RESUMEN
Most Arabidopsis long interspersed elements (LINEs, called ATLNs) have two open reading frames, orf1 and orf2. In the 5' untranslated regions (UTRs) located upstream of orf1, the most proximal segments of tens of base pairs long are not homologous even in two ATLN members with almost identical sequences. In this study, we first show that RT-PCR products from ATLN39, a member of ATLN, can be detected only in total RNA from the hypomethylation mutant ddm1 or from suspension-cultured cells treated with a DNA methylation inhibitor 5-azacytidine, indicating that the expression of ATLN39 is negatively regulated by DNA methylation. We then show that orf1 fused in frame with the luciferase (luc) gene is expressed in suspension-cultured cells of A. thaliana when the 5' UTR is present in the region upstream of orf1. Analysis of deletion in the 5' UTR revealed that the 5' UTR has two promoters, designated here as P1 and P2. Analysis of transcripts by 5' RACE showed that their 5' ends were located at sites immediately upstream of the P1 region or at sites downstream of the P2 region. This observation and the fact that the P1 region contains no TATA sequence indicate that P1 is an internal promoter that initiates transcription from sites upstream of the promoter. A sequence containing GGCGA with a CpG methylatable site is conserved in the P1 regions in members closely related to ATLN39. The P2 region, however, contains the TATA sequence as well as another sequence with a CpG site. The TATA sequence is conserved in members closely related to ATLN39 but not in the other ATLN members, suggesting that P2 is the promoter uniquely present in the ATLN39-related members. Transcripts from promoter P1 can be used as templates to give new copies proficient in retroposition, but those from promoter P2 cannot because of the lack of the proximal half region of the 5' UTR sequence. Transcripts from promoter P2, as well as those from promoter P1 can, however, be used for the production of a sufficient amount of proteins for retroposition. Only a short sequence of the non-homologous region is present at the 5' ends of transcripts from promoter P1, thus suggesting that the non-homologous regions seen in the most proximal regions in ATLN elements are not generated in transcription.
