MBE Advance Access originally published online on October 12, 2005
Molecular Biology and Evolution 2006 23(2):235-239; doi:10.1093/molbev/msj034
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Article |
A Test of the Master Gene Hypothesis for Interspersed Repetitive DNA Sequences
Institute of Genetics, University of Nottingham, Queens Medical Centre, Nottingham, United Kingdom
E-mail: john.brookfield{at}nottingham.ac.uk.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Many families of interspersed repetitive DNA elements, including human Alu and LINE (Long Interspersed Element) elements, have been proposed to have accumulated through repeated copying from a single source locus: the "master gene." The extent to which a master gene model is applicable has implications for the origin, evolution, and function of such sequences. One repetitive element family for which a convincing case for a master gene has been made is the rodent ID (identifier) elements. Here we devise a new test of the master gene model and use it to show that mouse ID element sequences are not compatible with a strict master gene model. We suggest that a single master gene is rarely, if ever, likely to be responsible for the accumulation of any repeat family.
Key Words: SINE • repetitive DNA • master gene • ID element • BC1
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Short Interspersed Elements and the Host Genome
Interspersed repetitive DNA families are very abundant, particularly in mammals, where they may comprise the majority of the genome (IHGSC 2001
). While some types of mobile element excise and transpose using only DNA intermediates, the most abundant mammalian families, short interspersed elements (SINEs) move via an RNA route: the template element remains in place, but a transcript is reverse transcribed and inserted elsewhere in the genome (Smit 1996; Rowold and Herrera 2000). However, the means by which SINEs have reached their current high genomic abundance are unclear. One opinion (Zeyl and Bell 1996) is that these sequences are parasitic; their copy number is in this view regulated by a balance between transposition, loss by deletion, and selection against individuals with a high number of copies. It has been suggested that mechanisms such as interfering RNA have indeed evolved to counter the effects of parasitic mobile DNAs (Tabara et al. 1999).
Others maintain (Brosius 1999; Kidwell and Lisch 2001) that interspersed repetitive elements have something more like a mutualistic relationship with the host genome, citing instances of adaptive SINE insertions or gene control regions derived from SINEs. Accumulating SINEs would certainly be expected to cause mutations, some of which might well be beneficial. According to the mutualistic view, then, SINEs provide valuable genomic variation, which contributes to the adaptive evolution of the host. Similar arguments have been put forward concerning the evolution of the mutation rate in general. Many studies have examined the evolutionary fate of mutator alleles that increase both adaptive and deleterious mutations: in general, theoretical work indicates that alleles increasing the mutation rate—such as SINE master genes—will not be selectively favored in sexual populations (for a review, see Sniegowski et al. 2000).
Although there are several well-documented instances of adaptive SINE insertions (Miller et al. 1999), these do not in themselves constitute compelling evidence for mutualism: it is not disputed that individual SINE insertions could occasionally be selectively advantageous. Clearly, most insertions will be harmful or neutral in their effects. The question is whether the existence of any SINE family can be viewed as a genomic adaptation at the level of the host rather than a parasitic "infection." A third possibility is that SINEs are a maladaptive but unavoidable property of a sequence selectively maintained for other reasons.
The Master Gene Hypothesis
SINE families exhibit another interesting feature: an unexpected structure to the variability of sequences. Typically, they show a pattern of the sequential replacement of subfamilies by others (Shen, Batzer, and Deininger 1991). This pattern is consistent with the expectations of a model in which only one element is capable of being copied to new locations. The proposed sole replicative locus is termed the master gene. In the master gene view, subfamily procession corresponds with sequence evolution at the master gene locus. This explanation has been advanced for many repeat families including human Alu and LINE (Long Interspersed Element) elements (Kass, Batzer, and Deininger 1995).
The master gene hypothesis states that only one locus in the genome is capable of spawning new copies and that these new copies are inert and do not inherit the master gene's propensity to replicate. We here refer to the alternative as a "transposon" model, in which members of the family capable of replication are present at multiple sites in the genome, although replicative potential may still vary among these sequences. The master gene hypothesis is relevant to the mutualism/parasitism question discussed above for the following reason. A master gene must have arisen as an allele, become fixed in the population, and maintained its replicative potential since that time. Therefore, it must be tolerated by selection, either because of its repeat-spawning behavior or in spite of it: any Mendelian gene has the same evolutionary "interests" as its host. Genomic parasitism is incompatible with a single master gene, although the existence of such a gene would not necessarily imply a benefit to SINE amplification: this could be a maladaptive but unavoidable property of a sequence selectively maintained for other reasons. By contrast, an element spawning copies that can also act as templates shows non-Mendelian inheritance and can therefore increase in frequency even if it inflicts a cost on the host (Bestor 1999).
