NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Coffin JM, Hughes SH, Varmus HE, editors. Retroviruses. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press; 1997.

Cover of Retroviruses


Show details

Mutational Consequences of Retroelements in the Germ Line

Endogenous Retroelements and the Host Genome: Genetic Consequences of Endogenous Retroviral Insertion

It has been estimated that perhaps 5% of the total vertebrate genome consists of retroelements and that up to sixfold more resulted from the reverse transcription of other RNAs (Baltimore 1985; Smit 1996). It is, then, not surprising that these genetic parasites have played a significant part in molding the genomes of their hosts. This section explores some of the effects of retroelement insertion as well as their potential uses, both for the organism and for the experimental scientist.

Any introduction of DNA into the host genome can have genetic consequences. Every retroviral integration event is potentially mutagenic. As originally demonstrated with bacterial transposons, the moveable element not only causes the mutation, but also provides a tag for cloning the afflicted gene (Kleckner et al. 1977). It might therefore be anticipated that genetic studies of endogenous retroviruses would prove richly rewarding. First, even if not mutagenic, every provirus is a potential source of useful polymorphism allowing molecular genetic access to previously undefined regions of the genome. Second, the identification of novel proviruses in spontaneously occurring insertion mutations might provide a means for identifying and cloning genes important in development and other processes. Third, a detailed understanding of the ways in which individual endogenous elements can affect the gene activity of their hosts should allow the development of better strategies for gene discovery.

Endogenous Retroviruses as Genetic Mapping Tools

Recently acquired endogenous proviruses, for example, the endogenous MLVs, show genetic polymorphism and will therefore segregate in a cross. With the development of strategies to allow the unambiguous identification of individual proviruses (see above), it became possible to turn to advantage the complex patterns shown on Southern blots of mouse DNA hybridized with MLV probes and employ endogenous proviruses as multilocus genetic markers, i.e., a number of independently segregating markers that can be analyzed simultaneously. Such markers can therefore facilitate rapid scanning of the genome to locate novel mutations, mapping of multigenic traits, and screening for novel tumor suppressor genes by reduction to homozygosity (Siracusa et al. 1991). For example, it should be possible to scan more than 85% of the murine genome using just the three nonecotropic oligonucleotide probes (Frankel et al. 1990). The maps generated are of low resolution, but they permit the rapid identification of regions of interest, which can then be examined further using more specific markers such as microsatellites (Dietrich et al. 1994). One approach has been to breed a tester strain, MEV, containing a large number of defined ecotropic MLV proviruses that can be typed simultaneously using the ecotropic-specific envelope probe (Taylor and Rowe 1989; Taylor et al. 1993). In addition, naturally occurring MMTV (Siracusa et al. 1991), nonecotropic MLV (Frankel et al. 1990), and IAP proviruses (Lueders et al. 1993b) are also very useful. A variety of genes and transgenes have been mapped in this way (Gerstein et al. 1990; Hanzlic et al. 1990; Rise et al. 1991; Messer et al. 1992; Peters and Barker 1993). A recent example is provided by the mapping of the stumbler (stu) mutation to mouse chromosome 2 (Frankel et al. 1994). Initially, linkage of stu to an IAP was established using a small cross in which stu was segregating. The IAP was then mapped to chromosome 2, using a well-characterized mapping panel. Finally, the map position was refined in the initial cross, using a series of microsatellite markers from the proximal region of chromosome 2.

The drawback to using endogenous proviruses for mapping is that only the proviral allele is detected, not the preintegration site. This means that the utility of the approach is determined by the number and distribution of the proviruses detected. If there are too few proviruses (probably the case with ALVs and endogenous FeLVs), the information gained is hardly worth the effort. The case of too many proviruses poses the technical question of how to resolve them. This can be resolved by using specific probes (as with MLVs), but the question of distribution cannot be solved as easily. Viruses that have been present in the germ line since before speciation will be present in all individuals of a species and will therefore not (usually) show polymorphism. Therefore, it seems unlikely that probes to, say, human endogenous retroviruses will be useful in genetic studies. (This would change if there has been recent movement of these viruses, a question that deserves reexamination in light of the data implying partial activity of the HERV-K family.)

Insertional Mutagenesis

Before examining the potential role of endogenous retroviruses as mutagenic agents, it is useful to consider the ways that endogenous elements can interact with their hosts. It must be emphasized that the provirus does not consist merely of inert DNA causing mutation by interrupting gene sequences but comes equipped with cis- and trans-acting sequences designed to drive and control the synthesis and processing of its own genes. These signals are also capable of acting on the host genome. Inserted proviruses can therefore affect the gene expression of neighboring genes by a variety of mechanisms, several of which are also important in viral oncogenesis (see Chapter 10. These include providing novel promoters or enhancers, altering RNA splicing, and interrupting coding sequences. Table 9 lists a number of germ-line mutations in the mouse that have been shown to result from insertion of retroviruses or retrovirus-like elements. Some are spontaneous, whereas others arise from experimental infection. In some cases, there is at least a superficial understanding of the mode of action of the provirus. These studies permit a few general conclusions, although it should be stressed that in the absence of direct experimentation, some of these conclusions should be regarded as tentative.

Table 9. Germ Line Mouse Mutations Caused by Retroviral Insertion.

Table 9

Germ Line Mouse Mutations Caused by Retroviral Insertion.

The analogy with transposon mutagenesis of bacteria (Kleckner et al. 1977) suggests that endogenous retroviruses might prove to be responsible for a significant fraction of spontaneous mutations, a view supported by the fact that perhaps 50% of Drosophila mutations are caused by insertion of moveable genetic elements (Ramel 1989). Systematic surveys of the MLV content of inbred and mutant mouse strains have positively identified two recessive mutations, dilute (Jenkins et al. 1981) and hairless (Stoye et al. 1988), that were caused by proviral insertions. Genetic proof of the causation of both mutations was obtained by finding revertants that had also lost the provirus by homologous recombination leaving a solo LTR. These proviruses and their associated flanking sequences were cloned; the flanking sequences were then used to identify transcripts from the wild-type alleles of the mutated genes. dilute encodes a novel myosin-like protein (Mercer et al. 1991) and hairless is a potential zinc finger protein (Cachon-Gonzalez et al. 1994). A number of other mutations are closely associated with specific proviruses, initially prompting speculation that MLVs might be responsible for up to 5% of all mutations in mice (Stoye et al. 1988). The xenotropic provirus associated with the retinal degeneration mutation (Frankel et al. 1989a) was found to lie within the first intron of the gene encoding the β-subunit of the cGMP phosphodiesterase. Although it may well cause the mutation, it is not the only inactivating change present in the gene (Bowes et al. 1993). Detailed investigations have shown that several of the other associations detected represent very tight linkage rather than causality (Rinchik et al. 1993; Winkes et al. 1994; Stoye et al. 1995). The 5% figure therefore seems somewhat high; a more realistic estimate is probably that 1% of the classic, visible mutations of mice arose by MLV insertion. Systematic surveys of mutant strains of mice for extra members of other families of endogenous virus have not been attempted.

The cloning of the pink-eyed gene illustrates the utility of retroviruses in gene hunting. To test the hypothesis that the pink-eyed unstable allele was caused by a retroviral insertion, high-resolution Southern blots were hybridized with a variety of retroviral probes. One out of several hundred fragments gave an enhanced signal compared to controls after hybridization with an IAP probe. This IAP-containing fragment was cloned and found to lie within a duplication of genomic DNA associated with the pink-eyed unstable mutation (Brilliant et al. 1991).

Germ line mutations caused by proviral insertion mutations are usually recessive, in contrast to the usually dominant effects seen in tumors. Presumably, the majority of such mutations cannot be passed through the germ line. Inactivating insertions are usually, but not exclusively, in introns (see Table 9), but probably not at a higher frequency than would be expected given that introns are on average ten times larger than exons. Many of the insertions appear to have occurred in the first intron, but it is not clear whether this reflects a bias in integration near the 5′ ends of genes (Rohdewohld et al. 1987) or whether such mutations are more likely to be mutagenic. It should be noted that proviral integration within a gene does not necessarily affect its expression; in the case of a mouse substrain carrying the Mov10 locus, a Mo-MLV insertion in the first intron of a gene for a putative GTP-binding protein has no perceptible effect on gene expression (Mooslehner et al. 1991). Inhibition may be tissue-specific; the Mov13 insertion, which blocks α-collagen expression in most tissues, has no effect in odontoblasts and some osteoblasts (Kratochwil et al. 1989), and Emv3 shows a greater effect on the dilute message in melanocytes compared to neural tissue (Mercer et al. 1991).

Proviral integrations can interfere with gene activity in a number of ways. The Mo-MLV insertion at Mov13 disrupts DNA sequences in the first intron of the collagen gene and is thought to prevent the binding of factors necessary for transcriptional activity (Barker et al. 1991). The proviruses causing the dilute and hairless mutations perturb normal mRNA splicing (Seperack et al. 1995; B. Cachon-Gonzalez and J.P. Stoye unpubl.). Hybrid messages, apparently terminating in the 3′LTR, are generated by splicing from genomic exons upstream of the provirus to the viral env splice acceptor (Fig. 13). This implies that transcription must proceed through the 5′LTR polyadenylation signal. Reversions of dilute and hairless have been reported; in both cases, a solo LTR remains at the integration site (Fig. 13) (Copeland et al. 1983; Stoye et al. 1988), implying that normal mRNA transcription and/or splicing can take place despite the presence of a solo LTR. However, transcripts terminating at the solo LTR were detected (to a degree varying from tissue to tissue), indicating that solo LTRs can affect mRNA processing (Seperack et al. 1995). Given the number of solo LTRs present in the genome, a more extensive investigation of the extent of this phenomenon seems to be warranted.

Figure 13. Insertional mutagenesis by inhibition of splicing.

Figure 13

Insertional mutagenesis by inhibition of splicing. A proviral insertion into intronic sequences and in the same transcriptional orientation as the cellular gene introduces a splice acceptor site (the viral env splice acceptor). Utilization of this splice (more...)

The situation with the nonagouti allele, a, of the agouti coat color gene (for review, see Siracusa 1994) is more complicated, apparently reflecting the alternative use of untranslated exons 5′to the three coding exons. The mutation results from insertion, just upstream of the gene's coding exons, of a VL30 element. A 5-kb fragment of genomic origin flanked by two 526-bp direct repeats is present in the middle of the VL30 sequence (Bultman et al. 1994). Recombination can take place between either pair of repeats (Fig. 14), leading to different alleles with characteristic alterations in phenotype apparently reflecting different effects on the translation and/or splicing of the alternative mRNAs.

Figure 14. The nonagouti mutation.

Figure 14

The nonagouti mutation. The agouti gene contains four noncoding exons (ABCD) which are differentially regulated and spliced to the three coding exons (2,3,4) (for review, see Siracusa 1994). The nonagouti mutation is characterized by the presence of a (more...)

Endogenous retroviruses can also supply novel termination signals. dilute and hairless provide two examples and an insertion causing the dominant T Wis mutation of the Brachyury gene, a third. In this case, an ETn has integrated within the splice donor site of exon 7 of the wild-type gene, interrupting splicing (Herrmann et al. 1990) and presumably terminating mRNA at the ETn LTR poly(A) site. The truncated protein generated from the aberrant mRNA can apparently act as a dominant-negative factor to prevent mesoderm formation (Herrmann and Kispert 1994). Several human cDNAs with polyadenylation signals provided by HERV elements have been described (Wilkinson et al. 1994); in at least one case, the retroviral sequences may be playing a functional part in the processing of a single-copy, multiexon RNA with a long ORF (Goodchild et al. 1992).

