Assessment of genetic variation for the LINE-1 retrotransposon from next generation sequence data

The Long Interspersed Nuclear Element 1 (LINE-1) retrotransposon is a repeat sequence that comprises roughly 21% of the human genome [1]. As many as 100 full length copies of the LINE-1 element have been demonstrated to be active and capable of retrotransposition within the human genome [2]. These genetic elements provide a dynamic and pervasive mechanism by which the genome may mutate, and through which it may evolve [1, 3]. Several LINE-1 elements have been implicated as disease causing or mediating agents [4]. Since LINE-1 is retrotranspositionally active it is expected that analyses of emerging personal genome data would reveal a great many insertions and deletions relative to the published reference sequence for the human genome. Projects such as the 1000 genomes project provide an unprecedented opportunity to analyze the complete genomes of large populations in order to assess the amount of diversity present among them. In this work, pilot trace data from the 1000 genomes project (http://www.1000genomes.org) is analyzed for genetic diversity in terms of sequence elements that suggest full length, and therefore potentially active insertions of the LINE-1 retrotransposon relative to the published reference sequence for the human genome. Twenty-two of these elements have been identified in this work. Next generation sequencing technology is emerging as an affordable tool that will allow clinicians and researchers to develop a comprehensive picture of an individual’s genetic composition. The results of this study demonstrate clearly that as the medical field moves toward personalizing medical treatment, it will be important to assay an individual’s genome for LINE-1 elements unique to them that may prove disruptive to normal cellular function.

Retrotransposons are genetic elements that are capable of replicating themselves throughout the genome. The process of retrotransposition results in the de-novo insertion of either a full length, or truncated copy of the retrotransposing element into an individual’s genomic DNA. Previously unreported insertions can be found when analyzing a diverse human population [5], and have even been detected when comparing human genome assemblies from different sources [6]. In a study similar to the work presented here, it is reported that the majority of intermediate length structural variation that distinguishes inbred mouse strains from one another was caused by “…endogenous retrotransposition, predominantly by L1 retrotransposons” [7].

The full length LINE-1 retrotransposon is approximately 6000 nucleotides in length, and is characterized by two open reading frames, referred to in the literature as ORF1 and ORF2, which code for the proteins necessary to support the retrotransposition process [8]. Specific LINE-1 alleles have been identified as a disease causing or mediating agents (ref [4] and references therein). Although the function of the majority of the genome is not yet understood, it is clear that a retrotransposition event could introduce damage by inserting LINE-1 into an exon, or untranslated regions of the genome that may have regulatory function, and as a result modify a transcript, or modulate expression [9]. Also, it has been described [9–12] that LINE-1 contains the necessary functional elements, namely a 5′ UTR promoter that is capable of promoting transcription in either direction, that would drive not only the expression of LINE-1, but perhaps also of the genes that neighbor it. Although no specific evidence has been reported, this makes plausible the notion of ectopic expression of genes to which the LINE-1 insertion is proximal, or intragenic [9]. Furthermore, it has been reported that LINE-1 expression is increased dramatically under exposure to UV radiation [13], and under insult of the common environmental pollutant benzo(a)pyrene [14, 15]. Increased expression of this element either endogenously, or due to environmental insult would not only serve to increase its ability to effect the expression of neighboring genes, but would also increase the likelihood of retrotransposition and, therefore possibility of somatic mutation. [3] Since this mobile element has been tied to disease, and clearly has the potential to wreak havoc with each retrotransposition event (the rate of which has estimates of between 1 in 30 and 1 in 500 in human germ cells [2]), it is clear that this is an element that must be accounted for when assessing an individual’s genetic composition.

In this study, only putative full-length LINE-1 insertion events were investigated. The reason for this is twofold. First, the full-length LINE-1 insertions are likely to have the greatest functional impact on their region of the genome. Secondly, the vast majority of the LINE-1s that are found in the genome are truncated and are missing the 5′ end. Identifying those reads that contained an insertion site for a truncated copy of LINE-1 is a more complex challenge as it is not clear that a read that contains LINE-1 sequence will also contain an insertion site, or will otherwise be comprised completely of LINE-1 sequence. By limiting the search to the 5′ end of the LINE-1 transcript, it was guaranteed that each read that contained this sequence segment would also contain the insertion site. It should be noted that LINE-1 deletion events were not considered in this phase of the project.

Of the 22 insertion sites identified for NA12878, 10 of the inserts are intragenic, and of those, 3 were located on the sense strand relative to the gene in which they were inserted. We chose to validate the 3 sense intragenic insertions (PRKCA, GLCCI1, and TRDN in Table 1) via allele specific PCR, as well as an intergenic insertion (position 21243489 on chromosome 5) for which there was no strong evidence within the sequence data to conclude that the insertion was inherited. LINE1+ and LINE1- alleles identified in the sequencing data were confirmed via allele specific PCR using methods described in the methods section, similar to those reported in Bennett et al.[5] The gel images are shown in Figure 1. The PCR products produced confirmed the results derived from the sequencing data, and confirmed that the insertion at position 21243489 on chromosome 5 for the daughter was inherited.

