship between LINE-1 hypomethylation and retrotransposition frequency, using tissue samples from a larger cohort of patients with upper GI cancer than previous studies, and observed that LINE-1 insertions also occurred in multiple cancer-related genes. Epigenetic changes, such as LINE-1 hypomethylation, can be acquired and have attracted considerable attention as promising targets for disease therapy. In this context, our results may be of clinical significance because we highlight that LINE-1 hypomethylation inversely correlates with an increase in LINE-1 retrotransposition events and the subsequent emergence of malignant tumor characteristics.

Methylation of the LINE-1 promoter region typically inhibits its transcription. However, reduced methylation can enhance LINE-1 transcription, potentially leading to increased LINE-1 RNA production and more insertions in the cancer genome.¹³ In this study, we demonstrated a connection between LINE-1 hypomethylation and high copy number amplification, which may support this phenomenon. Nevertheless, it is worth noting that LINE-1 hypomethylation may not always be a prerequisite for the amplification of LINE-1 promoter loci in cancer progression. Cancer cells may persist by favoring specific clones with concurrent LINE-1 amplification and hypomethylation. Further research, including in vitro experiments, is needed to delve into these aspects.

In addition, we observed that LINE-1 was frequently inserted within regions corresponding to regulatory sequences of tumor-suppressive genes, such as those involved in the RORA activates gene expression and p300-CBP-p270-SWI/SNF complex pathways. The peptide product of RORA functions as a transcriptional activator in response to circadian changes. Of particular note, disruption of circadian regulators, such as RORA, reportedly plays a critical role in tumorigenesis.⁴² A genome-wide association study conducted in a Han Chinese subgroup from East China revealed that RORA is negatively associated with esophageal squamous cell carcinoma (ESCC).⁴³ Regarding the p300-CBP-p270-SWI/SNF complex, Zhu et al. have demonstrated the tumor suppressive function of the P300/CBP-associated factor at the frequently deleted 3p24 region in ESCC.⁴⁴ Intronic L1 insertions in human tumors often result in decreased expression of the mutated genes and as such they could theoretically contribute to tumorigenesis by reducing or abolishing the expression of tumor-suppressive genes.¹¹ Taken together, our study showed that LINE-1 hypomethylation in upper GI cancers may at least partially contribute to the acquisition of aggressive tumor features through tumor-specific insertion in tumor-suppressive genes.

Our study has some limitations. First, while we successfully established a connection between LINE-1 hypomethylation, amplification, and tumor-specific insertions using clinical samples, we have yet to confirm whether LINE-1 insertions indeed impact the expression of tumor-suppressive genes and contribute to the acquisition of tumor malignancy. Further investigations are warranted in this regard. Second, the representation of esophageal adenocarcinoma in our study was limited, leaving us uncertain about the generalizability of these findings to esophageal squamous cell carcinoma and adenocarcinoma. Moreover, we have not ascertained if similar observations can be made in esophagogastric junction cancer. Validation in a larger cohort is imperative.

In summary, in this study we observed that LINE-1 hypomethylated tumors presented with increased LINE-1 copy numbers and frequent retrotransposition events. In addition, we found that LINE-1 is often inserted into tumor-suppressive genes such as those involved in the RORA activates gene expression and p300-CBP-p270-SWI/SNF complex pathways. Collectively, these findings suggest that LINE-1 hypomethylation in upper GI cancers may contribute to the acquisition of aggressive tumor features through the L1 insertion-dependent inactivation of tumor-suppressive genes. Further studies would be required to examine our findings in greater detail and explore additional mechanisms by which LINE-1 hypomethylation may affect tumor behavior.

AUTHOR CONTRIBUTIONS

Yoshifumi Baba: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; project administration; writing – original draft; writing – review and editing. Noriko Yasuda: Data curation; formal analysis; investigation; methodology; writing – original draft; writing – review and editing. Miki Bundo: Data curation; formal analysis; investigation; methodology; writing – original draft; writing – review and editing. Yutaka Nakachi: Data curation; formal analysis; investigation; writing – review and editing. Junko Ueda: Investigation. Takatsugu Ishimoto: Data curation; formal analysis; methodology. Masaaki Iwatsuki: Data curation; formal analysis. Yuji Miyamoto: Data curation; formal analysis. Naoya Yoshida: Investigation. Hiroyuki Oshiumi: Conceptualization; project administration; supervision. Kazuya Iwamoto: Conceptualization; data curation; formal analysis; project administration; supervision; writing – review and editing. Hideo Baba: Conceptualization; data curation; formal analysis; methodology; project administration; supervision.

ACKNOWLEDGMENTS

We thank Editage (www.editage.com) for English language editing.

FUNDING INFORMATION

This work was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (JSPS) (grant numbers 17KK0195 and 17H04273).

CONFLICT OF INTEREST STATEMENT

The authors declare no competing financial interests.

ETHICS STATEMENT AND CONSENT TO PARTICIPATE

Approval of the research protocol by an Institutional Reviewer Board: This study was approved by the institutional review board of Kumamoto University (#565).

Informed Consent: Written informed consent was obtained from all patients.

Registry and the Registration No. of the study/trial: N/A.

Animal Studies: N/A.