Novel retrotransposon analysis reveals multiple mobility pathways dictated by hosts

Abstract

Autonomous non-long-terminal-repeat retrotransposons (NLRs) proliferate by retrotransposition via coordinated reactions of target DNA cleavage and reverse transcription by a mechanism called target-primed reverse transcription (TPRT). Whereas this mechanism guarantees the covalent attachment of the NLR and its target site at the 3' junction, mechanisms for the joining at the 5' junction have been conjectural. To better understand the retrotransposition pathways, we analyzed target-NLR junctions of zebrafish NLRs with a new method of identifying genomic copies that reside within other transposons, termed "target analysis of nested transposons" (TANT). Application of the TANT method revealed various features of the zebrafish NLR integrants; for example, half of the integrants carry extra nucleotides at the 5' junction, which is in stark contrast to the major human NLR, LINE-1. Interestingly, in a cell culture assay, retrotransposition of the zebrafish NLR in heterologous human cells did not bear extra 5' nucleotides, indicating that the choice of the 5' joining pathway is affected by the host. Our results suggest that several pathways exist for NLR retrotransposition and argue in favor of host protein involvement. With genomic sequence information accumulating exponentially, our data demonstrate the general applicability of the TANT method for the analysis of a wide variety of retrotransposons.

Figure 1.

Schematic representation of the NLRs analyzed in this study. ORF1 and ORF2 are shown in gray and black, respectively. The terminal repeats are shown on the right. The endonuclease, reverse transcriptase, and esterase domains are shown as EN, RT, and ES, respectively. The percentages indicate amino acid identities between any two proteins.

Figure 2.

Analysis of L1 integrants within other transposons. (A) An example of a genomic L1 integrant. Only junction regions are shown. The consensus sequence of the host transposon, MER11A (magenta), the sequence of a part of human chromosome 4 (black), and the L1PA4 consensus sequence (green) are aligned. Asterisks indicate identical sequences and the magenta and green boxes indicate MER11A and L1PA4 regions, respectively, inferred from the alignments. The target-site duplication (TSD) is shown in red letters. (B) Length distribution of TSD. Both full-length and 5′-truncated copies are included. (C) Length distribution of the 5′ microhomology (MH). Both full-length and 5′-truncated copies are included. (D) Length distribution of the 3′ MH. Both full-length and 5′-truncated copies are included.

Figure 3.

Analysis of integrants of CR1_DRs. (A,B) Two examples of CR1–2_DR integrants. The consensus sequences of the host transposons (magenta, TDR2 in A and SINE3–1 in B), the sequences of parts of the zebrafish genome (black), and the CR1–2_DR consensus sequence (green) are aligned. Asterisks indicate identical sequences, and the magenta and green boxes indicate the regions of CR1–2_DR and host transposons inferred from the alignments. The TSD is shown in red letters. The two integrants have TSDs and 3′ MHs. The example in A has 5′ MH, whereas the example in B has an insertion of extra 5′ nucleotides. (C,D) Length distributions of TSD (C) and TST (D) of CR1_DRs. The colors that discriminate each CR1_DR are indicated in the key in each panel, with the total numbers of CR1_DRs analyzed in parentheses. (E,G) Length distributions of MH at the 3′ (E) and 5′ (G) junctions. The probability distribution of the MH length for two random sequences was calculated as described previously (Roth et al. 1985) and is shown in black. The P values for expected and observed numbers were ≪0.01 in all χ² tests. (F,H) Length distributions of extra nucleotides at the 3′ (F) and 5′ (H) junctions.

Figure 4.

Two-dimensional matrix analysis for the interrelatedness of the junction features. (A) 5′ junction vs. 3′ junction. Integrants are categorized into those having MH, direct joining (direct), or extra nucleotides (extra) at the 5′ junction. In each category, integrants are further categorized into those having MH (blue), direct joining (light blue), and extra nucleotides (red) at the 3′ junction. The number of copies collected is indicated inside each rectangle and the P values by χ² tests of independence are shown at the right. (B) Target-site alterations vs. 3′ junction. Integrants are categorized into those with TSTs and the others (TSD+blunt), and further categorized by their 3′ features. The number of copies collected is indicated inside each rectangle. (C) Target-site alterations vs. 5′ junction. Integrants are categorized into those with long TSTs or others (short TSTs, TSDs, or blunt insertion), and further categorized by their 5′ features. The number of copies collected is indicated inside. Integrants with MH and those joined directly were combined for the χ² analysis to enhance the power of validation.

Figure 5.

Retrotransposition assay in HeLa cells. (A) Construction of pZfL2–2/mneoI/ColE1. A full-length CR1–2_DR was placed under the control of the CMV promoter (P_CMV) in the pCEP4 vector carrying the hygromycin-resistance gene (Hyg). The mneoI marker and ColE1 origin were inserted in the 3′ UTR of CR1–2_DR. The mneoI marker is a neomycin-resistant gene (Neo) interrupted by an insertion of an intron in the antisense orientation. This marker itself is in an antisense orientation relative to the NLR transcript. Thus, the vector does not confer the Neo phenotype, whereas CR1–2_DR retrotransposition, which includes splicing of the transcript, reverse transcription of that spliced RNA, and insertion of the synthesized cDNA into the genomic DNA, restores an intact Neo gene, converting the host cell to Neo. (B) Statistics for various NLR integrants. Integrants of CR1–2_DR in the zebrafish genome (top), those in HeLa cells using pZfL2–2/mneoI/ColE1 (second), L1 copies in transposons in the human genome (third), and de novo L1 insertions in human cultured cells (bottom) were categorized with regard to 5′ junctions. The frequency of extra 5′ nucleotides in de novo L1 insertions were reported previously (Symer et al. 2002; Gilbert et al. 2005). Because MH and direct joining were not distinguished in these reports, these integrants are represented as MH. The number of copies collected is indicated inside each rectangle. The P-values by χ² tests for each pair are indicated at the right.

Figure 6.

Possible pathways for NLR retrotransposition. Major pathways that result in TSD and 3′ MH stretches are indicated by bold arrows (A,B), and the others are indicated by thin arrows (C,D,E). Some other pathways, although not shown, are possible; for example, one that generates long TSTs and 5′ MH. The numbered arrows indicate the following reactions: (1) first-strand cleavage by NLR-encoded ENs, (2a) reverse transcription initiated with the help of annealing of target DNA and NLR RNA, (2b) reverse transcription primed by extra nucleotides at the 3′ end, (3) second-strand cleavage, (4) annealing of nascent NLR cDNA and target DNA, (5) addition of extra 5′ nucleotides, (6) sense-strand synthesis and ligation, (7) addition of extra 3′ nucleotides, (8) nucleolytic digestion of overhanging sequences, (9) introduction of a double-strand break (DSB), and (10) exonucleolytic digestion from the DNA ends. (EXTRA) Extra nucleotides added either the 5′ or 3′ end.