SURVEY AND SUMMARY Does SINE evolution preclude Alu function?

The evolution, mobility and deleterious genetic effects of human Alus are fairly well understood. The complexity of regulated transcriptional expression of Alus is becoming apparent and insight into the mechanism of retrotransposition is emerging. Unresolved questions concern why mobile, highly repetitive short interspersed elements (SINEs) have been tolerated throughout evolution and why and how families of such sequences are periodically replaced. Either certain SINEs are more successful genomic parasites or positive selection drives their relative success and genomic maintenance. A complete understanding of the evolutionary dynamics and significance of SINEs requires determining whether or not they have a function(s). Recent evidence suggests two possibilities, one concerning DNA and the other RNA. Dispersed Alus exhibit remarkable tissue-specific differences in the level of their 5-methylcytosine content. Differences in Alu methylation in the male and female germlines suggest that Alu DNA may be involved in either the unique chromatin organization of sperm or signaling events in the early embryo. Alu RNA is increased by cellular insults and stimulates protein synthesis by inhibiting PKR, the eIF2 kinase that is regulated by double-stranded RNA. PKR serves other roles potentially linking Alu RNA to a variety of vital cell functions. Since Alus have appeared only recently within the primate lineage, this proposal provokes the challenging question of how Alu RNA could have possibly assumed a significant role in cell physiology. 1. CONTINUOUS SUCCESSION OF DISTINCT SINES Mammalian DNAs typically contain hundreds of thousands of copies of short interspersed repeated sequences called SINEs (1). The number of SINEs that are fixed in the mammalian genome is all the more remarkable when it is recognized that these sequences transposed into their genomic loci through RNA intermediates (retrotransposition). Thus SINEs must have been a tremendous source of insertional mutagenesis throughout mammalian evolution. Given their abundance and mobility, evolutionary considerations have naturally dominated research on SINEs. Results from those studies provide the starting point for considering other aspects of SINEs, including their possible functionality. Excellent reviews, including a recent monograph edited by Maraia, document generally accepted background information (Sections 1–5). The most extensively studied mammalian SINE, human Alu, exemplifies most features of this unusual class of sequences. There are nearly 1 000 000 Alus per haploid genome (1), corresponding to an average genomic spacing of 3 kb. Individual Alus share a 282 nt consensus sequence which is typically followed by a 3′ A-rich region resembling a poly(A) tail (Fig. 1). The Alu consensus sequence is a divergent tandem dimer in which the two monomer units are separated by a short A-rich region, a vestige of what must have been a 3′ A-rich region that flanked the ancestral monomer (1,2; Fig. 1). Except for a 30 nt insertion in the right monomer, Alu monomers are homologous to SRP RNA, also known as 7SL RNA (Fig. 1). Most Alus are flanked by short direct repeats which are the duplicated insertion site (1; Fig. 1). Except for rodents and primates, SINEs in all other animals examined are unrelated to SRP RNA but are instead homologous to tRNAs; even plants contain tRNA SINEs, indicating that the earliest eukaryotic SINEs must have been derived from tRNAs (1,3; Fig. 2). Moreover, all highly repetitive eukaryotic SINEs belong to either the SRP or a tRNA superfamily. (Different tRNA superfamilies are not distinguished here; 3.) Rodents contain both SRP RNA and tRNA related SINEs, usually called B1 and B2 repeats respectively (1,4; Fig. 2). Rodent B1 repeats essentially resemble the left human Alu monomer (Figs 1 and 2). Prosimian SINEs include full-length dimeric Alus, B1-like Alu monomers, B2-like/tRNA SINEs and composite elements consisting of both SINE superfamilies. This intermediate composition suggests a transition between the SINEs in rodents and higher primates (1; Fig. 2). Sequence analysis indicates that rodent B1 and primate Alu repeats are ultimately derived from a single founder (5). As discussed later (Section 4), human Alu subfamilies also result from individual founders. The question of whether the very earliest tRNA SINEs might have been rooted in a single primordial ancestor (Fig. 2) remains to be answered (Section 11). The most recent common ancestor of human and rodent must have contained tRNA SINEs (Fig. 2). Decrepit, fossil tRNA *Tel: +1 530 752 3003; Fax: +1 530 752 3085; Email: [email protected] Nucleic Acids Research, 1998, Vol. 26, No. 20 4542 Figure 1. Consensus Alu structure (1). Direct repeats (dr) flanking an Alu result from duplication of its empty genomic insertion site. Similarly, 5′ and 3′ sequences flanking an Alu are contributed by the unique genomic locus in which it resides. Four or more T residues are sufficient to terminate pol III-directed transcription so that termination and length of the primary Alu transcript are determined by its unique 3′ flanking sequence. The consensus Alu sequence is usually followed by an A-rich region resembling a poly(A) tail. As depicted by the two solid arrows, the 282 nt Alu consensus sequence consists of an inexact duplication of two monomer units which are homologous to SRP RNA. These two monomers are separated by a mid A-rich region and, as depicted by the wavy line, the right monomer contains an additional sequence (∼30 nt) that is absent in the left monomer. Rodent B1 repeats essentially resemble left Alu monomers. SINEs, called MIRs, that pre-date rodent–human divergence are buried in human DNA (1,6,7; Fig. 2). Sequence database searches, hybridization analysis and library screening with rodent B2 probes have failed to identify other tRNA SINEs in human DNA (6–8; unpublished results). Thus all available evidence, albeit negative, indicates that the previously successful mammalian tRNA SINEs are now either extinct or severely reduced in copy number within the human genome. The reasons why Alus flourished while tRNA SINEs died in the higher primate genome are unknown. Perhaps Alus are merely better genomic parasites than tRNA SINEs since ‘No cellular function...is required to explain...the behavior or persistence of middle repetitive sequences as a class’ (9–11). However, there is no a priori reason to dismiss the possibility that Alus provide a selective advantage to their host which drives their retrotranspositional success. While some evidence suggests that Alus may serve one or more functions (Sections 8, 10 and 11), the two explanations are not exclusive, since a successful parasite optimizes its requirements with those of the host. As extraordinarily successful genomic symbionts, Alus may have established a state of nearly complete neutrality, a ‘genomic peace’ (Section 3), or may instead compensate their host with selective advantages. In the extreme, Alus could serve a vital function that precedes their genomic proliferation. Competing themes throughout this review are how either the retrotransposition pathway or the host’s requirements might select for SINEs. 2. SINE FAMILIES GROW BY ACCUMULATING NEW MEMBERS To understand the dynamics of SINE evolution, the fate of human Alus has been traced by comparisons with their orthologs in other primates (1). (Primate phylogeny is qualitatively depicted in Fig. 2 to follow this and subsequent discussions.) These comparisons, which emphasize Alus mapping within globin gene clusters, indicate that the great majority of human Alus post-date the Figure 2. SINE phylogeny and accumulation within a human DNA (1). The straight time line on the left qualitatively traces the evolutionary history and sequence of events by which human Alus and SINEs accumulated in the DNA from a particular tissue from a particular person. Solid arrows with names in bold indicate the times at which SINE families appeared, and open arrows with names in italic indicate particular human SINEs as inferred from phylogenetic comparisons. Branching compares either definite species or hypothetical populations, individuals and tissues as outgroups (lower case) to trace the lineage of accumulated SINEs and human Alus within a particular tissue from a particular person (upper case). Although most of the branchings depicted have been established by comparing orthologous loci in outgroups, common ancestry has not been established for inheritance of either the tRNA superfamily in plants and animals or the SRP superfamily in rodent and primate. In principle either of these two events could have resulted from convergent evolution and not common inheritance, as assumed for simplicity in this tree. Somatic insertion of the Mlvi-2 Alu has not been proven. divergence of the prosimian lineage but pre-date chimpanzee divergence (Fig. 2). Some older Alus even pre-date human– monkey divergence (Fig. 2). Alus are rarely subject to sequence conversion (12) and, except for the special case of CpG dinucleotides, accumulate point mutations at the rate expected for unselected DNA sequences (1,2). DNA methylation accelerates the mutation rate of Alu CpG dinucleotides (1,2,13,14; Sections 6 and 7). In summary, phylogenetic comparisons generally indicate that Alus are rather immobile, are stably inherited over relatively long evolutionary times and, like the fossil SINE example, are eventually obliterated by accumulated point mutations (Fig. 2).