Most retroelements have been considered harmful because they cause accumulation of insertion and deletion mutations in the host genome [1]. Mutation of retroelements could affect gene transcription and translation. However, recent investigations have shown that HERV and Alu elements in the intron or flanking regions of functional human genes provide alternative promoters, splicing sites and polyadenylation signals [2, 3]. Unlike HERV and Alu, LINE elements tend to contain multiple potential splice sites (ESE) [4] and polyadenylation signals [5] in their sequences. There are four types of transposable elements in the human genome: long interspersed nuclear elements (LINEs or L1s) or non-long terminal repeat retrotransposons, short interspersed nuclear elements (SINEs), LTR retrotransposons (endogenous retroviruses) and DNA transposons [1], which together constitute 45% of the total genome. Most of these elements are inactive. However, a few LTR elements have been shown to contain intact open reading frames (ORFs) [6], and LINE elements also have the capacity for autonomous retrotransposition [7, 8]. SINE elements cannot be expressed by themselves and depend on L1 elements for active mobility [9]. The L1 elements constitute about 17% of the human genome and are present in an estimated 79% of human genes in at least one copy [10].

The full length of L1 is about 6 kb. It consists of a 5' untranslated region (5'UTR); two nonoverlapping open reading frames (ORF1 and ORF2) encoding an RNA binding protein [11], an endonuclease [12] and a reverse transcriptase [13]; and a 3'UTR that ends in an AATAAA polyadenylation signal and a polyA tail [9]. The Alu and SVA transposable elements and processed pseudogenes are believed to have been inserted into the genome by borrowing the endonuclease and reverse transcriptase from L1 elements [14–16]. The L1 element itself has also been inserted into new genomic locations during mammalian evolution. Such elements are mostly truncated and rearranged to form inactive copies of their progenitors. These insertional mutations are reported to be associated with twelve genetic diseases [17] and also contribute to protein variability or versatility [18].

Active or functional L1 elements, which are involved in shaping the human genome, are differentiated into three types depending on where they are inserted into the genome. First, a 6 kb-long full-length or variable-length 5'-truncated L1 element is inserted into the 5'UTR or introns of a gene, affecting its expression. In this process, LINE elements are probably reverse transcribed and integrated in the new location by target-primed reverse transcription (TPRT) [19]. LINE elements have provided not only many internal promoters at new genomic locations, but also 5'-UTR-located internal promoters, which could guide the transcription of many adjacent genes [20]. Second, retrotransposition of the L1 element results in the transduction of a 3'-UTR flanking fragment to a new genomic location; this is due to the effect of the ambiguous L1 polyadenylation signal [21]. Third, the L1 components are shuffled into exons, affecting the splicing site at transcription and consequently leading to the production of alternative mRNA transcripts [22].

Assembling genomic information and constructing a web-database of genome annotations and genes with particular functions is generally useful for implementing functional studies and for understanding evolutionary genomic organization. Representative web-databases of transposable elements in the human genome have been reported: a database of Alu elements incorporated within protein-coding gene [2], an HERV expression and structure analysis system [3] and a system for extrapolating functional annotation to the prediction of active LINE-1 elements [23]. Although it is well established that information about the structure and position of LINE elements in genes is important for functional studies of genetic diseases, such data are limited and are not included in any database that allows large amounts of scattered information to be searched easily. To address this deficiency, we developed a database for LINE expression and structure in the human genome, LINE FUSION GENES. Our database provides the structures and expression patterns of LINE elements including their relative positions in the genes, and additional information such as the tissue distribution and chromosomal location of the genes and their domain structures. To enhance ease of access for subsequent users, we linked all of the accession numbers to the NCBI data bank to provide mRNA sequences.

LINE FUSION GENES uses JSP technology; the data come from a primary database. Users can efficiently retrieve three modes of information concerning LINE expression within genes. First, they can search LINE expression within a gene by typing a gene ID or clicking on the gene name listed on the view page according to its chromosomal location. Second, the database provides type information in which LINE expression is classified into three types (alternative promoter, alternative polyadenylation signal and exonization). The type information can help users to speculate more readily about the effects of LINE expression within interesting genes. Third, users can search interesting genes using accession numbers from the NCBI data bank or from the HUGO symbol name provided on the view page, and even acquire mRNA sequences from the NCBI data bank for further study.

The result pages are listed in a tabular format that provides the evidence for and information about LINE expression within genes. As shown in Figure 2, the LINEs are visualized by colors: red (5' UTR elements), blue (3' UTR elements) and green (ORF1 and ORF2). LINE fusion regions within mRNAs are indicated in red. Moreover, detailed information about the LINE fusion regions are displayed in the table on the result page. Occasionally, LINE incorporation results in domain changes in a protein. In order to speculate about these domain changes, users can check the domain description on the page. The domain information includes the results obtained from searching queries about genes with LINEs by RPS-BLAST [29].

From our in silico analysis of the human genome, 1,329 genes were identified as being affected by LINE elements during expression. LINE FUSION GENES is continually supplemented with new human gene data from the available sources. We are planning to update the database with full length human cDNA data obtained from various clinical samples representing human diseases. Through this update, we will be able to profile the patterns of LINE expression in various diseases and to identify LINEs that affect the expression of functional human genes. We will also supplement the database with LINE fusion genes from other mammalian species and compare them with those of humans. We also envision the integration of our HESAS [3] and LINE FUSION GENES databases, intended for release in 2007. We believe that our work will help us to gain insight into the implications of LINE expression for human evolution and disease.

LINE FUSION GENES is publicly available at the URL http://www.primate.or.kr/line. Questions and comments are welcomed through the site.