The number of replicating sequences in a family need not be great for genomic parasitism to explain their presence—elements must merely produce copies that may themselves be copied. The evolutionary interests of a transposing family of replicable elements are drastically different from a repeat-forming allele at a single genomic locus. This is true however few copies of the element exist in the genome at any one time. However, the more frequently copies are made, the more damage they may cause the host: the advantage gained through replication will counter selection at the level of the host (Orgel and Crick 1980). Quantifying the number of loci to which replication is confined is therefore extremely important in understanding the evolution of interspersed repetitive elements. It is occasionally suggested that an element family has "several master genes" (e.g., Kass, Raynor, and Williams 2000). However, discriminating between several master genes and a small family of self-replicating elements in a larger family of inert copies requires a rigorous definition of a master gene that is currently lacking.
Newly arising SINE families are a special case. By definition, a new family of SINEs is at first confined to a single locus, which could be considered a master gene. Some of the features allowing the element to be copied may be specific to that locus. For example, adjacent control regions may favor the transcription, reverse transcription, or integration of the new element, but these regions would not be part of the element and hence would not be transmitted to its copies (Roy et al. 2000). However, a new family of selfish elements can emerge if some of the copies themselves have the potential to replicate.
The existence or otherwise of a master gene is therefore vitally important in understanding both the origin and the accumulation of SINEs within genomes. The answer will also affect the interpretation of changing rates of SINE accumulation; a past increase in accumulation rate need not be due to increased activity of the master gene or the effects of host population size but could simply be due to an increased number of template elements.
Here we suggest a new type of test to investigate whether the pattern of variation in an interspersed repetitive sequence family is as expected under the master gene model.
Rodent ID Elements
The test is applied to the ID (identifier) element in mouse, a SINE specific to rodents (Kass and Kim 1996).
Rodent ID elements are the best-supported example of master gene–style expansion of an interspersed repeat family because the purported master gene has been identified; persuasive evidence supports BC1 being a major source of new sequences. BC1, a noncoding RNA thought to have a function in certain neurons (Kim et al. 1994), is abundantly expressed in the testis (Muslimov et al. 2002) and capable of priming its own reverse transcription (Shen, Brosius, and Deininger 1997), both of which would promote self-replication in the germline.
The BC1 gene and ID elements have been studied not only in the mouse but also in other rodents to look at the evolution of the family between species. The number of ID elements is extremely variable between species (Kass and Kim 1996). Further evidence pointing to BC1 as the master gene is found from these interspecific studies: differences in the predominant ID element of different rodent species correspond to interspecific differences at this locus. The guinea pig, moreover, has two BC1 genes, and its genome contains ID elements corresponding to both sequence variants. In the rat, however, it appears that BC1 has been superseded as the master gene. Newer subfamilies have emerged whose sequence does not correspond to that at the BC1 locus (Kim and Deininger 1996), and transcripts from several loci are present in rat testes (Kim, Kass, and Deininger 1995).
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
The process by which element copies are generated will leave its mark in their phylogeny: that is, the relationships of the different elements of elements within a single genome. A master gene will produce a tree of elements in which all bifurcations occur in a single lineage: a pectinate, or comblike, tree (fig. 1a). Non–master gene amplification will result in bifurcations not involving the master gene lineage (fig. 1b). A transposon model can also give a pectinate tree under certain conditions (Brookfield and Johnson, in press
), and this effect will be exacerbated by variation in element activity. However, given a large enough sample size, bifurcations not involving the master gene should still be observed, particularly among recent coalescences.
|
Given a true phylogeny of the elements, the method for distinguishing between these two hypotheses would simply be to construct a phylogenetic tree of all sequences and see whether the phylogeny resembles that shown in figure 1a. Unfortunately, many SINE sequences are so numerous, and the number of variable sites are so low, that establishing the true phylogeny is not feasible. The chances of repeat mutation are high, especially as SINEs are often found in heterochromatic regions and are liable to show mutations resulting from methylated CpG. Therefore, we cannot simply produce a tree of the sequences and see if it corresponds to the shape expected under the master gene model.
