|
|
Genomics. Author manuscript; available in PMC 2008 August 1. Published in final edited form as: Published online 2007 May 22. doi: 10.1016/j.ygeno.2007.04.003. | PMCID: PMC2065750 NIHMSID: NIHMS27599 |
Computational and Biological Inference of Gene Regulatory Networks of LINE-1 Retrotransposon Kenneth S. Ramos, Qiang He, Ted Kalbfleisch, Diego E Montoya-Durango, Ivo Teneng, Vilius Stribinskis, and Marcel Brun Department of Biochemistry and Molecular Biology and Center for Genetics and Molecular Medicine, University of Louisville School of Medicine Computational approaches were used to define structural and functional determinants of a putative genetic regulatory network of murine LINE-1 ( Long Intersperced Nuclear Element-1), an active mammalian retrotransposon that uses RNA intermediates to populate new sites throughout the genome. Polymerase (RNA) II polypeptide E AI845735 and mouse DNA homologous to Drosophila per-fragment M12039 were identified as primary attractors. siRNA knockdown of the aryl hydrocarbon receptor NM_013464 modulated gene expression within the network, including LINE-1, Sgpl, SynBP and MGST1. Genes within the network did not exhibit physical proximity and instead were dispersed throughout the genome. The potential impact of individual members of the network on the global dynamical behavior of LINE-1 was examined from a theoretical and empirical framework. Keywords: aryl hydrocarbon receptor, genetic regulatory networks, LINE-1, retrotransposon, siRNA knockdown The availability of complete genome sequences for several species has opened the door to advances in our understanding of genomic structure and function. Nevertheless, relatively little is understood about the regulation of ubiquitous gene-sized segments of DNA known as transposable elements, repetitive sequences that populate different chromosomal locations in the host genome. Approximately 46% of the human genome represents transposable elements which serve as intergenic sequences that interrupt protein encoding genes and participate in the regulation of gene expression [ 1]. A similar pattern is observed in the mouse where approximately 38% of the mouse genome is transposon-derived [ 2]. Transposable elements are found in multiple genomic compartments, including pericentromeric, heterochromatin, telomeres, gene regulatory regions, exons and introns [ 3]. Retrotransposon activation plays a role in genomic diversity and evolution [ 4], double strand break DNA repair [ 5], exon shuffling [ 6, 7], gene silencing and transgenesis. The relative contributions of LINE-1 to biological processes other than genomic insertions are not yet well understood and continue to be debated. Retrotransposition has been implicated in several human diseases including, hemophilia [ 6] and breast and colon cancers [ 8, 9]. The fact that transposable elements are prone to stochastic epigenetic silencing makes them ideal candidates to explain phenotypic variations that cannot be explained by differences in DNA sequence. A functional LINE-1 element in both mice and humans is ~6 kb in length and consists of a 5′ UTR (untranslated region), two open reading frames (ORF1 and ORF2), and a 3′ UTR terminating in poly (A) tail (). The mouse contains monomers in its 5′ UTR that function as promoters and exhibit redox-sensitivity [ 10], while the human contains an internal promoter with SOX and YY1 DNA binding regulatory sites [ 11, 12] (). Although the functional significance of ORF1 is not clear, ORF1 protein is a non-sequence-specific RNA binding protein [ 13]. ORF2 contains three domains critical for LINE-1 propagation: endonuclease [ 14], reverse transcriptase [ 15], and a C-terminal cysteine-rich motif [ 16]. The regulation of retrotransposon expression in somatic cells is mostly directed at repression and silencing via DNA methylation and RNA interference (RNAi) [ 17, 18], presumably to preclude the negative genomic consequences of retrotransposition within the host genome. Other than its role as an insertional mutagen, little is known about the genetic interactions of LINE-I with other genes within the mammalian genome. In silico analyses of the mouse and human genomes have shown that LINE-1 may coordinate gene transcription by providing regulatory sequences that direct the expression or silencing of gene expression [ 17, 18]. This theoretical framework is consistent with the finding that over 3% of the mRNAs in germ cells and preimplantation embryos contain retrotransposon regulatory sequences[ 19], and that retrotransposons participate in control of developmental gene expression in mouse oocytes and preimplantation embryos[ 20]. Hence, highly repetitive elements, including retrotransposons, are no longer viewed solely as parasitic entities, but rather as sequences that play active roles in the control of gene expression. Retrotransposons are actively transcribed when plants and animals are subjected to cellular stress, suggesting that retroelements function as integral components of the global genomic response to environmental stress [ 21- 26]. Members of the SRY family of transcription factors positively regulate human LINE retrotransposons by binding to the LINE-1 promoter [ 11]. Moreover, LINE-1 transcription is upregulated by steroid hormones and steroid hormone-like molecules [ 27- 29], as well as chemical carcinogens and UV light [ 26]. These functional interactions suggest that a genetic regulatory network exists that controls reciprocal interactions between LINE-1 and other genes within the mammalian genome. The underlying conceptual framework of the present study is that a gene regulatory network exists that coordinates LINE-1 expression as part of the adaptive response of the organism to changes in its environment. Because LINES exist under tight genetic control, and are silenced under most conditions, the existence of such a gene regulatory network would either be functionally linked to host silencing mechanisms or simply coincident with periods of LINE-1 expression. To elucidate putative genetic regulatory networks of LINE-1, we applied computational algorithms that identified genes predictive of LINE-1 expression based on statistical evaluation of co-expression profiles from large-scale simultaneous measurements of gene expression made using DNA microarray technology. The chromosomal location of genes within the LINE-1 genetic regulatory network was examined to determine if patterns of co-regulation within the network are related to physical proximity or involve other commonalities in molecular regulation. Functional genomics analyses were completed to examine the biological connectivity of genes within the LINE-1 regulatory network. The identification of relevant components within the LINE-1 regulatory network was modeled using the normalized intensity values for greater than 12,000 genes from 235 independent DNA microarray hybridizations. The expression of genes with the highest coefficient of variation ( cv) was used to create predictor/training data sets with transcript levels categorized into ternary expressions and tested for all possible predictor combinations. The quality of prediction for each set was quantified using a multivariate non-linear measure of determination termed Coefficient of Determination (CoD). A number of three-clone combinations met the selection criteria [ ![[theta w/ macron]](/corehtml/pmc/pmcents/x03B8x0304.gif) ≥ 0.5 and ![[epsilon with macron]](/corehtml/pmc/pmcents/x03B5x0304.gif) [ψ opt] < 0.05], with one or two clones identified as predominant predictors within the sample pool ( Table 1). Among the best predictions (highest CoD values) of L1 target by three-clone combinations were: chemokine (C-C motif) ligand 2, chitinase 3-like 3, chloride intracellular channel 3, chemokine (C-X-C motif) ligand 1, extracellular link domain-containing 1, microsomal glutathione S-transferase 1, Mouse DNA homologous to the Drosophila per locus, Mus musculus 10 days neonate medulla oblongata cDNA, phenylalanine hydroxylase, protein tyrosine phosphatase receptor type B, polymerase(RNA)II (DNA directed) polypeptide E, solute carrier family 34 (sodium phosphate) member 2, syndecan binding protein, endothelial-specific receptor tyrosine kinase, vascular cell adhesion molecule 1. Of note, is that LINE-1 was most often predicted by apolipoprotein D (Apo D) (selected 10.7% of the time) for all three-gene combinations followed by cysteine dioxygenase 1 (Cdo 1) (3.9%), and microsomal glutathione S-transferase 1 (Mgst 1) (2.0%). shows the use of individual gene predictor values for construction of three gene predictor set models. The predominant linkages were between polymerase (RNA) II (DNA directed) polypeptide E, microsomal glutathione S-transferase 1, M12039, syndecan binding protein, and chemokine (C-X-C motif) ligand 1. Genes whose transcription levels exhibited coupling by CoD methodology were hypothesized to be predictive of one another, whether lying upstream or downstream within the biological network, or distributed about the network such that their relation to LINE-1 was only based on chains of interactions among intermediate genes. | Table 1Predictors / L1 target Coefficient of Determination Values Computed by Modeling Methodology. |
Genes identified as predictors exhibited stronger independent correlations to LINE-1 than to non-predictors, as best exemplified by polymerase (RNA)II (DNA directed) polypeptide E, with a coefficient of 0.689108 and M12039 with a coefficient of 0.41982 ( Table 1). Most significant, however, was the identification of predictor sets involving genes which alone exhibited low correlation, but that in combination with each other displayed strong predictive power. This is best exemplified by the chemokine (C-X-C motif) ligand 1 with a CoD of 0.0787 or endothelial-specific receptor tyrosine kinase with a CoD of 0.0132, but in combination with other genes yielding a CoD of 0.5. Thus, CoD methodology identified interactions between LINE-1 and genes within the computational network that would be missed if the analysis was carried out solely on a gene-by-gene basis. Thirty-four transcripts were identified as putative members of the LINE-1 regulatory network and the structure of the network was inferred from the relationship among these elements (). Members of the network included: Agl (amylo-1,6-glucosidase, 4-alpha-glucanotransferase), Ccl2 (chemokine (C-C motif) ligand 2), Cdo1 (cysteine dioxygenase 1, cytosolic), Chi3l3 (chitinase 3-like 3), Clic3 (chloride intracellular channel 3), Col3a1 (procollagen, type III, alpha 1), Cxcl1 (chemokine (C-X-C motif) ligand 1), Cyp2a4 (cytochrome P450, family 2, subfamily a, polypeptide 4), Dbp (D site albumin promoter binding protein), Dnajb9 (DnaJ (Hsp40) homolog, subfamily B, member 9), ELD (extra cellular link domain-containing 1), EST (expressed sequence AI747096), Evi2a (ectopic viral integration site 2a), Gja1 (gap junction membrane channel protein alpha 1), Gnal2 (guanine nucleotide binding protein, alpha 12), Icam1 (intercellular adhesion molecule), Mgst1 (microsomal glutathione S-transferase 1), M12039 (Mouse DNA homologous to the Drosophila per locus), Myl7 (myosin, light polypeptide 7, regulatory), NMO (Mus musculus 10 days neonate medulla oblongata cDNA), Pah (phenylalanine hydroxylase), Pkia (protein kinase inhibitor, alpha), Prot Ty (protein tyrosine phosphatase, receptor type, B), Rgs2 (regulator of G-protein signaling 2), RIKEN cDNA [RIKEN cDNA: 2310032D16], RiKEN [RIKEN cDNA: 1110032A03], RNA Pol II (polymerase(RNA)II (DNA directed) polypeptide E), Rnpc2 (RNA-binding region (RNP1, RRM) containing 2), Sgpl1 (sphingosine phosphate lyase 1), Slc34a2 (solute carrier family 34 (sodium phosphate), member 2), SynBP (syndecan binding protein), TEK (endothelial-specific receptor tyrosine kinase), Vamp3 (vesicle-associated membrane protein 3), VCA (vascular cell adhesion molecule 1). Several genes predominated as predictors of LINE-1 and these relationships were denoted in the connectivity of the network. By interconnecting overlapping sub-networks we built complexity into the interactions, such that any node could have multiple incoming edges and consequently its regulation can depend on complex interactions between input edges. The interactions predicted computationally by the CoD algorithm for murine LINE-1 were examined using a functional genomics approach (). In this experiment, we evaluated the influence of RNAi knockdown of the aryl hydrocarbon receptor (Ahr), a ubiquitous PAS domain transcription factor implicated in LINE-1 inducibility, on the expression of genes within the putative regulatory network. HeLa cells were chosen for analysis to evaluate the integrity of the predicted network structure as well as the conservation of the response between mouse and humans. Cells were cultured under standard conditions and transfected with 50 nM AhR-specific or scrambled siRNA oligonucleotides. In control experiments, RT-PCR measurements of LINE-1 were completed in cells treated with 3uM benzo(a)pyrene (BaP), a ligand of the Ahr known to activate LINE-1 in mammalian cells [ 10]. Also, β-actin, cyp1a1, gapdh or gadd45 mRNAs were measured under silencing conditions to monitor genes lying outside the regulatory network, and expressed either constitutively or following periods of genomic stress by BaP. As expected, siRNA markedly downregulated Ahr expression, while treatment with BaP increased LINE-1, cyp1a1 and gadd45 mRNAs (). Cyp1a1 and gadd45 were included as positive controls for hydrocarbon inducibility of gene expression [ 26]. The stress associated with transfection of non-specific, scrambled oligonucleotide was itself sufficient to induce Ahr and LINE-1 mRNAs, confirming the role of these genes in the integrated cellular response to stress [ 10]. Ahr knockdown markedly suppressed LINE-1 inducibility and modulated the expression of genes within the LINE-1 regulatory network, with marked upregulation of Sgpl, and downregulation of SynBP and Msgt1 observed following Ahr knockdown. These responses were specific for genes within the LINE-1 network since β- actin, gapdh or gadd45 were unchanged by Ahr knockdown. The specificity of the primers used in A-C and the absence of genomic DNA contamination was confirmed in experiments where cDNA was prepared from untreated, scrambled and siRNA-treated cells. Reverse transcriptase (RT) was included only in one positive (NT) control sample, while the enzyme was excluded from all other samples (NT, Scrambled and siRNA) and processed as negative controls in the cDNA mix. The regulation of genes within the LINE-1 network by Ahr knockdown is consistent with the identification of M12039 as a primary node within the network given that M12039 is a mammalian homolog of the Drosophila protein PER, a member of the PAS-domain transcription factors that includes Ahr. These data suggest that PAS proteins lie upstream or within the genetic network of LINE-1, Sgpl, SynBP and MGST1, and established the connectivity of the genes and species conservation of the biological response. LINEs are silenced in somatic, differentiated, non-dividing cells, and transcribed in developing organisms and cells without the need for mobilization to new places in the host genome [ 25]. As such, L1 may normally be tailored for activation at points best suited for their propagation without harm to the host. LINEs are also actively transcribed in plants and animals subjected to stress, implicating retroelements as integral components of the global genomic response to environmental stress and a coping mechanism to adapt to