Transposon insertion profiling by sequencing (TIPseq) identifies novel LINE-1 insertions in human sperm

Abstract

Purpose

Long interspersed nuclear element-1 (LINE-1 or L1) comprises 17% of the human genome. Retrotransposons may perturb gene integrity or alter gene expression by altering regulatory regions in the genome. The germline employs a number of mechanisms, including cytosine methylation, to repress retrotransposon transcription throughout most of life. Demethylation during germ cell and early embryo development de-represses retrotransposons. Intriguingly, de novo genetic variation appearing in sperm has been implicated in a number of disorders in offspring, including autism spectrum disorder, schizophrenia, and bipolar disorder. We hypothesize that human sperm exhibit de novo retrotransposition and employ a new sequencing method, single cell transposon insertion profiling by sequencing (scTIPseq) to map them in small amounts of human sperm.

Methods

Cross-sectional case–control study of sperm samples (n=10 men; ages 32–55 years old) from consenting men undergoing IVF at NYU Langone Fertility Center. scTIPseq identified novel LINE-1 insertions in individual sperm and TIPseqHunter, a custom bioinformatics pipeline, compared the architecture of sperm LINE-1 to known LINE-1 insertions from the European database of Human specific LINE-1 (L1Hs) retrotransposon insertions (euL1db).

Results

scTIPseq identified 17 novel insertions in sperm. New insertions were mainly intergenic or intronic. Only one sample did not exhibit new insertions. The location or number of novel insertions did not differ by paternal age.

Conclusion

This study for the first time reports novel LINE-1 insertions in human sperm, demonstrating the feasibility of scTIPseq, and identifies new contributors to genetic diversity in the human germ line.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-023-02852-6.

Introduction

Transposable elements (TEs) constitute a significant part of the human genome. TEs are subdivided into two main classes: retrotransposons (class I) that use the “copy and paste” process for their replication and expansion and DNA transposons (class II) that use the “cut and paste” mechanism [1]. Retrotransposons are classified into two categories: LTR (long terminal repeat), including endogenous retroviruses (ERVs), and non-LTR transposons, including LINEs (long interspersed elements), SINEs (short interspersed elements) and in humans the SINE-VNTR-Alu (SVA) [1]. Retrotransposons transcribe an intermediate RNA, which is then transcribed back into DNA and inserted into new locations within the genome, a process called retrotransposition [1]. In the mammalian genome, most retrotransposons have become incompetent for transposition, but some retrotransposons, especially LINEs, can become active [2]. Moreover, repressive epigenetic marks inhibit LINE-1 retrotransposition.

Cytosine demethylation removes most of the repressive epigenetic marks [3–5] during gamete formation and early embryo development. This epigenetic reprograming, geared towards more transcriptionally active chromatin, also promotes LINE-1 retrotransposition [6–8]. Retrotransposon activation may influence genomic stability by disrupting coding regions, or more commonly, regulatory regions in the host genome [8–13].

Male gametes develop more copy number variants (CNVs) compared to female gametes [14, 15] and paternally derived translocations arise more commonly in the offspring of older men [16, 17]. More generally, age alters chromatin structure and post transcriptional gene regulation pathways in many cell types [18–20]. With the age of parenthood increasing in most societies [21], understanding the effects of advancing paternal-age on the genomes’ offspring becomes an urgent priority. De novo genetic variation [22–24] [25, 26] contributes to a number of neuropsychiatric syndromes, including autism, schizophrenia, and bipolar affective disorder [27, 28]. Altered DNA methylation has been found in both somatic and germ cells [29] and may mediate some of the paternal age effects [30]. Similarly, LINE-1 becomes more transcriptionally active during aging and senescence [31, 32].

The precise location of de novo LINE-1 insertions is vital to understand their potential role as a source of structural variation and mutation. Because LINE-1 is a highly repetitive sequence found in high copy number throughout the human genome, identification of de novo LINE-1 insertions has been technically challenging. Transposon insertion profiling by next-generation SEQequencing (TIP-seq), and a machine-learning computational pipeline, TIPseqHunter, a next-generation sequencing-based method, were designed to overcome these challenges [33]. Recently this technique was adapted to profile LINE-1 insertions with limited amounts of DNA available for analysis, including individual cells [34]. The present study applied scTIPseq, and its associated bioinformatics pipeline, TIPseqHunter, to identify and map novel LINE-1 insertions in human sperm.