Despite these difficulties, we are able to test the master gene model using the following logic. Consider a pair of elements, A and B, neither of which is the master gene. Figure 2a shows the topology of the tree expected under the master gene model, in which four sequences are shown. M represents the master gene, while A and B represent the sequence pair under consideration. O represents the sequence in the data set that is the most diverged from the master gene, which is used as an outgroup.
|
When compared to the master gene and outgroup, each informative nucleotide site can fall into three types we shall refer to as type 1-3 categories. Figure 2b–d shows the trees that would be inferred from these informative nucleotides, taken individually. Sequence A could group with the outgroup and sequence B with the master as in figure 2a, (Type 1) site (fig. 2b) or vice versa, which we shall call type 2 (fig. 2c), or sequences A and B could group together, called type 3 (fig. 2d). If we do not know which of the sequences, A and B, is the younger of the pair—that is, the positions of A and B could be reversed—then we cannot say that a type 1 or a type 2 site requires multiple mutations. However, a type 3 site always requires two mutations to have occurred given that the master gene model is true: for example, as shown in figure 2e.
Figure 3 illustrates this: for the pair of sequences shown, there is one type 1 base, three type 2, and one type 3.
|
In a data set with no back mutation, all informative sites will reflect the true relationship between the four sequences, so only one type of site will be observed for any pair of elements. However, where there have been repeated mutations to a site, the pattern shown at any one site may not correspond to the true relationship between sequences, so a pair of elements can contain bases in more than one of these three types.
Even in a fully master gene–like situation, then, some pairs of elements will show type 3 sites, but these will be generated only by repeat mutation. If the master gene model is false, however, there will be pairs of sequences that do have the relationship shown in figure 2d. Here, a single mutation would be expected to cause A and B to be different from O and M. Thus, if there were more type 3 sites than expected, this would represent evidence against the master gene. Unfortunately, because one cannot assume that all sites are equally mutable, it is difficult to determine the expected numbers of type 3 sites given that the master gene model is correct.
However, note that for pairs of elements showing the relationship seen in figure 2d, type 1 or 2 sites now require two mutations as no branch now groups either with the master or outgroup. Thus, such pairs would be expected to have many type 3 sites, and few type 1 or 2 sites, resulting in a negative relationship between the number of type 3 bases and the number of type 1 or 2. We can therefore test the master gene model by calculating, for each pair of sequences, the number of type 3 sites and regressing this against the number of sites of type 1 plus type 2. If several sequences are replicating, as in figure 1b, a negative slope is expected. Note that this test requires that the master gene be identified correctly. However, where there are several potential master genes, testing each in this way may help determine which has contributed most to sequence replication.
The 75-bp sequence of the purported master gene, from mouse BC1 (GI|401724|GB|U01310.1), was used in a Blast search of the mouse genome (Mouse Genome Sequencing Consortium 2002). Sequences scoring highly were aligned by eye and any duplicates removed, so that in the final data set no sequence was identical to any other. This would potentially remove some evidence of recent non–master gene replication but also meant that there was less chance of spurious shared changes from the master gene resulting from the same genomic region being sequenced more than once. It was unclear whether mismatches at the very extremes of the sequences were due to the insertion of a truncated product or to later mutation in situ. Therefore, 3 bp were trimmed from each extreme of the sequences to leave an unambiguous alignment. The resulting data set contained 212 sequences, each 69 bp long. It was not possible to determine which, if any, of these 212 sequences were potentially autonomous, as it is not well understood how linked sequences affect the potential for elements to replicate. The sequence with the most (13) changes from the potential master gene, GI|26292109|GB|AC125144.3, was used as the outgroup.
Sequences other than BC1 and the outgroup were then taken in pairs, and for every such pair the number of sites of each type were counted.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Figure 4 shows, for all possible pairs of sequences, the number of sites that show the third pattern (type 3 sites) plotted against the sum of the number of sites showing the first and the second patterns (type 1 and type 2 sites) for the two sequences. Note that each point may represent many sequence pairs. Regressing the number of type 3 bases against the total number of type 1 or 2, it is found that the slope is very significantly negative (P = 1.34 x 10–32).
|
While this relationship is highly significant and while it would be unexpected according to the master gene model, the P value is misleading because not only may the residuals be nonnormal but also, more importantly, the pairs of sequences are not independent of each other. Instead, we test the significance of this slope using simulated data.
