br Materials and methods br Results br Discussion I

Materials and methods
Results
Discussion I have identified an alternative transcript of the integrin αE gene that is abundantly expressed in testis. In humans, this transcript (hAED) includes part of intron 26 and exons 27, 29, 30, and 31 of the conventional αE gene and Northern analysis indicates that a similar transcript (mAED) exists in mice. Neither the hAED transcript nor the equivalent region in the murine gene contain an ATG translation start codon in the same reading frame as the conventional αE coding sequence. So, it appears unlikely that a truncated form of the αE integrin would be encoded in either species. A potential coding sequence of 30 D-erythro-Sphingosine (synthetic) with a suitable Kozak sequence does exist in the hAED transcript, but an anti-serum to this peptide failed to detect a protein of the appropriate size in tissues where hAED RNA is abundantly expressed. Furthermore, the equivalent mAED transcript could not encode a similar protein. Clearly, these results do not rule out translation of the hAED and mAED RNAs, but it seems unlikely that a functional protein product is produced. Some poly(A)+ RNA molecules are thought to function as non-coding RNAs. For example, the Xist and the antisense Tsix RNAs are involved in regulation of X chromosome inactivation (Lee and Lu, 1999). The H19 and insulin-like growth factor-2 receptor antisense (Igf2r AS) genes also encode non-translated RNAs. These genes are implicated in genomic imprinting of the insulin-2/insulin-like growth factor-2 genes and Igf2r gene, respectively, although the role of the RNA molecules themselves remains unclear (Tilghman, 1999). Interestingly, non-coding RNAs (Pgc and the Xlsirts) that are critical for germ cell establishment have been characterized in Drosophila and Xenopus oocytes (Kloc et al., 1993, Nakamura et al., 1996). It is possible that hAED/mAED may also function as a non-coding RNA. In addition, transcription of the hAED/mAED RNA may be related to its proximity to another gene within the αE locus. In both humans and mice, the previously undescribed gene for haspin is located within the same intron of the integrin αE gene close to the initiation site of hAED transcription. Such intronic genes are rare in group III introns, but not unprecedented. Indeed, large introns (of 32, >40, and 64 kb) in the human Factor VIII, neurofibromatosis type 1 and thyroglobulin genes contain unrelated genes (Levinson et al., 1992, Xu et al., 1990, Meijerink et al., 1998). However, it is remarkable that intron 26 of the αE gene, which is less than 4.4 kb in humans and mice, harbors a gene that itself covers 2.8 kb. In fact, the 5′ end of hAED is less than 70 bp away from the 5′ end of the haspin gene, and the two transcripts are generated in a head-to-head orientation from opposing DNA strands (see Fig. 2, Fig. 8). Several examples of head-to-head genes are found in higher eukaryotes. In many cases such bi-directional promoters allow co-ordinated expression of related genes such as collagen IV α1 and α2 (Pöschl et al., 1988). Indeed, both haspin and hAED/mAED are most strongly expressed in the testis, suggesting that the two transcripts are generated concurrently. However, the two RNAs are clearly not always co-expressed. In fact, hAED is widely expressed in tissues where haspin is not expressed, particularly in muscle, and haspin can be transcribed where mAED cannot be detected, such as in MTC-1 cells. Other genes that are in close head-to-head orientation yet can be expressed differentially have been described (Linton et al., 1989, Schilling and Farnham, 1994, Brenner et al., 1997). It is possible that the hAED RNA plays a role in regulation of haspin gene expression but how this would occur is unclear. It is conceivable that transcription of hAED upstream of the haspin gene might facilitate maintenance of an open chromatin structure at the 5′ end of the haspin gene, or prevent CpG methylation (see below). It is notable that the promoter region of the haspin and hAED/mAED genes is within a CpG island. Such islands are closely correlated with the 5′ ends of most housekeeping genes, and are associated with many tissue-specific genes. The promoter activity of many genes is dependent on hypomethylation of 5′ CpG islands, and it is possible that CpG islands also serve as origins of DNA replication (Delgado et al., 1998). The Xist, Tsix, H19 and Igf2r AS genes are all associated with 5′ CpG islands or ‘differentially methylated regions’, and DNA methylation is clearly involved in regulation of imprinting (Tilghman, 1999). Since global denovo methylation takes place during mammalian gametogenesis and embryonic development, mechanisms to protect CpG islands from methylation must exist. It has recently been proposed that active transcription from CpG promoters in germ cells and early embryonic cells may be required to prevent DNA methylation (Macleod et al., 1998). Consistent with this, even CpG islands not obviously associated with the 5′ ends of genes are found to have promoter activity in testis and embryonic cells. For example, CpG islands in introns of the mouse pro-opiomelanocortin and MHC class II I-Aβ genes contain such promoters that drive transcription of RNAs derived from the 3′ ends of the two genes, (Gardiner-Garden and Frommer, 1994, Macleod et al., 1998). The finding that the CpG island in intron 26 of the αE gene also contains a promoter that is active in these tissues lends support to this model. Additionally, it has been noted that many bi-directional promoters are associated with CpG islands (Lavia et al., 1987) and this is true, for example, of the collagen IV α1 and α2 genes (Pöschl et al., 1988) and the F8A and F8B transcripts that originate from intron 22 of the human Factor VIII gene (Levinson et al., 1992), as well as the haspin and hAED/mAED genes. Perhaps bidirectional transcription from CpG islands is particularly effective in preventing DNA methylation.