Material and methods

Study population and sample preparation

Cross-sectional case–control study including sperm samples (n=10; ages 32–55 years old) obtained from consenting ten men undergoing in vitro fertilization at NYU Langone Fertility Center, between February to July 2020. Sperm parameters were evaluated (Table 1), and samples washed for 10 min and re-suspended in global G1 medium. Sperm genomic DNA was extracted using DNeasy Blood and Tissue Kit (QIAGEN) following manufacturer’s recommended protocol, with an additional lysis step using X2 buffer (20 mM Tris-Cl, pH 8.0, 20 mM EDTA, 200 mM NaCl, 4% SDS); 80 mM dithiothreitol (DTT) and 12.5 μL/mL proteinase K [35]. DNA concentration and purity were measured using Qubit Fluorometer (ThermoFisher Scientific, cat. no. Q33226) (Table 1). Samples were stored at −80°C until TIPseq method.

Table 1.

Descriptive data of each semen parameters and DNA concentration pre and post multiple displacement amplification (MDA)

Sample ID	Age (years)	Concentration (mL)	Progressive motility (%)	DNA concentration (ng/ul) pre MDA	DNA concentration (ng/ul) post MDA
20163	55	48	51	20.4	86.0
13613	45	211	83	15.6	88.0
20196	32	140	57	20.0	78.4
20171	34	76	55	29.0	92.6
20120	32	78	79	31.6	79.0
20132	33	160	62	31.0	76.4
20145	49	53	44	25.8	83.6
20146	32	79	58	28.2	82.6
20173	55	90	44	33.4	90.2
20199	49	42	31	33.2	98.8

ID Identification; % percentage; concentration (×10⁶/mL)

Whole genome amplification and quality control

Single-cell TIPseq was performed according to a recently published protocol [36], but adapted for use with low input DNA [34] as we were unable to get large amount of sperm DNA even using bulk sperm sample (Table 1). Briefly, samples were diluted to 10 ng in 2.5 ul and submitted to a preliminary step — multiple displacement amplification (MDA) using REPLI-g single cell kit (QIAGEN, cat no. 150343), to produce larger quantities of high quality DNA necessary for the TIPseq assay (Table 1). MDA provide a high level of coverage across the genome [37]. Placenta genomic DNA was used as positive control and double distilled water used as negative control. MDA products were purified following the protocol of the purification kit Agencourt AMPure XP Magnetic Beads (Beckman Coulter, cat. no. A63882 and DynaMag-TM Side Magnetic -ThermoFisher Scientific, cat.no. 12331D) and eluted in 50 μL of TE buffer. A qPCR was performed as a quality control (QC) to evaluate the efficiency of the MDA. The purified products were aliquoted into each well of a 96-well plate to perform the quantitative real-time PCR. Each reaction included 10 μL 2× SYBR Green mix (Bio-Rad), 2 μL each of 4 μM forward and reverse primers, 1 μL molecular grade water, and 5 μl diluted (1:300) MDA product, to yield a 20μL reaction. For this, twelve pairs of primers were designed for qPCR to amplify regions downstream of the 3’ end of fixed L1s insertions on different chromosomes (Supplementary table S1). Primers were synthesized by Integrated DNA Technologies (IDT). A Bio-Rad thermocycler (CFX96) was used to run program “CFX_3stepAMP” with anneal temperature set as 60°C. If all 12 regions are amplified, the sample is suitable to continue to the next step.

Vectorette PCR

An initial amount of 10 μg of the whole-genome amplified DNA was divided into aliquots that were independently digested with a panel of restriction enzymes: AseI (cat. no. R0526S), BspHI (cat. no. R0517S), BstYI (no. R0523S), HindIII (cat. no. R0104S), NcoI (cat. no. R0193S), and PstI (cat. no. R0140S) (NewEngland Biolabs). A pair of vectorette oligonucleotides (synthesized by IDT) corresponding to each restriction enzyme or T tail was annealed to form vectorette adaptors with a sticky end created (Supplementary table S2) [36]. Then the digested amplified genomic DNA were ligated with the vectorette adaptors using T4DNA ligase (NewEngland Biolabs, M0202S) at 4°C overnight. After ligation, PCR was performed with the L1 primer (5’- AGATATACC TAATGC TAG ATG ACACA -3’) and the vectorette primer (5’- CTC TCC CTT CTC GGATCT TAA -3’), using Hot-Start Taq DNA polymerase (Takara, cat no. RR006A) with a touchdown program (95°C 5 min; 5 cycles: 95°C 1 min, 72°C 1 min, 72°C 5 min, 5 cycles: 95°C 1 min, 68°C 1 min, 72°C 5 min, 5 cycles: 95°C 45 s, 64°C 1 min, 72°C 5 min, 15 cycles: 95°C 45 s, 60°C 1 min, 72°C 5 min, 72°C 15 min; 16°C hold).