Insertion of retroviral elements can also provide promoters or enhancers that direct transcription and cause dominant mutations. The three mutations of agouti, A iapy, A iy, and A vy, show similar phenotypes and result from promoter insertion by IAP elements (Fig. 15) in a manner similar to that seen for the somatic activation of c-mos in a mouse myeloma (Horowitz et al. 1984). In each case, a cryptic promoter in the U3 region of the 5′LTR apparently produces an mRNA in the opposite transcriptional orientation to the provirus (Duhl et al. 1994; Michaud et al. 1994; Perry et al. 1994). The cryptic promoter has not been mapped, although bidirectional transcription from IAP LTRs has been reported elsewhere (Christy and Huang 1988). Interestingly, levels of message vary among animals carrying the same mutation. Levels of message also correlate inversely with host-mediated methylation of the IAP LTR; methylation, in turn, depends on whether the mutant allele is inherited through the male or female germ line, a phenomenon known as germ line imprinting (Efstratiadis 1994). The dominant mutations of agouti might provide a valuable system for studying this effect.

Figure 15. IAP insertions at the agouti locus.

Figure 15

IAP insertions at the agouti locus. Three dominant mutations of the agouti locus are caused by insertion of IAP elements (Duhl et al. 1994; Michaud et al. 1994). In all three cases, the provirus integrates upstream, and in the opposite transcriptional (more...)

The mouse sex-linked protein (slp) requires androgen for expression. The androgen apparently acts on a hormone-responsive enhancer within the LTR of an ancient and otherwise uncharacterized retroviral element inserted 2 kb upstream of the slp gene. The element then imposes androgen responsiveness on its neighboring gene (Stavenhagen and Robins 1988).

Control of gene expression by endogenous elements is also seen in humans. HERVs can regulate the expression of a number of human genes (Wilkinson et al. 1994). The best studied example concerns the amylase gene cluster (Samuelson et al. 1990; Ting et al. 1992). Insertion of an HERV-E element, in the opposite transcriptional orientation, into the upstream region of a duplicated copy of the pancreatic amylase gene has led to a switch from pancreatic to parotid expression and may be responsible for salivary amylase expression in primates. A further duplication event followed the HERV-E insertion; the provirus flanking one gene was lost, leaving behind a solo LTR that does not confer parotid-specific amylase expression. A second example of gene control by endogenous retroviruses concerns a phospholipase A2-related gene (Feuchter-Murthy et al. 1993). Expression of this gene is driven by an HERV-H LTR and initiates within the 5′LTR of an intact element. The mRNA is formed by splicing from the viral splice donor to a cellular splice acceptor.

A final example of the potential impact of endogenous elements comes from chickens with the henny-feathering trait, a dominant autosomal condition that leads to the enhanced expression of aromatase mRNA. Animals with this trait express an additional aromatase mRNA compared to control animals. This RNA contains at its 5′end sequences homologous to the U5 region of RAV-0 (Matsumine et al. 1991).

To assess the genetic impact of retroviral infection, it is of interest to ask what percentage of proviral insertions result in phenotypic change. Unfortunately, we cannot use endogenous viruses for this analysis as integrations with deleterious effects may be counterselected. Therefore, this analysis must be confined to novel proviruses and not include the proviruses present in standard strains of mice. Systematic inbreeding of mice carrying new MLVs, resulting either from experimental infection of embryos or by replication of spontaneously expressed endogenous MLV, indicates that approximately 5% (8/142) of all novel insertions result in clear-cut mutations (Spence et al. 1989; Weiher et al. 1990; Taylor et al. 1993). Several of these mutations are recessive prenatal lethal mutations and were initially recognized only by the absence of homozygous animals in progeny of crosses between two heterozygous provirus-containing mice. Presumably, such integrations will be rapidly counterselected during normal breeding and will not be found in collections of mutant animals.

Two lines of evidence suggest that this 5% figure represents a minimum estimate for the frequency of viral integrations affecting gene activity. First, mice carrying ablations of specific genes frequently appear to be superficially normal, suggesting that gene function can be lost without obvious phenotype (Brandon et al. 1995). The second line of evidence comes from studies using vectors (gene trap vectors, enhancer trap vectors, etc.) that have been designed to allow selection of insertion mutations (Skarnes 1990; Friedrich and Reddy et al. 1991; Soriano 1991; Skarnes et al. 1992). Integration of these vectors within a transcription unit activates a reporter gene (Fig. 16) and prevents proper splicing of the target gene. Approximately 12% of unselected integrations into embryonal stem cells allow splicing events that cause reporter gene activation and host gene inactivation. These cells can be selected and used to regenerate chimeric animals heterozygous for the inserts which can then be tested for an affect by breeding to homozygosity. Approximately 40% of these selected insertions give an overt phenotype when animals are bred to homozygosity (Friedrich and Soriano 1991; von Melchner et al. 1992). Whether the remaining genes are really dispensable or whether the silent insertions represent leaky mutations remains an open question. Nevertheless, this approach represents a significant improvement over standard retroviral insertional mutagenesis protocols for mammalian gene discovery. Mutations in several interesting genes have been identified in this way (Chen et al. 1994; DeGregori et al. 1994), and it is anticipated that this approach will prove a fruitful source of novel genes for developmental studies.

Figure 16. Use of a retroviral promoter trap vector.

Figure 16

Use of a retroviral promoter trap vector. The retroviral vector ROSAβ-geo (Friedrich and Soriano 1991) consists of a reporter gene with a neo gene fused to the 3′end of the lacZ gene, flanked by splice acceptor and polyadenylation signals (more...)

Insertional Alteration of Host Gene Expression by Retrotransposons

Insertional inactivation by retrotransposons is a common phenomenon that can result from insertion into coding regions, with the production of a truncated protein. In addition, many retrotransposons contain terminators that interfere with expression of the target gene at the transcriptional or translational levels. Finally, insertions within promoter regions can disturb the spacing between normal promoter elements to eliminate or severely reduce gene expression.

Retrotransposons can insert into promoter, 5′UTR, intronic, or 3′UTR regions of target genes, with wide-ranging effects. Some of the more interesting effects on gene expression result from the activation of target gene expression mediated by cis elements that bind host-cell factors important for gene expression within the transposon. These are analogous to the steroid-response elements in MMTV that can activate oncogene expression in a hormone-dependent manner. Well-studied examples include the yeast Ty1 and Ty2 elements, which contain sequences adjacent to the 5′LTR that activate adjacent gene expression in a manner that depends on the yeast mating type. Activation is observed in mating-proficient yeast but not in MAT a/α diploid yeast (for review, see Boeke and Sandmeyer 1991). One well-studied example from Drosophila is the gypsy element, which has been shown to mediate adjacent gene expression via the Su(Hw) protein. This regulation is mediated by binding sites for this host protein near the 5′end of the gypsy element. Binding of the Su(Hw) protein to these sequences appears to “insulate” the target gene downstream from the gypsy element from the action of enhancers located far upstream (Geyer and Corces 1992).

Hybrid Dysgenesis Associated with Retrotransposons

Hybrid dysgenesis is a genetic phenomenon in which the hybrids resulting from mating of two dissimilar strains show reduced fertility, increases in mutation rate, nondisjunction, male recombination, segregation distortion, and chromosome aberrations (Table 10). Hybrid dysgenesis has been associated with at least five transposon families of different types from Drosophila. Usually, hybrid dysgenesis is apparent only in one sexual “orientation,” i.e., hybrids will have different behaviors depending on whether the transposon is introduced via the maternal or paternal germ line. One of two classic systems of hybrid dysgenesis in Drosophila, called the I/R system, is associated with I factor retrotransposition (Bucheton et al. 1984). In dysgenic crosses, strains that donate an active I factor are termed I (inducer) strains, whereas R (reactor) strains lack expressed I factors. To obtain a strong dysgenic phenotype in this system, I males must be crossed with R females; the resultant hybrid females show a temperature-dependent sterile female (SF) phenotype. Unlike the more extensively studied P-M dysgenesis system (Engels 1989), in which actual atrophy of the gonads in the affected hybrids is readily observed, the I-R dysgenic hybrid females simply lay dead eggs. The SF phenotype is temperature- and age-dependent; older mothers can lay eggs with much higher levels of hatching. The surviving progeny of the sterile females show high incidences of mutation, and these mutations are caused by new I factor insertions (Finnegan 1989; Bucheton 1990). Although active I factor is obviously a major determinant of this type of dysgenesis, the mechanism by which the eggs die is not obvious. Finally, hybrid dysgenesis-like phenomena are associated with the Drosophila elements gypsy (Song et al. 1994), as well as the D. virilis elements Ulysses and Penelope (Scheinker et al. 1990; Evgen'ev et al. 1997).

Table 10. Hybrid Dysgenesis and Related Phenomena in Drosophila melanogaster.

Table 10

Hybrid Dysgenesis and Related Phenomena in Drosophila melanogaster.

Heterochromatin and Retrotransposons

A relatively unexplored area relevant to retrotransposons is the connection between retrotransposons and heterochromatin. It is clear that the centromeric heterochromatin in Drosophila is extremely rich in transposon sequences as well as other repeat sequences of a bewildering variety. For this reason, the heterochromatin has been shunned by many molecular biologists as an impenetrable evolutionary junkyard. Is this simply a gathering place for transposable elements, because they are selected against in all other genomic regions? Even students of retrotransposons tend to dismiss the importance of these large chromosomal regions because many transposon copies located in these regions appear to be inactive. For example, the inactive I factor elements found in R strains are located in the pericentric heterochromatin (Bucheton et al. 1984).

Spradling (1993) has postulated that the many transposon sequences located in heterochromatin might be responsible for large-scale elimination of heterochromatin (a form of DNA elimination). This hypothesis, while interesting, rests largely on the assumption that element-encoded transposases or other transposon proteins specifically catalyze such excisions. Specific excision is unknown in retransposable elements, and the majority (if not all) of deletional events between different retrotransposon copies appear to be mediated by the host general homologous recombination machinery and not by element-encoded proteins. Thus, such elimination, if it occurs, is unlikely to be effected by retroelements.

The LTR retrotransposon copia can respond to a variety of cellular signals; one such signal relevant to a discussion of heterochromatin is a dosage-compensation response. Dosage compensation is the mechanism used to equalize X chromosome transcription in male (XY) and female (XX) cells. Hiebert and Birchler (1992) reported that copia elements on the X chromosome can dosage-compensate, suggesting yet another way that retrotransposons have tapped into host regulatory circuits.

Retrotransposons and Telomere Repair

Heterochromatin is also associated with telomeres, and recent developments suggest that two families of Drosophila poly(A) retrotransposons, HeT-A (Biessman et al. 1992) and TART (Levis et al. 1993), that inhabit telomeric heterochromatin can have special roles. Both of these elements have been shown to be capable of transposing onto broken chromosome ends. Furthermore, recent analysis of a natural Drosophila 14-kb telomeric sequence shows it to consist of two copies of each of these elements. The proposed mechanism for poly(A) retrotransposon insertion can be modified slightly to allow attachment of new (retrotransposon) DNA sequences to the termini of chromosomes via a reverse transcription process (Levis et al. 1993). This is especially interesting given that the telomeres of other organisms are synthesized by a specialized reverse transcription process mediated by telomerase, a reverse transcriptase carrying its own specialized template (discussed below). Most insects have a version of the short direct telomerase-dependent repeats found at the extreme chromosome termini of mammals and fungi. Interestingly, a poly(A) retrotransposon of the silkworm, Bombyx, is apparently sequence-specific for this telomeric repeat (Okazaki et al. 1995). Drosophila and other Diptera appear to lack a conventional telomerase-generated repeat sequence (Okazaki et al. 1992). Thus, it may be that retrotransposons took over telomerase function in Diptera.

