Horizontal Gene Transfer (HGT) is a hallmark of prokaryote evolution and has extensively been studied [1-3]. The vehicles of these events, Mobile Genetic Elements (MGE), transfer functional genes through taxonomic borders, which promote adaptation of bacteria to new habitats and accelerate evolution [1, 4]. Insertions of foreign DNA, known as Genomic Islands (GIs), are readily identified in prokaryotes by various methods and catalogued in multiple, easily accessible platforms for further research and analysis [5-7].

Research on eukaryotic GIs and in particular human GIs are not as comprehensive. Various publications have delved into the contentious issue of HGT in the human genome with varying degrees of HGT detected [8-10]. Even less is known about the possible role that HGT could play in evolution of eukaryotes. The contribution of HGT to the diversification of microbial eukaryotes was demonstrated [11], but it remains disputable whether the acquired DNA fragments can be of any use to higher eukaryotes. One interesting consideration was proposed by Iyer et al. [12] that cell-to-cell signalling molecules could evolve by a lateral exchange from bacteria to multicellular eukaryotes. In this case, the authors suggested that these events took place during the early stages of eukaryote speciation, or possibly after the divergence of chordates from other animal phyla as the boldest hypothesis. There are several publications reporting gene exchange between higher organisms and their bacterial symbionts, for example between Drosophila ananassae and Wolbachia [13], or between eukaryotic chromosomes and genomes of organelles that may be exemplified by multiple inserts of mitochondrial DNA in the nuclear human genome [14]. The latter events are important for tracing down the paths of human evolution from apes and human migrations, but the functional aspect of these events is rather controversial.

As less than 2% of human genomic DNA is ultimately translated into proteins [15-17], it seems highly optimistic that any functionally annotated, large protein would have arisen due to an HGT event. It is also rather difficult to imagine the process of adaptation of acquired genes to the complex regulation of gene expression and alternative splicing in vertebrates. However, the lateral transfer of genetic material is not restricted to functional genes and may include non-coding sequences [18-20].

Historically, long non-coding RNAs (lncRNAs) were perceived as junk DNA. Recently these genetic elements spanning regions on chromosomes larger than 200 bp have been reconsidered and are currently perceived as transcriptionally and functionally important [21, 22]. The origin and location of lncRNAs in vertebrate genomes have been found to be highly influenced and driven by transposable elements due to their ability to incorporate regulatory sequences Upon chromosomal insertions with a substantial proportion of mature lncRNAs containing transposable elements [23]. Recent studies have indicated a direct involvement of lncRNAs in mammalian host cell response to bacterial infections [24-26]. LncRNAs are highly variable across genomes with rapid loss or gain occurring during the course of evolution [27, 28].

The objective of this study was to determine whether the laterally acquired fragments of DNA can play any functional and/or evolutionary role in vertebrates and particularly in human evolution. Functional genetic elements which are readily distributed by HGT are genes for regulatory RNA. An attempt was made in this work to investigate possible association of previously identified lncRNA with predicted GIs. GI prediction was performed by the program SeqWord Gene Island Sniffer (SWGIS v.2.0) [7, 29-31] A modified version of SWGIS v2.0 designed specifically for GI prediction in eukaryotic genomes has been published recently with a database of predicted eukaryotic GIs [32]. SWGIS uses a sliding window approach to exploit the variance of tetranucleotide frequencies across the genome to determine GIs based on their sequence composition. The modified version SWGIS v2.0 accounts also for a higher compositional diversity of eukaryotic chromosomes comprising long stretches of non-coding DNA and frequent repeats [32].

Only sufficiently large GIs, longer than 5 kbp, can be identified by this method. Readers should be informed that this work was not aimed at the creation of a comprehensive inventory of GIs in the human genome, but rather to identify sufficiently large horizontally acquired inserts which potentially serve as vehicles for functional genes.

The availability of predicted GIs in the human genome will enable further research in the developing and exciting field of HGT and GIs in the human evolution. All predicted GIs are freely accessible from http://hislands.bi.up.ac.za.

SWGIS v2.0 was optimized for GI predicting in eukaryotic genomes to alleviate the burden of horizontal transfer detection in massive genomes such as the human genome [32]. The GIs locational preference was found to be unbiased, which is in contrast with Huang et al. [9] who use an alternative approach of HGT prediction based on pairwise sequence comparison. Furthermore, it was found that GIs clustered on the chromosomes with tRNA genes. Besides encoding functional tRNA molecules, these genes are important for regulation of transcription of nearby protein coding genes and in maintenance of chromatin domain architecture by functioning as chromatin barrier elements [47]. Moreover, tRNA genes are preferable sites for integration of mobile genetic elements in prokaryotes, archaea and eukaryotes [5, 48, 49]. Thus the neighbouring position of protein and tRNA coding genes in predicted GIs may reflect both: insertion preference and better functional integration of lcnRNA into transcriptional machinery of the host organism. HGT is common in bacteria with a rather high rate of turnover of GIs by bacterial genomes. Many sequenced strains of microorganisms were characterized with a unique set of GIs shared only by the closest relatives [7]. HGT is important for fast adaptation of bacteria to new eco-niches by sourcing important enzymes from ecotype-specific bacterial consortia [50, 51]. However, many bacterial GIs were crammed with selfish plasmid and phage born genes and probably played no role in the adaptation. Foreign DNA inserts in bacterial genomes are either subjects to excision, if not functional; or they quickly gain the host specific characteristics in terms of codon usage and preferable usage of oligonucleotides in the genomic DNA amelioration process [52]. Contrary, in the higher metazoan organisms a successful HGT into germline cells is a rare event which occurs maybe once in an epoch, whilst retaining the specific oligonucleotide usage patterns for even longer periods of time indicates an absence of genome amelioration processes. At the same time, 80.91% of predicted GIs were functional as they contained lncRNAs genes. GI inserts also contributed to the evolution of protein coding genes, but to much lower extent. Only a few small genes related to the immune system functioning were acquired horizontally in the human genome. Several genes of defence against pathogens and associated with gonadal mesoderm development responsible for sexual dimorphism possibly acquired their exons through HGT.