Using the program Seq-Gen (Rambaut and Grassly 1997), we simulated 1,000 data sets, each containing 212 sequences 69 bp in length, related according to a completely master gene–like tree, and calculated slopes similarly. Branch lengths were taken from the real data set such that the expected number of mutations between the master gene and the oldest sequence in the simulated data sets was the same as in the real data set. Internal branch lengths were calculated in two ways: firstly, by assuming constant replicational activity of the master gene and therefore regular intervals between master gene replications. Where this was the case, 34 of 1,000 slopes were more negative than the –0.065 observed from the real data. Alternatively, the internal branch lengths were all set at values calculated from the real data set. This gave 16 of 1,000 slopes less than –0.065. To mimic extensive back mutation, another data set was simulated in which half of the sites were assumed to be invariant (although in fact all sites were mutated in some sequences). This gave 21 of 1,000 slopes less than –0.065.
The above calculations assume that the master gene branch has evolved at the same rate as other branches, which is unlikely. If, as the opposite extreme, the master gene sequence is assumed to be highly conserved, such that it has not itself mutated since the oldest sequence in the data set was produced, none of 1,000 slopes was less than –0.065 under any of the above models. All data sets used the JC69 model of nucleotide substitution: equal frequencies of the four bases and equal rates of transitions and transversions (Jukes and Cantor 1969). The sequence most divergent from the master was used as the outgroup, rather than the true outgroup defined in the tree file. Simulated and real data sets were therefore subject to identical analysis.
Segmental duplication of large stretches of the genome could also give the appearance of non–master gene replication. To rule out this possibility, we took the pairs of ID elements with high numbers of type 3 sites, then went back to the original genomic sequence and examined the adjacent regions. Similarity was confined to the ID element itself (fig. 5).
|
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Here we have presented evidence that BC1, the purported master gene of mouse ID elements, is not the only source of new elements. Although rodent ID elements have been seen as one of the best examples of likely master gene replication, in rats also, BC1 is not the major source of new transpositions. Rats also have more ID element copies than any other rodent species studied, which may not be coincidental: a locus creating many insertional mutations would be very unlikely to survive. There are also examples of non–master gene replication in other SINE families (Leeflang et al. 1992
; Johanning et al. 2003). Another potential cause of shared differences from the master gene among SINE copies is the presence, at some time in the past, of multiple simultaneously active alleles at the master gene locus. However, given the small effective population sizes of mammals and the high number of shared changes between some element pairs, we think it unlikely that this is responsible for the above result. Similarly, while gene conversion could potentially also create patterns that we find, the length of the sequence is too short to form a tract for gene conversion in mammals.
A parasitic explanation for the presence of ID elements is therefore plausible in this case, and perhaps appropriate for all such elements. While the discovery of major source genes such as BC1 can give us valuable insight into how SINE families arise, the subsequent accumulation of SINEs in the genome is often inconsistent with a master gene model.
Our test could be applied to other element families suspected of accumulating via a master gene process and for which a potential master gene has been identified.
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
This work was supported by a Biotechnology and Biological Sciences Research Council grant, reference number 42/GI 3767. The authors thank Adam Eyre-Walker for helpful suggestions.
|
Footnotes |
|---|
Pierre Capy, Associate Editor
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Bestor, T. H. 1999. Sex brings transposons and genomes into conflict. Genetica 107:289–295.[CrossRef][ISI][Medline]
Brookfield, J. F. Y., and L. J. Johnson. 2005. The evolution of mobile DNAs—when will transposons create phylogenies that look as if there is a master gene? Genetics (in press).
Brosius, J. 1999. RNAs from all categories generate retrosequences that may be exapted as novel genes or regulatory elements. Gene 238:115–134.[CrossRef][ISI][Medline]
[IHGSC] International Human Genome Sequencing Consortium. 2001. Initial sequencing and analysis of the human genome. Nature 409:860–921.[CrossRef][Medline]
Johanning, K., C. A. Stevenson, O. O. Oyeniran, Y. M. Gozal, A. M. Roy-Engel, J. Jurka, and P. L. Deininger. 2003. Potential for retroposition by old Alu subfamilies. J. Mol. Evol. 56:658–664.[CrossRef][ISI][Medline]
Jukes, T. H., and C. R. Cantor. 1969. Evolution of protein molecules. Pp. 21–123 in H. N. Munro, ed. Mammalian protein metabolism. Academic Press, New York.
