Search

Transposable elements (TEs) are endogenous components of eukaryotic genomes, constituting 45% of human DNA. The human genome project revealed that human endogenous retroviruses (HERVs) constitute about 8% of the sequence. HERVs are derived from sequences integrated into germ cells during retrovirus infections, up to 25 million years ago. However, most copies of HERVs are defective in multiple ways. The HERV-K family is the youngest family and likely has significant biological activity because of its protein coding capacity. HERV-K activity may be involved in a variety of cancers, and in particular may play an important role in human melanoma. In this project, HERV activity in melanoma and different cancer cell lines was investigated. Our results showed the HERV-K pol gene is expressed in melanoma and in breast, prostate and colon cancers. Furthermore, the response to serum starvation conditions is not simply related to increased expression of HERV-K genes, and changes in cellular phenotype under serum starvation are limited to particular melanoma cell lines, rather than being a general phenomenon. HERV-K env protein expression in melanoma cells was compared to normal primary melanocytes - 1% FBS serum starvation can increase expression of this protein in SKMel5 cells that retain an adherent phenotype. Moreover, env protein expression is significantly increased in T47D breast cancer cells under 1% FBS. The analysis of HERV-K LTR methylation state demonstrated that HERV K env and gag proteins in melanoma cells under 10% FBS and 1% FBS conditions decreased. The most interesting finding was the detection of 98 candidate loci as novel proviral insertions where previously these loci were annotated as solo LTRs. This result suggests that proviruses are systematically excluded from assemblies and the census of HERV-K proviruses is much greater than represented in assembled genomes. The Genome-wide amplification of proviral sequences combined with Next Generation Sequences (GAPS –NGS) is established as an effective approach to discover new proviral loci.

With the advent of Next Generation Sequencing (NGS), we have witnessed the generation of enormous volumes of short read sequence data, cheaply and on short time scales. Nevertheless, the quality of genome assemblies generated using NGS technologies has been greatly affected by this innovation, compared to those generated using Sanger DNA sequencing. This is largely due to the inability of short read sequence data alone to scaffold repetitive structures, creating gaps, inversions and rearrangements and ultimately resulting in assemblies that are, at best, draft forms (by draft we mean, assembly that is only a preliminary result that will require more work to be done to make it a more complete and accurate representation of the genome). Single molecule long-read sequencing (SMS) technologies on the other hand, address this challenge by generating sequences with greatly increased read lengths, offering the prospect to better recover these complex repetitive structures, concomitantly improving assembly quality. Following this development, we evaluate the ability of SMS data (specifically Pacific Biosciences SMRT data and Oxford Nanopore MinION data from human genomes) to recover poorly represented repetitive sequences (specifically, GCrich human minisatellites), identify novel transposable element insertions and enable the closing of gapped regions. Our results show that by using single molecule sequencing and long read technology, poorly represented repetitive sequences (specifically, minisatellites and L1s) and other missing elements in published human genome assemblies can be characterized by developing custom software, scalable for the analysis of single molecule long-reads (particularly, Pacific Biosciences’ SMRT technology). The tool designed is cross-platform, thus, giving computational and non-computational biologists a straightforward approach and less technical platform for local analysis of specific poorly characterized DNA sequences.