DNA purification and quality control

PCR products were purified using the purification kit (Agencourt AMPure XP beads and DynaMag^TM-96 Side Magnet) and were eluted in 21 μL of TE buffer. DNA concentration was measured using Qubit. An amount of 2 ug of purified DNA on 1.5% agarose — 1× TAE gel stained with SYBR Green was run at constant 100 V for 1 h or until separation of the ladder was clearly visible. Vectorette PCR amplicons should appear as a smear on the gel averaging around 1–3kb.

Next-generation sequencing

About 2 μg of the vectorette PCR products were sheared to fragments of around 300 bp. Sequencing libraries were prepared using the KAPA HTP DNA Library preparation Kit (Roche, cat no. KK8234). Libraries were sequenced on an Illumina HiSeq 4000 with paired-end 150 bp reads. Sequencing was performed in NYU Langone’s Department of Systems Genetics [34].

Data analysis using TIPseqHunter

The bioinformatics pipeline, TIPseqHunter [33], was performed by the Bioinformatics Core from the NYU Genomics Center, with oversight by Professor David Fenyo, the computational biologist who developed TIPseqHunter. TIPseqHunter is comprised of five steps: (i) sequence read preparation and quality control, (ii) sequence alignment, (iii) candidate LINE-1 insertion site identification, (iv) LINE-1 insertion site modeling, and (v) LINE-1 insertion site prediction. Because the efficacy of the PCR assay and the type of background associated with each TIPseq dataset can vary, the machine-learning algorithm is run on each sample. The reliability of single cell TIPseq results from sperm samples was calculated using kappa coefficient to compare the total number of LINE-1 insertions and insertions into developmentally expressed genes. These parameters were compared between sperm samples from young and older men, and also with reference parameters from European database of human specific LINE-1 (L1Hs) retrotransposon insertions in humans (euL1db).

Validation of unknown insertions

Validation of the unknown insertions was performed using the sperm samples after MDA, by 3’ junction PCR [34]. The 3’ junction PCR is designed to amplify the 3’ of LINE-1 insertion and the downstream flanking sequence using LINE-1 primer in the 3’ of LINE-1 and the other primer in unique flanking sequence. The primers were designed using an in silico PCR (https://primer3.ut.ee/). The 20ul PCR reaction was set up with 50ng of DNA, 2ul of 10× ExTaq buffer, 1.6ul of dNTP mix, 0.1ul of Ex Taq HS, 0.4ul of L1 primer (10uM), and 4ul of left or right primer (1uM), and amplified on the thermocycler with setting as 1 cycle of 94°C for 5 min; 30 cycles of 94°C for 30 s, 60°C for 30 s, 68°C for 30 s; 1 cycle of 68°C for 7 min. PCR product then were loaded on 0.5% agarose gel for electrophesis in 1X TAE buffer to identify amplifications.

Statistical analysis

The descriptive values of the samples were calculated through measurements of central position and dispersion. The median was used as central position measures and the interquartile range (25–75), and the minimum and maximum values of the samples included in study were used as dispersion measures.

Analyzes were implemented in the SAS version 9.4 program.

Results

Baseline parameters of the samples included in this study

Sample characteristics are shown in Table 2 as median (interquartile range — IQR). The median age of the 10 samples included in this study was 39.5 (32–55) years, the concentration of spermatozoa per ml was 78.5 million (42-211), and the percentage of progressive motility was 56 (31–83). The median age of the 5 samples < 35 years was 32 (32–34) years old, the concentration of spermatozoa per ml was 79 (76–160), and the percentage of progressive motility was 58 (55–79). The median age of the 5 samples ≥ 45 years was 49 (45–55) years old, the concentration of spermatozoa per ml was 53 (42–211), and the percentage of progressive motility was 44 (31–83).

Table 2.

Baseline parameters of the samples included in this study

Samples	N	Age (years)	Concentration per ml	Progressive motility (%)
All samples	10	39.5 (32–49)	78.5 (53–140)	56 (44–62)
≤ 35 years	5	32 (32–33)	79 (78–140)	58 (57–62)
≥ 45 years	5	49.0 (49–55)	53 (48–90)	44 (44–51)

Data presented as median (interquartile range — IQR); N number, concentration (x10⁶/mL); % percentage

LINE-1 insertions in sperm samples

Total counts of LINE-1 insertions are shown in Fig. 1 and Table 3 as median (interquartile range — IQR). The median of all insertions observed in the 10 samples included in this study was [n: total number; median (minimum–maximum)]: n: 4831; 480.5 (415–565), the median of the known insertions n: 4606; 462 (400–538), the median of the unknown insertions n: 225; 24.5 (12–37) and, the median of the unique insertions n: 17; 1.00 (0–4). The median of all insertions found among the 5 samples age < 35 years included in this study was 476 (442–505), the median of the known insertions 457 (429–480), the median of the unknown insertions 24 (12–26), and the median of the unique insertions 1 (1–3). The median of all insertions found among the 5 samples age ≥ 45 years included in this study was 492 (415–565), the median of the known insertions 462 (400–538), the median of the unknown insertions 27 (15–37) and, the median of the unique insertions 1 (0–4).

