Systematic variation in mRNA 30-processing signals during mouse spermatogenesis

Gene expression and processing during mouse male germ cell maturation (spermatogenesis) is highly specialized. Previous reports have suggested that there is a high incidence of alternative 30-processing in male germ cell mRNAs, including reduced usage of the canonical polyadenylation signal, AAUAAA. We used EST libraries generated from mouse testicular cells to identify 30-processing sites used at various stages of spermatogenesis (spermatogonia, spermatocytes and round spermatids) and testicular somatic Sertoli cells. We assessed differences in 30-processing characteristics in the testicular samples, compared to control sets of widely used 30-processing sites. Using a new method for comparison of degenerate regulatory elements between sequence samples, we identified significant changes in the use of putative 30processing regulatory sequence elements in all spermatogenic cell types. In addition, we observed a trend towards truncated 30-untranslated regions (30UTRs), with the most significant differences apparent in round spermatids. In contrast, Sertoli cells displayed a much smaller trend towards 30-UTR truncation and no significant difference in 30-processing regulatory sequences. Finally, we identified a number of genes encoding mRNAs that were specifically subject to alternative 30-processing during meiosis and postmeiotic development. Our results highlight developmental differences in polyadenylation site choice and in the elements that likely control them during spermatogenesis. INTRODUCTION Messenger RNA cleavage and polyadenylation is a necessary processing step for almost all eukaryotic mRNA transcripts. The mechanisms involved in 30-processing have been studied in detail over the last 30 years, and the identities and roles of most of the proteins involved have been characterized (1,2). Comparative studies of the mechanisms and of the proteins involved in several organisms have shown a high level of primary sequence conservation between the homologous genes across a broad range of organisms (1,3). A number of cis-acting RNA sequences involved in the regulation, selection and processing of the 30 ends of mammalian mRNA transcripts have been previously described [Figure 1, (1,4–6)]. By convention, these sequence elements are described as being (50) or (30) relative to the site of poly(A) addition. The most widely observed sequence, the polyadenylation signal (PAS), was identified 30 years ago (7), and is usually the canonical hexamer, AAUAAA, or its most common variant, AUUAAA; however, a number of recent studies have indicated that this element is more variable than originally thought (8,9). The PAS is usually 15–30 nt upstream of the site of poly(A) addition (1). Early studies of the downstream element indicated the presence of two distinct elements (10,11), though subsequent studies [e.g. (12–14)] resulted in an ambiguous picture, such that the downstream elements were jointly referred to as U-/UG-rich. We recently completed a computational study of large sets of 30-processing sites in 10 different metazoans (5) that confirmed the presence of two separate elements, a proximal UG-rich element (typically 6–10 nt downstream of the 30-processing site), and a more distal U-rich element (typically 15–30 nt downstream of the 30-processing site). Our analysis also confirmed that these elements are highly degenerate, to the extent that functional 30-processing sites often lack strong matches to one or both elements. *To whom correspondence should be addressed. Tel: +1 207 288 6847; Fax: +1 207 288 6073; Email: [email protected] The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors