Heads or tails: L1 insertion-associated 5' homopolymeric sequences

Abstract

Background: L1s are one of the most successful autonomous mobile elements in primate genomes. These elements comprise as much as 17% of primate genomes with the majority of insertions occurring via target primed reverse transcription (TPRT). Twin priming, a variant of TPRT, can result in unusual DNA sequence architecture. These insertions appear to be inverted, truncated L1s flanked by target site duplications.

Results: We report on loci with sequence architecture consistent with variants of the twin priming mechanism and introduce dual priming, a mechanism that could generate similar sequence characteristics. These insertions take the form of truncated L1s with hallmarks of classical TPRT insertions but having a poly(T) simple repeat at the 5' end of the insertion. We identified loci using computational analyses of the human, chimpanzee, orangutan, rhesus macaque and marmoset genomes. Insertion site characteristics for all putative loci were experimentally verified.

Conclusions: The 39 loci that passed our computational and experimental screens probably represent inversion-deletion events which resulted in a 5' inverted poly(A) tail. Based on our observations of these loci and their local sequence properties, we conclude that they most probably represent twin priming events with unusually short non-inverted portions. We postulate that dual priming could, theoretically, produce the same patterns. The resulting homopolymeric stretches associated with these insertion events may promote genomic instability and create potential target sites for future retrotransposition events.

Figure 1

Classical target primed reverse transcription (TPRT), twin priming, variants of twin priming and dual priming mechanisms. (a) A schematic of classical TPRT. The poly(A) tail of an L1 mRNA anneals to the target site created by L1 endonuclease. L1 reverse transcription (RT) primes at the target site and synthesizes the bottom-strand cDNA. A subsequent second-strand nick and synthesis results in an L1 insertion with a 3' poly(A) flanked by TSDs. (b) Twin priming. In this variant of TPRT, after the second-strand nick, a site internal to the mRNA anneals to the top strand overhang. A second RT molecule primes at this site, generating an inverted L1 cDNA. (c) This twin priming variant involves the disengagement of the first RT before reaching the end of the poly(A) tail, resulting in an insertion with a 5' poly(T) stretch, but lacking a 3' poly(A) tail. Like classical twin priming, this mechanism results in an inverted L1 structure. (d) A second twin priming variant creates an insertion with both a 3' poly(A) tail and a 5' poly(T) stretch. The first RT falls off before reaching the end of the poly(A) tail. (e) Dual priming. Classical TPRT involving the first mRNA begins on the first strand. After the second strand nick, a second mRNA anneals to the second strand and undergoes classical TPRT. Note that this panel is rotated 180° relative to the orientation of all other panels. This is done to show that the resulting insertion will appear the same to computational filters as the above twin priming variant.

Figure 2

Investigation of candidate loci and variations within the homopolymeric stretches. (a) A triple alignment of pT684 to two outgroup species, the rhesus macaque and the common marmoset. The TSDs are highlighted in grey, the poly(T) stretch in green, and the L1 is highlighted in blue. (b) A gel chromatograph of polymerase chain reaction products depicting an insertion present in humans, chimpanzees, gorillas and orangutans, but absent in rhesus macaque and owl monkey. (c) Internal primers were designed around the poly(T) stretches for all human-specific loci; two loci are shown here. For each locus, HeLa DNA and a mixture of the DNA of 80 human individuals was run out on a 4% agarose gel with 100 bp and 20 bp ladders. No within-species variation in poly(T) length was observed.

Figure 3

Alignment of candidate L1s to their L1 consensus sequences. Schematic of the position of each candidate L1 when aligned against an L1 consensus sequence. Stars indicate that the 3' end of the locus aligns to a portion of the poly(A) tail in the consensus. Loci are color-coded to indicate in which species each was found.

Figure 4

Microhomology and comparison of insertion site characteristics of each locus. (a) An analysis of the microhomology of the six nucleotides adjacent to each insertion junction (with '1' indicating the nucleotide closest to the insert) to the corresponding sequence in the putative mRNA. Exclusion of a junction from analysis is indicated by a (-) and positions where microhomology is found are shaded grey. Those positions at which significant microhomology were found are highlighted in blue. (b) A binomial distribution analysis of the 6 bp at each junction revealed significant microhomology at both the 3' and 5' junctions of the insertions. No significant microhomology was found at the internal junction. P-values highlighted in blue are significant at p < 0.001. (c) A WebLogo analysis of the 6 bp found at the 3'junction. The logo supports our finding of microhomology at this junction, and is consistent with the expected motif at the L1 endonuclease cleavage site.

Figure 5

L1 endonuclease (EN) cleavage site analyses at the 5' and 3' junctions. For both the 5' and 3' target sites of each locus, the last four nucleotides of the target site and first nucleotide of the flanking sequence were compared to the canonical L1 EN cleavage motif (5'-TTTT/A-3'). To investigate the possibility that the candidate L1s were inserted in the antisense orientation, the 5' target site was reverse complemented and analysed. The black bars show the frequency of each divergence value at the 3' target site among our 39 loci, while the blue bars show values for the 5' target sites. The 3' target sites show more divergence from the typical EN cleavage motif than the 5' target site.