Human retrotransposons are genetic elements that copy themselves into new locations in the genome by way of an RNA intermediate. They are extremely numerous making up at least 45% of human DNA. Retrotransposon insertions are a major source of inter-human genetic variation, and have been known to cause disease. They are also intrinsically difficult to analyse in genomes due to their highly repetitive nature. In humans there are three currently active retro-transposable elements: LINE-1, Alu and SVA. LINE-1 is an independent element and Alu and SVA parasitise the LINE-1 retrotransposition machinery. There are experimental ways of discovering and analysing such elements, but they require significant investment, while human sequence datasets containing potentially usable data are multiplying at an ever increasing rate. In particular there are now many assembled human genome sequences as well as new sources of whole genome high throughput sequencing data, such as the 1000 Genomes Project. For this reason this study is devoted to using bioinformatic approaches to extract new knowledge about human retrotransposons from the existing datasets. Previous efforts, by past members of this research group, have been devoted to analysing the genomic variation of the LINE-1 element itself. However this study focuses on the extraction of presence / absence variation in the LINE-1 -dependent elements, Alu and SVA. In addition to building software to extract this information from a wide variety of data sources, this project has also involved making the information data available to non-specialist researchers in the form of a website. The tools developed and described here utilise generic design principles, enabling rapid, largely automated updating, necessary with the constant expansion of the underlying data.

Recent high throughput studies have led to the discovery of de novo L1 retrotransposition in malignant somatic cells, as well as large numbers of novel insertions, many of which are highly active in cell culture assays. These data suggest that L1 elements are robustly active, undergoing extensive diversification in contemporary human genomes. Despite this there is little direct evidence of endogenous L1 retrotransposition in the human germline or early embryogenesis: data from very rare disease causing insertions is indirect, subject to strong acquisition bias, and is often equivocal with respect to the origin of the insertions. For L1s to be evolutionarily successful they must retrotranspose during early human development or in the germline, in order to transmit copies to the next generation. The purpose of this thesis was to develop sensitive and yet robust methods to screen human embryos and embryonic cell models for de novo full-length endogenous L1 insertions. We developed a new high throughput sequencing technique, which was able to recover single molecule retrotransposition events. Based on this technique we identified 172 candidate novel L1 insertions in a total of three human embryos, represented by whole-genome amplified DNA of individually dissected blastomeres and the remaining blastocyst tissue. 57 of these insertions are potentially genuine de novo endogenous L1 insertions. Moreover, we have identified a candidate germline specific L1 insertion from a healthy adult donor. Therefore, this study has detected candidate de novo L1 retrotransposition events in human embryos and germlines, using an approach that enables complete validation and characterization of the insertions, despite operating at the single molecule and single cell level. We consider this technical innovation will be most significant in the ongoing dissection of how L1, the dominant human transposon, is actively driving the evolution of modern human genomes.

The human genome contains vast numbers of sequences that have copied themselves to new genomic locations by retrotransposition. Long Interspersed Nuclear Element-1 (LINE-1 or L1) is the only sequence in the human genome still capable of autonomous retrotransposition. L1 elements have contributed to the evolution of the human genome via insertional mutagenesis, pseudogene formation, sequence transduction, and recombination events (producing insertions, deletions and inversions). Currently general and L1- specific sequence databases do not reflect the true level of Full Length Human Specific L1 (FL-L1HS) variation, due to the polymorphic nature of these elements and the way the databases were compiled. Methods to identify FL-L1HS were applied to three sequence assemblies (Reference, Celera and HuRef) and the nucleotide accession database from NCBI. A non-redundant set of 533 FL-L1HS was discovered in these four sources, of which 164 resided in genes. The trace archives from Ensembl were also searched and a further 48 potential FL-L1HS were found. Computational analyses showed 154 FL-L1HS were potentially capable of retrotransposition, including 54 that resided in genes. Alongside these analyses a Target Site Duplication (TSD) detection and analysis tool, TSDmapper was developed to automatically detect TSDs in FL-L1HS sequences and provide annotation on sequence transduction. TSDmapper was used to predict the pre-insertion sequence of all 533 unique L1s, which facilitated in-silico genotyping. A new informatic resource, baseLINE (http://baseline.gene.le.ac.uk), was created to display and enable searching of all the L1 annotation information generated. Data can be viewed in a genomic context in chromosome ideograms or can be exported via the Distributed Annotation Service (DAS) on to the Ensembl genome browser. TSDmapper is also provided as a web application at baseline for users to perform TSD annotation of their sequences of interest.