Kass, D. H., M. Batzer, and P. L. Deininger. 1995. Gene conversion as a secondary mechanism of short interspersed element (SINE) evolution. Mol. Cell. Biol. 15:19–25.[Abstract]
Kass, D. H., J. King, and P. L. Deininger. 1996. Sporadic amplification of ID elements in rodents. J. Mol. Evol. 42:7–14.[CrossRef][ISI][Medline]
Kass, D. H., M. E. Raynor, and T. M. Williams. 2000. Evolutionary history of B1 retroposons in the genus Mus. J. Mol. Evol. 51:256–264.[ISI][Medline]
Kidwell, M. G., and D. R. Lisch. 2001. Transposable elements, parasitic DNA and genome evolution. Evolution 55:1–24.[CrossRef][ISI][Medline]
Kim, J., and P. L. Deininger. 1996. Recent amplification of rat ID sequences. J. Mol. Biol. 261:322–327.[CrossRef][ISI][Medline]
Kim, J., D. H. Kass, and P. L. Deininger. 1995. Transcription and processing of the rodent ID repeat family in germline and somatic cells. Nucleic Acids Res. 23:2245–2251.
Kim, J., J. A. Martignetti, M. R. Shen, J. Brosius, and P. L. Deininger. 1994. Rodent BC1 RNA gene as a master gene for ID element duplication. Proc. Natl. Acad. Sci. USA 91:3607–3611.
Leeflang, E., W. Liu, C. Hashimoto, P. V. Choudary, and C. W. Schmid. 1992. Phylogenetic evidence for multiple Alu source genes. J. Mol. Evol. 35:10–16.
Miller, W. J., J. F. McDonald, D. Nouaud, and D. Anxolabehere. 1999. Molecular domestication—more than a sporadic episode in evolution. Genetica 107:197–207.[CrossRef][ISI][Medline]
Mouse Genome Sequencing Consortium. 2002. Initial sequencing and comparative analysis of the mouse genome. Nature 420:520–562.[CrossRef][Medline]
Muslimov, I. A., Y. Lin, M. Heller, J. Brosius, Z. Zakeri, and H. Tiedge. 2002. A small RNA in testis and brain: implications for male germ cell development. J. Cell Sci. 115:1243–1250.
Orgel, L. E., and F. H. C. Crick. 1980. Selfish DNA: the ultimate parasite. Nature 284:604–606.[CrossRef][Medline]
Rambaut, A., and N. C. Grassly. 1997. Seq-Gen: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees. Comput. Appl. Biosci. 13:235–238.
Rowold, D. J., and R. J. Herrera. 2000. Alu elements and the human genome. Genetica 108:57–72.[CrossRef][ISI][Medline]
Roy, A. M., N. C. West, A. Rao, P. Adhikari, C. Aleman, A. P. Barnes, and P. L. Deininger. 2000. Upstream flanking sequences and transcription of SINEs. J. Mol. Biol. 302:17–25.[CrossRef][ISI][Medline]
Shen, M. R., M. Batzer, and P. L. Deininger. 1991. Evolution of the master Alu genes. J. Mol. Evol. 33:311–320.[CrossRef][ISI][Medline]
Shen, M. R., J. Brosius, and P. L. Deininger. 1997. BC1 RNA, the transcrpit from a master gene for element activation, is able to prime its own reverse transcription. Nucleic Acids Res. 25:1641–1648.
Smit, A. F. A. 1996. The origin of interspersed repeats in the human genome. Curr. Opin. Genet. Dev. 6:743–748.[CrossRef][ISI][Medline]
Sniegowski, P. D., P. J. Gerrish, T. Johnson, and A. Shaver. 2000. The evolution of mutation rates: separating causes from consequences. Bioessays 22:1057–1066.[CrossRef][ISI][Medline]
Tabara, H., M. Sarkissian, W. G. Kelly, J. Fleenor, A. Grishok, L. Timmons, A. Fire, and C. C. Mello. 1999. The rde-1 gene, RNA interference, and transposon silencing in C. elegans. Cell 99:123–132.
Zeyl, C., and G. Bell. 1996. Symbiotic DNA in eukaryotic genomes. Trends Ecol. Evol. 111:10–15.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||