changing environments. LINE-1 activation profiles may correspond to specific patterns of gene expression that can be defined using the CoD algorithm. Patterns of co-regulation within the LINE-1 network may involve common mechanisms of transcriptional control. Although additional experimentation will be required to evaluate molecular mechanisms of co-regulation, siRNA experiments provided evidence for biological connectivity of genes within the computationally-predicted genetic network. Because only a limited number of genes within the network have been examined to date, it is important to consider that patterns of co-expression for all predictor genes within the putative LINE-1 network may not reflect biological connectivity. Genes within the predicted network may in fact exhibit coordinate expression as a function of physical proximity within the genome such that they are transcribed as transcriptional units, or share common trans-regulatory domains [ 30]. To evaluate these possibilities, we investigated the physical proximity of L1Md-A5, a Mus musculus domesticus (Md) element identified as a hydrocarbon inducible clone in mammalian cells [ 10], to the 34 transcripts identified as predictors by CoD analysis. The L1Md transcript used for chromosomal analyses was RNP-58B [GenBank: U15647.1]. Although U15647.1 is nearly identical to a large percentage of the LINE-1 genes in the mouse genome, the 3’ region against which Affymetrix microarray probes were designed is a unique sequence mapping to Chr 4 (). All other 34 transcripts of interest spread over 13 of the mouse chromosomes, with the nearest transcript located ~3.5 million bases away from the specific locus corresponding to U15647.1. This suggests that physical proximity is not likely a major factor contributing to co-expression and that alternate mechanisms must be considered. The mouse L1 sequence contains two open reading frames, ORF1 and ORF2. In the mouse, the 5’ UTR is bipartite, consisting of tandemly repeated units called monomers followed by a non-repeating sequence leading into ORF1. An investigation of the RNP-58B genomic locus revealed features typical of other L1 retrotransposons in mice, most notably, a 745 base region corresponding to 3 2/3 A type monomeric units that are nearly identical to the L1Md-A3.6, now reclassified as L1Md-A5, pattern first reported by this laboratory[ 10], followed by 202 bp of 5’UTR, a 1074 bp ORF1, a 40 bp intergenic region, 3846 base ORF2, followed by a 3’UTR. This sequence, from the beginning of the monomer region through to the end of the mapped transcript, was used as a target sequence to search the genome for nearly identical sequences that may share the same regulatory signature. To search the mouse genome for copies of L1Md that were nearly identical to the target transcript examined here ( U15647.1), it was necessary to construct a putative regulatory region by concatenating the four units L1MdA2 A (140 bases), L1MdA2 C (208 bases), L1MdA2 D (208 bases), and L1MdA2 E (198 bases). The additional sequence in the regulatory region was necessary so that those copies of L1Md with 4 2/3 monomer units could be distinguished from those copies with 3 2/3 monomer units. When compared to the sequence extracted directly from the RNP-58B locus, the sequence alignment begins at base 10 of the putative monomer, which corresponds to base 78 of a 208 base A type monomer. While the significance of truncation of the 2/3s A type monomer is not clear, other examples of truncated A type monomers exist where near identity extends from the 3’ end of the monomer, 5’ toward to bases 69 in one instance, and 86 in another[ 31]. As such, BLAST alignments were considered that begin within ±10 bases of the 78 th base of the target sequence. The evidence for co-regulation due to proximity was sparse, but three examples are worth noting. On Chr 7, two transcripts, Mouse DNA homologous to the Drosophila per locus, [GenBank: M12039, or X02966], and extra cellular link domain-containing 1 [GenBank: AA880988, NM_053247], were located 1.35 Mbases of each another, with the L1Md target sequence located 1.00 Mbases downstream of M12039 and each of the sequences mapping to the minus strand. On Chr 3, two transcripts, vascular cell adhesion molecule 1 [GenBank: U12884] and amylo-1,6-glucosidase, 4-alpha-glucanotransferase [GenBank: AA681807], were located 0.61 Mbases of each another, with the L1Md target sequence located 1.85 Mbases downstream of U12884 and each of these sequences mapping to the minus strand. On Chr4, the target L1Md sequence mapped to an intronic region for transcript X71426 (endothelial-specific receptor tyrosine kinase). However, X71426 mapped to the positive strand, and the target sequence to the negative. The degree to which the features translate into functional relationships is not known. The resolution of LINE-1 regulatory networks is needed to ascertain the impact of repeated DNA sequences on developmental programming and genome stability. LINE-1 elements and their kin in other eukaryotes rearrange genes and chromosomes via a wide variety of mechanisms. In addition, their presence within 5’-regulatory and intronic regions is associated with