Finally, it should be noted that the decision to utilize the exact string match criteria described in the methods section will prove restrictive with respect to the comprehensive identification of all novel LINE-1 insertions in these datasets, both in terms of full length inserts that vary, even by a single base, from the search subsequence, as well as for even modestly truncated LINE-1 inserts. The manifold of subsequences that would have been necessary for a comprehensive search using this exact string match criteria would have been very large. This is a shortcoming of the search strategy that will soon be alleviated as next generation sequencing technologies improve and read lengths of 70 to 500 nucleotides become available from the 1000 Genomes project and other whole genome sequencing efforts. At that point, more comprehensive search strategies as described in the methods section that are tolerant of both sequence variation, and LINE-1 truncation will be possible since the longer reads that span LINE-1 splice sites will be more likely to contain enough LINE-1 sequence to confidently identify high fidelity matches to LINE-1, as well as sufficient flanking sequence to uniquely place the insert site in the genome. Specifically, we will be able to employ robust search and alignment tools such as BLAST[16, 17] and BLAT[18] instead of the exact string match. However, even using the exact string match criteria we successfully identified 22 putative full length insertions that do not appear in the reference genome which is suggestive of hundreds of insertions that will be identifiable when using more comprehensive search strategies.

As the price of sequencing is driven down by next generation technologies, the reality of personal genomes for use in research, the clinic, and personal edification is destined to become a reality. In 2007 and 2008 the results of the deep sequencing of Craig Venter [19], and James Watson [20] were reported respectively. The former was produced using Sanger technology, and the latter, next generation technology. Since then, larger scale sequencing projects have begun such as the 1000 genomes project, the personal genome project (http://www.personalgenomes.org/), and additional individual projects are being undertaken [21, 22]. A finding common in each of the published individual sequencing efforts is the discovery of hundreds of thousands of previously unreported polymorphisms. Conservative estimates from the data produced in the Venter and Watson studies place the number of new SNPs in excess of 600,000, and the number of new IN/DELs in excess of 22,000 [20].

It has previously been reported that there are in excess of 7000 full-length copies of LINE-1 in the reference genome sequence [23]. In this work, 22 insertions are being reported in an individual that are consistent with full-length LINE-1 elements (i.e. putative) known to be retrotranspositionally active that have not been identified previously. However, no de-novo insertions of putative full-length LINE-1s were detected in this sample. The insertions detected here were identified using an exact match to a 14 nucleotide sequence for reasons that are specific to the short reads available at the time of analysis (see methods). As longer reads become available, more flexible alignment searches will be possible, and the numbers of detectable insertions will, almost certainly, increase significantly.

A limitation of current next generation sequencing technology is the ability to fully characterize these insertions. Current sequence lengths are on the order of several hundred nucleotides, and the full-length LINE-1 element containing both ORF1 and ORF2 is greater than 6000 nucleotides in length. A full characterization would include the assignment of the LINE-1 to one of the many evolutionary families known to exist within humans [24]. Full-length sequencing of the entire insert is necessary not only to assign the LINE-1 to a family, but also in order to be able to assess the integrity of the LINE-1 open reading frames, and the internal promoter elements. Mutations in these regions will significantly affect the activity of the LINE-1 element, and thus, its impact on cellular function.

Several of the LINE-1 elements identified here are intragenic, may be full length, intact elements, and thus, may be of clinical significance. It is critical that as next generation data is analyzed, care is taken not to obscure the data necessary to perform analyses such as these. It is common to mask/trim known repeats prior to mapping a trace. If masked, the high quality traces selected here would not appear anomalous, and the identification of these insertions would be impossible in the mapping process. However, if the repeat region is annotated for the read (even if it is masked), the fact that it aligns to a region with no currently known repeat region would signal the presence of an insertion. Such considerations are important, as it is currently not clear how scientists will utilize this information, whether at the level of the assembly, or if the called reads will be the basis of further analysis.

The findings of this study and those published previously make clear that as more individual genomes are sequenced there will be vast increases in the number of reported polymorphisms. The technology now exists for scientists and clinicians to perform a comprehensive assay of genetic makeup of individuals. The next step is to assess the clinical relevance of these individual differences, as evidenced by diseases/disorders of altered gene expression, genomic instability or genetic toxicity. Of general interest are mutations in genes encoding for proteins involved in basal transcriptional control and responsible for reordering regions of broad local enrichment and control of chromatin structure (BLOCS). Epigenetic silencing of LINE-1 by way of DNA methylation of CpG loci within the LINE-1 promoter [25] via DNA methyltransferases as well as the involvement of E2F/Rb/HDAC complexes in the regulation of chromatin status [26] allow for the possibility of new treatment of LINE-1-mediated human diseases, once they are identified. Epigenetic drugs may help in disease treatment via modulation of the epigenetic status of BLOCS where LINE-1 sequences act as centers for epigenetic nucleation and heterochromatin formation. For example, DNA demethylating agents such as the antineoplasic 5-AzaC, and histone deacetylase inhibitors alone or in combination could restore gene expression levels of eukaryotic genes silenced as a result of LINE-1 insertion and LINE-1-mediated epigenetic silencing.

LINE-1 is an active retrotransposon that has been demonstrated to affect clinical phenotype [4, 27]. In this study, twenty two previously unreported insertion sites have been identified for an individual that are consistent with full length insertions of the LINE-1 retrotransposon, 10 of them intragenic. As we move toward the paradigm of personalized medicine, it is clear that more work is necessary to fully catalogue the diversity with respect to this element within the human population. This will allow us to assess the allele frequency of all existing LINE-1s, and begin to classify them with respect to clinical relevance. This understanding will put us in a much better position to evaluate the impact of de novo insertions as they are discovered since this is an element that should be accounted for when performing genetic tests for disease causing mutations.