Fig. 1.

Total count of LINE-1 insertions in sperm samples

Table 3.

LINE-1 insertions found on the samples included in this study

Samples	N	Age (years)	All insertions	Known insertions	Unknown insertions	Unique insertions
All samples	10	41.5 (32–55)	480.5 (453–505)	462 (30–480)	24.5 (15–27)	1.5 (1–3)
≤ 35 years	5	32 (32–33)	476 (453–483)	457 (430–463)	24 (13–25)	1 (1–3)
≥ 45 years	5	55 (49–55)	492 (478–522)	462 (462–485)	27 (16–30)	2 (1–4)

Data presented as median (interquartile range — IQR); N number

Novel LINE-1 insertions

TIPseq identified 17 novel insertions in sperm, 8 from older (≥ 45 years) and 9 from younger men (< 35 years) (Fig. 1). New insertions were mainly intergenic or intronic, including TMEM163 (2/7), CTTNBP2NL (3/5), TMC2 (2/19), MACROD2 (2/6), RAB3C (3/4), LINC02664 (1/1), AC007402 (2/10), AC107023 (3/3), AC079052 (2/3), and AC017091 (4/4) (Table 4). One novel insertion (< 35 years old) hits a known regulatory element. Only one sample (≥ 45 years old) did not exhibit any new insertion (Fig. 1, Table 4). The number of novel insertions did not differ by paternal age.

Table 4.

Descriptive data of the new and unique insertions found in sperm samples

Sample ID	Age	Chromosome	Gene	Repeat	Regulatory element
13613	45	chr20
20120	32	chr2	AC007402 intron (2/10)	L1MB5
20120	32	chr2	TMEM163 intron (2/7)	AluSp
20120	32	chr13
20132	33	chr1	CTTNBP2NL intron (3 of 5)	AluJb
20132	33	chr2
20132	33	chr12	AC107023 intron (3/3)
20145	49	chr20	TMC2 Intron (2/19)
20145	49	chr20	MacroD2 intron (2/6)	L1PB1	CEBPB TF binding site
20146	32	chr4
20171	34	chr5	RAB3C intron (3/4)	L1PA8A
20173	55	chr6		L2
20196	32	chr11			Enhancer
20199	49	chr1	Intergenic	L1PA13
20199	49	chr4	LINC02664 intron (1/1)	L1ME3
20199	49	chr10	AC079052 intron (2/3)
20199	49	chr18	AC017091 intron (4/4)
20163	55		No new insertion

Validation of unique insertions using PCR

We performed 3’ junction PCR validations on 5 of these 17 unknown insertions (primers sequences are shown in table S3, Fig. 2, 3, 4). Specific PCR amplicons aligned to the appropriate location in the genome. This is consistent with the findings of McKerrow et al. (2019) that TIPseq and TIPseqHunter analysis provide infrequent false-positive calls after filtering, showing near-complete profiles of LINE-1 insertion sites.

Fig. 2.

Novel and unique LINE-1 in sperm samples

Fig. 3.

Comparison of LINE-1 insertions and men’s age

Fig. 4.

Validation of unique LINE-1 insertions using PCR

Discussion

Transposable elements shaped the human genome in the past and presumably continue to do so. The retrotransposon activity of LINE-1 is repressed during most stages of life, but epigenetic changes during gametogenesis and early embryo development can unleash activity to create novel insertions that provide structural variation to our genome [36]. The highly repetitive nature of these sequences make them especially challenging to locate within the genome [33], but here we apply a sensitive strategy to amplify and sequence LINE-1 retrotransposon insertions in human sperm using single cell transposon insertion profiling by next-generation sequencing (scTIPseq). Our study is the first to report novel LINE-1 insertions in human sperm, demonstrating that TIPseq is a feasible technique to identify and map LINE-1 insertions in human gametes.

For the first time, we report total counts of novel LINE-1 insertions in sperm samples. That specific PCR amplicons aligned to the proper genomic location of all five novels insertions validates our findings.