regulation of methylation status and epigenetic control of gene expression[ 32]. As such, identification of genes within the LINE-1 network is relevant to our understanding of biological complexity and the role of retroelements in human pathogenesis. Members of the LINE-1 regulatory network included genes involved in estrogen signaling, nucleotide excision repair, purine metabolism, G-protein coupled receptors, SAPK/JNK signaling, taurine metabolism and chemokine signaling. The co-determination of genes within the LINE-1 network may involve some level of transcriptional co-regulation and involve PAS homology domain transcription factors. For instance, a novel gene of LINE-1 lineage was identified in murine vascular smooth muscle cells and shown to participate in the mammalian stress response[ 33]. Enhanced LINE-1 gene expression and activated monomer-driven transcription is mediated via a redox-sensitive mechanism by ARE/EpRE-like elements (5′-GTGACTCGAGC-3′) within the A2/3 and A3 region (). ARE/EpRE-like elements were identified by MatInspector in the regulatory region of genes within the LINE-1 network including, Ccl2 (chemokine (C-C motif) ligand 2), Rgs2 (regulator of G-protein signaling 2), Prot Ty (protein tyrosine phosphatase, receptor type, B), Gja1 (gap junction membrane channel protein alpha 1), Pkia (protein kinase inhibitor, alpha), TEK (endothelial-specific receptor tyrosine kinase), ELD (extra cellular link domain-containing 1), Dnajb9 (DnaJ (Hsp40) homolog, subfamily B, member 9), Slc34a2 (solute carrier family 34 (sodium phosphate), member 2), Gnal2 (guanine nucleotide binding protein, alpha 12), Chi3l3 (chitinase 3-like 3), Pah (phenylalanine hydroxylase), RNA Pol II (polymerase(RNA)II (DNA directed) polypeptide E). Of note, Ahr has been shown to be present within the macromolecular complex that assembles on ARE/EpREs [ 34]. Constitutive LINE-1 retrotransposon expression in vivo is limited almost exclusively to germ line and embryonic cells [ 35, 36]. The factors that determine germ line specificity of LINE-1 expression have not been identified, but members of the SOX family of transcription factors and the ubiquitous nuclear transcription factor YY1 have been implicated [ 35]. LINE-1 elements are heavily methylated in normal somatic cells and the large majority of 5-methylcytosine (m 5C) in the genome actually lies within repetitive elements, including LINE-1 [ 37]. Induction of LINE-1 has been observed in embryonic carcinoma cells, testicular germ line tumors and ovarian carcinomas [ 38- 40], and to a lesser degree in other tumors [ 26, 41]. The underlying principle driving creation of the LINE-1 network was that pair-wise interactions connecting genes can be conceptualized as “nodes” connected to each other by “edges”, such that the edges represent the interactions between any two components. A gene in and of itself may not be highly correlated to a target, but in combination with other genes may be predictive of the behavior of the target. When used individually by the computational algorithm each predictor gene exhibited a CoD measure that was comparable to the value obtained using a linear correlation coefficient model. In contrast, CoD detected multivariate non-linear influences on gene expression within complex genetic networks and enabled calculation of a value that mathematically reflected interactions among multiple predictors. Transposons may provide a selective evolutionary advantage to cells or instead, participate actively in cycles of genomic assault by insertional mutagenesis. Approximately 26% of the sequence on the X chromosome is LINE-1, whereas the average among the other autosomal chromosomes is 13% [ 42]. Actual genomic target sites for insertion have not been defined with certainty, and the extent to which LINE-1 modulates cell biology via non-insertional mechanisms remains undefined. A key unanswered question is the extent to which endogenous reverse transcriptase activity in the host cell influences LINE-1 retrotransposition. LINE-1 activation events may contribute to the appearance of new phenotypes and play a central role in human and animal pathogenesis. As such, activation of LINE-1 may be a critical step in the adaptive response of the genome to environmental stress. The concept that LINE-1 functions are integral to organismal biology and under genetic regulation is intriguing, but speculative at this time. Thus, the existence of the network must be approached with caution as it may be inconsistent with retrotransposon biology and the dynamics of their evolution. Because retroelements exist in a repressed state under most conditions, and yet connectivity for genes in the network was established, network genes may either be functionally linked to host silencing mechanisms, or simply be co-incident with periods of LINE-1 activity. The latter possibility would predict that the positive regulation of L1NEs by external factors involves disruption of the very same mechanisms engineered toward silencing of retrotransposon activity. If instead, the relationship is co-incident, L1 itself may be tailored for activation at time points that are synchronized to the expression of genes within the