Sources of GIs and mechanisms of their acquisition by metazoan organisms remain unclear. Comparison against the NCBI nr database did not reveal any links to bacterial DNA sequences or proteins implying that bacteria were not the sources of these elements. High sequence similarities against a few bacterial proteins were identified but these were determined to be likely due to human contamination in the subject bacteria. The most plausible sources of GIs are viruses that corroborates with the latest discoveries in this field [53-56]. A total of 244 lncRNAs located in GIs were found to have sequence similarity (e-value = 1E-30) with viral hosts in the NCBI nt database. These related to 161 different viral sequences. Human endogenous retrovirus sequences were found to be the most abundant viral hits Fig. (4). For example, Fig. (5) shows an alignment of a lncRNA ZBTB11-AS1:8 located in a GI on chromosome 3 that was found to have 99.928% identity over the entire length of the Human endogenous retrovirus HERV-K(II) genome sequence (NCBI accession AB047240.1). Furthermore, this was the second best hit after an alignment to Homo sapiens 3 BAC RP11-454H13 (Roswell Park Cancer Institute Human BAC Library) complete sequence (NCBI accession: AC084198.31). Another lncRNA P3H2-AS1:7 located in a GI on chromosome 3 displayed 91.206% identity over 3,821 bp of Human endogenous retrovirus H HERV-H/env59 (clone 916F3, NCBI accession: AJ289711.1). The only other hits found were against Homo sapiens, Pan troglodytes, Pongo abelii, Gorilla gorilla and Nomascus leucogenys (data not shown).

Fig. (4). Distribution of best BLASTN hits revealed for 244 lncRNA inserts in the human genome against viral sequences in NCBI nt database.

Fig. (5). Alignment of the lncRNA ZBTB11-AS1:8 found in a GI on human chromosome 3 against the entire length of the Human endogenous retrovirus HERV-K(II) complete genome sequence (NCBI accession AB047240.1).

Close relations between several lncRNA inserts and Human endogenous viruses were exemplified by sequence similarity comparison. For instance, a lncRNA insert lnc-AP1S2-2:7 located in a GI on chromosome X displayed high sequence similarity (90.2% identity) with a 2,550 bp region of Human endogenous retrovirus H HERV-H/env59 (clone 916F3, AJ289711.1). Sequence similarity analysis of this region aligned to other sequences found in Homo sapiens, Pan troglodytes, Pongo abelii, Gorilla gorilla and Nomascus leucogenys is displayed as a dendrogram in Fig. (6A). Another tree was designed for sequences sharing similarity with the lncRNA MIR3681HG:27 found in a GI on chromosome 2. This insertion also displayed high sequence similarity (91.2% identity) with a 2,033 bp region of Human endogenous retrovirus H HERV-H/env59 (clone 916F3, AJ289711.1) as shown in (Fig. 6B).

Fig. (6). Dendrograms of sequence similarity revealed between sequences of lncRNA A) lnc-AP1S2-2:7; B) MIR3681HG:27 and homologous loci of other chromosomes and Human endogenous retroviruses (highlighted in the dendrograms). Bootstrapping numbers of sequence clustering are shown.

While retroviral genes and sequences are common elements of repeats of the human genome, a recent study reported fossils of other prehistorical non-retroviral genomes in human chromosomes [53]. The conceptually novel perception of viruses and their role in the evolution of life was proposed by Moelling [54]. According to this theory, only a small portion of viruses are strict pathogens causing specific diseases or cancer, while the majority of viruses exist in the environment as a well-balanced ecosystem and contribute significantly to the evolution of single cell and multicellular organisms. This hypothesis was supported in a later publication by Durzyńska and Goździcka-Józefiak [55], where more details on the role of HGT as carried out by environmental viruses in the formation of tree of life were given. More specifically, the role of genomic inserts of viruses in formation of the antiviral immunity in vertebrates by HGT was reported by Villarreal [56].

It may be concluded that HGT played an important role in the formation of regulatory networks of vertebrates and particularly in human evolution. The rate of acquisition of novel lncRNA increases when organisms dramatically change their lifestyle, for example at the time of appearance of first terrestrial vertebrates; or during rapid expansion and speciation as a part of the founder effect. Acquisitions of new lncRNA are rare events yet the GIs retain their compositional specificities throughout epochs and as such may be used for timing and reconstruction of evolutionary events in eukaryotes.

The most recent possible lncRNA acquisition event with regards to the Chimpanzee genome was located in a GI on X chromosome (55,978,495 – 156,028,276). This GI (GI0530:244, chrX: 55,978,495 – 156,028,276 in hislands.bi.up.ac.za) contains 13 lncRNAs, one of which is WASIR1 (WASH and IL9R antisense RNA 1, ENSG00000185203). Gene expression was found to be the highest in human ovary samples with no known orthologues in the UCSC Genome Browser. Furthermore, it has been reported that a non-coding sequence located and within this GI on the X chromosome may be involved in the pathogenesis of X-linked calvarial hyperostosis [57]. Unfortunately, the rather high level of paralogy of lncRNA genes in the human genome prevents us from making more specific conclusions regarding the role of every individual lncRNA insert.

The importance of lncRNAs in innate and adaptive immune responses in combination with varying degrees of conservation across species may indicate a rapid evolutionary acquisition of these enigmatic regulators by means of HGT in the human genome.