Our experiments did not identify increased novel LINE-1s in sperm from older men. The limited sample size and age ranges likely would be inadequate to identify age effects on novel LINE-1 insertions. Moreover, not all sperm within a sample may exhibit novel insertions. A small fraction of sperm within sample-inheriting novel insertions would not have been detected by the method we employed. Additional studies using larger samples sizes are needed to map insertions in a wider range of men.

In human and mouse sperm, EGFP expression indicated that retroelements can be internalized in sperm cells, independently of the sperm origin and penetrate the tightly packed chromatin [38–40]. This permeability is a distinctive feature of both, epididymal and ejaculated spermatozoa (after removing the seminal fluid), from all animal species including humans [41, 42]. Mouse spermatozoa not only can integrate exogenous retroelements, but they also transfer them into the oocyte during fertilization, affecting preimplantation embryo development [40]. To proper compare if the insertions occur during spermatogenesis or in mature epididymal spermatozoa, it would be necessary to compare sperm samples extracted from the testis and epididymis, which would be extremely challenging.

The occurrence of the integration suggests that the sperm chromosomal DNA is not uniformly and tightly packed with protamine, implying the existence of genomic sites where the chromosomal DNA is accessible to foreign molecules [43]. The integration of endogenous retrotransposons and of foreign DNA molecules is proposed to occur in the nucleohistone chromatin spacer within two adjacent protamine domains [43]. The nucleosomal sperm chromatin fraction interacts with the nuclear scaffold and is supplied with the enzymatic machinery required to support DNA integrations [43] that could interfere with the genome, possibly influence the sequence of certain loci, and give rise to new retrotransposition events [38–40].

According to NCBI database, some of the genes where these new and unique insertions were found have uncharacterized functions [44]. The transmembrane protein 163 gene (TMEM163) has been predicted to enable zinc ion–binding activity, to be involved in zinc ion import into synaptic vesicle, to be located in early endosome membrane, to be active in intracellular vesicle and plasma membrane, and to be integral component of synaptic vesicle membrane [44]. This gene also is associated with hypomyelination leukodystrophy genetic disorder [45]. The CTTNBP2 N-terminal like gene (CTTNBP2NL) located in actin cytoskeleton enables protein phosphatase 2A binding activity. Acts upstream of or within negative regulation of transmembrane transport [44]. Overexpression of CTTNBP2NL inhibited embryo implantation in mice whereas knockdown of CTTNBP2NL decreased phosphatase activity [46]. Regard the transmembrane channel like 2 gene (TMC2), mutations on this gene may underlie hereditary disorders of balance and hearing [44]. The protein encoded by mono-ADP ribosylhydrolase 2 gene (MACROD2) is a deacetylase involved in removing ADP-ribose from mono-ADP-ribosylated proteins. The encoded protein has been shown to translocate from the nucleus to the cytoplasm upon DNA damage [44]. Deletion in MACROD2 is suggested to affect correct embryonic development in a family with complex phenotypes such as microcephaly, intellectual disability, polydactyly, renal, and pancreatic malformations [47]. RAB3C, a member of the RAS oncogene family, encodes a small GTPase. Other similar small GTPases are known to be involved in vesicle trafficking, and the encoded protein was shown to play a role in recycling phagocytosed MHC class 1 complexes to the cell surface [44]. RAB3C gene was also suggested as a potential marker for post-thaw boars sperm quality for cryopreservation [48].

These associations are intriguing, but much remains to be understood regarding the novel LINE-1 insertions found by scTIPseq. We believe that new powerful analytical methods, such as scTIPseq, will help clarify the role of retrotransposition events in human gametes, and possibly to further elucidate the role of retrotransposons in sperm function and male infertility.

Funding

Coordination for the Improvement of Higher Education Personnel (CAPES, Finance Code: 001, Brazil; process no. 88887.371487/2019-00 and process n. 88887.597054/2021-00 to TSB), Brazilian National Council for Scientific and Technological Development (CNPq, Brazil; Grant number 204747/2018-0 to FBK) and the Stanley H. Kaplan Fund of the NYU Grossman School of Medicine (to DLK).

Data availability

The data underlying this study will be shared on reasonable request to the corresponding author.

Declarations

Ethics approval

The New York University Langone Medical Center Institutional Review Board (IRB) approved this study under the IRB number S#16-00154.

Consent for publication

The author agree to its submission to the Journal of Assisted Reproduction and Genetics and, if accepted, to its publication in this journal. We warrant that this article is original, does not infringe on any copyright or other proprietary right of any third party, is not under consideration by another journal and has not been previously published. Ethical approval has been sought and obtained as necessary and any conflicts of interest stated.

Consent to participate

Each participant provided written informed consent.

Competing interests

The authors declare no competing interests.