computationally-derived network. In this scenario, coordinate gene expression may be coupled to periods of transcription factor availability that can be resolved by COD methodology. Future studies must be designed to examine these interesting hypotheses. The fundamental biological mechanisms governing retrotransposition and the complexity of regulatory networks involved in the regulation of this process are not yet clear and continue to be debated. This is partly due to the elusive nature of the transposed elements, the low frequency of transposition in normal cells and a lack of understanding of the genes that participate in the process. Taking advantage of computational methodologies created to study regulatory networks in cancer [ 43], we have identified genes predictive of LINE-1 expression. The algorithm is based on the Boolean formalism as a framework to define the complex and dynamical network of LINE-1. Data Filtering Affymetrix DAT files were analyzed using Affymetrix MAS 5.0 and a 500 scaling factor for all arrays. The CEL files were input into RMA Express [ 44] where signal calculation, background adjustment, quantile normalization and log transformation were completed and exported to a text file. The objective was to identify highly conserved biological interactions of the mouse L1 using Mus musculus domesticus (ORF1) and reverse transcriptase (ORF2) gene [GeneBank: U15647] as biological targets irrespective of genomic context. Computational Methodology The coefficient of determination (CoD) algorithm employed was based on the co-expression of genes coupled to probabilistic relationships that identify gene sets predictive of putative biological interactions. A completed description of the computational methodology was published elsewhere [ 45], also see Supplement 5. LINE-1 Network Plots The relationships predicted by the CoD algorithm were illustrated using the Interaction Explorer™ Software (Stratagene Corporate, La Jolla, CA). The LINE-1 predictor gene pathways were constructed upon examination of common regulators for the 34 genes within the network and the shortest link between those genes. Each of 22 combinations of three predictors was linked to the LINE-1 target and the schematics overlaid to form a linkage diagram that depicts the relationship of LINE-1 target gene to predictor genes, as well as connections among predictor genes. The final product depicts the all inclusive hypothetical interrelationships among all predictor genes and LINE-1. LINE-1 Predictor Gene Annotations Different programs were utilized for the annotation of predictor genes. The Database for Annotation, Visualization and Integrated Discovery 2.0 (DAVID 2.0) from the National Institute of Allergy and Infectious Disease was used to acquire annotation information for predictor genes and to integrate functional genomic annotations with intuitive graphical summaries (see Supplement 1). Mapping of Predictor Genes Gene accessions were mapped to the mouse genome using Genbank annotation. The Entrez Gene ID corresponding to the accession from the gene2accession file was retrieved. The gene2accession file was downloaded from Genbank ( ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2accession.gz). Using the Entrez Gene ID, the gene’s chromosomal location was identified from the file seq_gene.md downlodaded from Genbank ( ftp://ftp.ncbi.nih.gov/genomes/M_musculus/mapview/seq_gene.md.gz). This file was searched for records where the gene Id is the feature_id, the feature type is “GENE”, and the group_label was not “C57BL/6J”. Twenty two of the 34 accessions could be found in the gene2accession file. For the remaining 12, unigene clusters were identified. Accessions were found within the unigene cluster to which the accession of interest mapped, with coverage over the entire length of the sequence of interest. The accessions produced by the CoD analysis, are listed with those that were ultimately used in the mapping process in Table 2, along with their mapping information to build 35 of the mouse assembly. | Table 2Location of transcripts related to the expression of L1MD by CoD analysis. |
Construction of Target Sequence The target sequence was extracted directly from the locus to which the transcript reported in Genbank accession U15647.1 (GI: 558906) was mapped. This accession’s sequence mapped to chromosome 4 in the reverse orientation to position 9848137-9854979 with 99.9% identity. The target ultimately spanned bases 9848137 to 9855624 of chromosome 4 from the mouse genome, build 35. The sequence was reverse complimented to match the orientation of the transcript. The target sequence and the alignments to monomer, transcript, and microarray probe sequences are presented in Supplement 2. Genome BLAST Searches The genome sequence used for sequence analysis was Build 35 of the mouse genome from the Mouse Genome Sequencing Consortium. The sequence searches were performed for L1Md reference against the mouse genome sequence using the Washington University BLASTN1 program. Monomer Sequence Analysis The A type monomer sequence data used for analyzing the 5’ monomer region of the target sequence, as well as for padding the target sequence was transcribed from Loeb (1986). The monomer elements used in this study correspond to L1MdA2 A, L1MdA2 C, L1MdA2 D, and L1MdA2 E. These sequences were included in fasta format in Supplement 3. Matching Criteria The match began within ± 10 bases of base 78 of the target sequence, since matches starting prior to that are likely those containing 4 2/3s copy of the monomer. As similarity in the regulatory region is critical to the assertion made here, the match needed to be 98% identical in the 745 bases of the monomer region, bases 78-822 of the target. Additionally, the entire target sequence from the start of monomer region through to the end of ORF2 was considered, bases 78-5984 of the target. It has been demonstrated in human LINE-1 that a high degree of similarity exists amongst active copies[ 31]. In those studies a 98% threshold was used to identify potentially active copies of L1[ 31, 32]. Therefore, the same criteria was applied such that there must be a contiguous match from base 78 (±10) through base 5984 of the target sequence. Sequence Manipulation Basic sequence manipulation was carried out using methods from the biojava framework for processing biological data. siRNA silencing of Ahr HeLa cells (2 × 104/cm2) were seeded in 6 well plates and 24 hrs later transfected with siRNA targeting exon 5 of the Ahr using Lipofectamine 2000™ (Invitrogen, Carlsbad, CA). Control cells were transfected with scrambled siRNA to monitor non-specific changes in gene expression. After 48 hr, cells were either harvested or treated with BaP or vehicle (DMSO) and analyzed for patterns of mRNA expression via qRT-PCR. Ahr protein knockdown was confirmed via western blotting of samples in parallel experiments. Semi-Quantitative Reverse Transcriptase PCR (qRT-PCR) Total RNA was extracted using Triazol reagent (Invitrogen), quantified and treated with DNAse to remove residual DNA contamination. Total RNA (1ug) was used for cDNA synthesis followed by 30 cycles of PCR using primers directed at selected targets as indicated in Supplement 4. PCR products were resolved on a 1% agarose gel. 01: Supplement 1 DAVID Summary on L1 CoD genes.htm. The DAVID summary provides Information from the Gene Ontology (GO) classifications and KEGG, BioCARTA pathways. 02: Supplement 2 AlignmentOutputData.txt. The L1 target sequence and the alignments to monomer, transcript, and microarray probe sequences are shown. 03: Supplement 3 PromoterSeq.doc. The monomer elements’ sequences used in this study are shown in fasta format. The monomer elements correspond to L1MdA2 A, L1MdA2 C, L1MdA2 D, and L1MdA2 E. 04: Supplement 4 PCR primer sequences for selected targets. 05: Supplement 5 Supplementary details on Methodology The work was supported in part by research grants ES 04849 to Kenneth Ramos. The authors are grateful to Eric Rouchka for acquiring and preparing the Mouse genome sequence. 1. Lander ES, et al. Initial sequencing and analysis of the human genome. Nature. 2001;409:860–921. [PubMed] 2. Waterston RH, et al. Initial sequencing and comparative analysis of the mouse genome. Nature. 2002;420:520–562. [PubMed] 3. Kazazian HHJ, Goodier JL. LINE drive retrotransposition and genome instability. Cell. 2002;110:277–280. [PubMed] 4. Han JS, Szak ST, Boeke JD. Transcriptional disruption by the L1 retrotransposon and implications for mammalian transcriptomes. Nature. 2004;429:268–274. [PubMed] 5. Tremblay A, Jasin M, Chartrand P. A double-strand break in a chromosomal LINE element can be repaired by gene conversion with various endogenous LINE elements in mouse cells. Mol Cell Biol. 2000;20:54–60. [PubMed] 6. Kazazian HHJ, et al. Haemophilia A resulting from de novo insertion of L1 sequences represents a novel mechanism for mutation in man. Nature. 1988;332:164–166. [PubMed] 7. Kidwell MG, Lisch D. Transposable elements as sources of variation in animals and plants. Proc Natl Acad Sci. 1997;94:7704–7711. [PubMed] 8. Nangia-Makker P, et al. Galectin-3 and L1 retrotransposons in human breast carcinomas. Breast Cancer Res Treat. 1998;49:171–183. [PubMed] 9. Suter CM, Martin DI, Ward RL. Hypomethylation of L1 retrotransposons in colorectal cancer and adjacent normal tissue. Int J Colorectal Dis. 2004;19:95–101. [PubMed] 10. Lu KP, Ramos KS. Redox Regulation of a Novel L1Md-A2 Retrotransposon in Vascular Smooth Muscle Cells. J Biol Chem. 2003;278:28201–28209. [PubMed] 11. Tchenio T, Casella JF, Heidmann T. Members of SRY family regulate the human LINE retrotransposons. Nucleic Acid Res. 2000;28:411–415. [PubMed] 12. Athanikar JN, Badge RM, Moran JV. A YY1-binding site is required for accurate human LINE-1 transcription initiation. Nucleic Acids Res. 2004;32:3846–3855. [PubMed] 13. Kolosha VO, Martin SL. High-affinity, non-specific RNA binding by the open reading frame 1 (ORF1) protein from long interspersed nuclear element 1 (LINE-1). Journal of Biochemistry. 2003;278:8112–8117. 14. Feng Q, Moran JV, Kazazian HHJ, Boeke JD. Human L1 retrotransposon encodes a conserved endonuclease required for retrotransposition. Cell. 1996;87:905–916. [PubMed] 15. Mathias SL, Scott AF, Kazazian HH, Boeke JD, Gabriel A. Reverse transcriptase encoded by a human transposable element. Science. 1991;254:1808–1810. [PubMed] 16. Fanning TG, Singer MF. The LINE-1 DNA sequences in four mammalian orders predict proteins that conserve homologies to retrovirus proteins. Nucl Acids Res. 1987;15:2251–2260. [PubMed] 17. Jordan IK, Rogozin IB, Glazko GV, Koonin EV. Origin of a substantial fraction of human regulatory sequences from transposable elements. Trends Genet. 2003;19(2):68–72. [PubMed] 18. Thornburg BG, Gotea V, Makalowski W. Transposable elements as a significant source of transcription regulating signals. Gene. 2006;365:104–110. [PubMed] 19. Peaston AE, et al. Retrotransposons regulate host genes in mouse oocytes and preimplantation embryos. Dev Cell. 2004;7(4):597–606. [PubMed] 20. Shapiro JA. Retrotransposons and regulatory suites. BioEssays. 2005;27:122–125. [PubMed] 21. Grandbastien MA. Activation of plant retrotransposons under stress conditions. Trends Plant Sci. 1998;3(5):181–187. 22. Hagan CR, Rudin CM. Mobile genetic element activation and genotoxic cancer therapy: potential clinical implications. American Journal of Pharmacogenomics. 2002;2(1):25–35. [PubMed] 23. Banerjee G, Gupta N, Tiwari J, Raman G. Ultraviolet-induced transformation of keratinocytes: possible involvement of long interspersed element-1 reverse transcriptase. Photodermatology, Photoimmunology & Photomedicine. 2005;21(1):32–39. 24. Kuchen S, et al. The L1 retroelement-related p40 protein induces p38delta MAP kinase. Autoimmunity. 2004;37(1):57–65. [PubMed] 25. Arnholdt-Schmitt Birgit. Stress-Induced Cell Reprogramming. A Role for Global Genome Regulation? Plant Physiol. 2004;136(1):2579–2586. [PubMed] 26. Stribinskis V, Ramos KS. Activation of human long interspersed nuclear element 1 retrotransposition by benzo(a)pyrene, an ubiquitous environmental carcinogen. Cancer Res. 2006;66:2616–2620. [PubMed] 27. Morales JF, Snow ET, Murnane JP. Environmental factors affecting transcription of the human L1 retrotransposon. I. Steroid hormone-like agents. Mutagenesis. 2002;17(3):193–200. [PubMed] 28. Schiff R, Itin A, Keshet E. Transcriptional activation of mouse retrotransposons in vivo: specific expression in steroidogenic cells in response to trophic hormones. Genes Dev. 1991;5(4):521–532. [PubMed] 29. Trelogan SA, Martin SL. Tightly regulated, developmentally specific expression of the first open reading frame from LINE-1 during mouse embryogenesis. Proc Natl Acad Sci USA. 1995;92:1520–1524. [PubMed] 30. Yamashita T, et al. Genome-wide transcriptome mapping analysis identifies organ-specific gene expression patterns along human chromosomes. Genomics. 2004;84(5):867–875. [PubMed] 31. Brouha B. Hot L1s account for the bulk of retrotransposition in the human population. PNAS. 2003;100(9):5280–5285. [PubMed] 32. Badge RM, Alisch RS, Moran JV. ATLAS: a system to selectively identify human-specific L1 insertions. Am J Hum Genet. 2003;72(4):823–838. [PubMed] 33. Lu KP, Ramos KS. Identification of genes differentially expressed in vascular smooth muscle cells following benzo(a)pyrene challenge: implications for chemical atherogenesis. Biochem Biophy Res Commun. 1998;253:828–833. 34. Chen YH, Ramos KS. A CCAAT/enhancer-binding protein site within antioxidant/electrophile response element along with CREB-binding protein participate in the negative regulation of rat GST-Ya gene in vascular smooth muscle cells. J Biol Chem. 2000;275(35):27366–27376. [PubMed] 35. Trelogan SA, Martin SL. Tightly regulated, developmentally specific expression of the first open reading frame from LINE-1 during mouse embryogenesis. Proc Natl Acad Sci. 1995;92:1520–1534. [PubMed] 36. Ergun S, et al. Cell type-specific expression of LINE-1 ORF1 and ORF2 in fetal and adult human tissues. Journal of Biochemistry. 2004;278:27753–27763. 37. Luedi PP, Hatemink AJ, Jirtle RL. Genome-wide prediction of imprinted murin genes. Genome Res. 2006;15:875–884. [PubMed] 38. Skworonski J, Fanning TG, Singer MF. Unit-length line-1 transcripts in human teratocarcinoma cells. Mol Cell Biol. 1988;8:1385–1397. [PubMed] 39. Bratthauer GL, Fanning TG. Active LINE-1 retrotransposons in human testicular cancer. Oncogene. 1992;7:507–510. [PubMed] 40. Menendez L, Benigno BB, McDonald JF. L1 and HERV-W retrotransposons are hypomethylated in human ovarian carcinomas. Mol Cancer. 2004;3:12. [PubMed] 41. Florl AR, Lower R, Schmitz-Drager BJ, S WA. DNA methylation and expression of LINE-1 and HERV-K provirus sequences in urorethelial and renal cell carcinomas. British Journal of Cancer. 1999;80:1312–1321. [PubMed] 42. Prak ET, Kazazian HH., Jr Mobile elements and the human genome. Nat Rev Genet. 2001;2(1):78. 43. Shmulevich I, ER D, Zhang W. From Boolean to probabilistic Boolean networks as models of genetic regulatory networks. Proceedings of the IEEE. 2002;90:1778–1792. 44. Bolstad BM, Irizarry RA, Astrand M, Speed TP. A Comparison of Normalization Methods for High Density Oligonucleotide Array Data Based on Bias and Variance. Bioinformatics. 2003;19(2):185–193. [PubMed] 45. Johnson CD, et al. Unraveling Gene–Gene Interactions Regulated by Ligands of the Aryl Hydrocarbon Receptor. Environmental Health Perspectives: Toxicogenomics. 2004;112(4):403–412.
|
The publisher's final edited version of this article is available at 