Drosophila teleomeres appear to represent the first case in which retrotransposition is directly beneficial in the everyday processes of the host. Because this process is known only in Drosophila and because there appears to be an alternative mechanism operating to generate telomere synthesis in most eukaryotic cells, it is not clear how widespread this property of poly(A) retrotransposons is. However, poly(A)-type retrotransposons could function as a back up to telomerase, because unlike telomerase, retrotransposon RTs do not have a sequence-specific primer requirement. Thus, these elements might also serve a “broken chromosome repair system” in other organisms. Recent data indicate that both Ty1 and heterologously expressed human L1 can patch double-strand breaks in appropriately engineered yeast cells (Moore and Haber 1996; Teng et al. 1996). Many other retrotransposons and repetitive elements (such as Y′ in yeast; Lundblad and Blackburn 1993) are associated with chromosome termini, but their role is less clearly defined.

RIPping and Retrotransposons in Neurospora

Some species may have evolved mechanisms specifically to protect their genomes against the onslaught of transposable elements. For example, a phenomenon called RIP (repeat-induced point mutation) exists in the filamentous fungus Neurospora in which duplicated sequences are subject to a very high rate of mutation of the GC→AT-type prior to meiosis. These changes are correlated with extensive changes in the degree of methylation of these regions as well. This system, the detailed molecular mechanism of which remains uncertain, may be a system designed to inactivate transposons. Interestingly, some regions of the genome containing repeated sequences, such as the rDNA, are immune to RIP (Selker et al. 1987; Cambareri et al. 1989).

Despite this potent DNA repeat-destroying system, a poly(A) retrotransposon family with many members called Tad was identified in a single strain of Neurospora called “Adiopodoumé” after the African location where it was originally isolated. Curiously, Tad is absent from hundreds of other Neurospora isolates. When the Adiopodoumé strain is crossed to other Neurospora strains, elevated genetic instability is observed among the progeny, a phenomenon not dissimilar to aspects of hybrid dysgenesis. Among these alterations are mutations caused by transpositions of the Tad element (Kinsey 1989; Kinsey and Helber 1989).

Recent sequencing of an endogenous Tad element from the Adiopodoumé genome and comparison to a newly transposed copy resulting from a cross between Adiopodoumé and a laboratory strain revealed ten nucleotide substitutions; of these, seven were GC→AT. These may have resulted from the RIPping process, although the expected frequency of point mutation in a sequence as long as Tad (based on studies of RIPping in other repeat sequences) would be expected to be considerably higher. This phenomenon suggests that Tad elements somehow avoid high levels of RIP that might inactivate other transposons. Possible explanations include the following: (1) The Adiopodoumé strain may carry a dominant mutation that greatly reduces the RIPping frequency, (2) the Tad element may have evolved a resistance to RIPping, and (3) Tad “master copies” located in RIP-immune regions of the genome maintain the Tad family.

Retrotransposons Can Masquerade as Introns

Because transposons can insertionally inactivate genes, various mechanisms have arisen to minimize the potentially deleterious consequences of such insertions. One mechanism that appears to operate both in plants (Wessler 1989; Varagona et al. 1992) and in insects (Fridell et al. 1990; Pret and Searles 1991) is the ability of transposons (including retrotransposons) to be spliced out (imprecisely) at the RNA level from the transcripts that they interrupt. In addition, a number of cases exist in which there is a low-level transcriptional readthrough of retrotransposons inserted within introns of a target gene (Bingham and Zachar 1989). There may be a selective advantage to transposons to have slightly leaky 3′-end formation signals; in this way, insertions that would otherwise be lethal may be tolerated. These may represent additional retrotransposon adaptations, allowing relatively random transposition without completely abrogating host gene expression.

Retroelements as Portable Regions of Homology That Destabilize Host Genomes

Like all repeated DNA sequences, retroelements are potent agents of genomic instability. This is not merely a consequence of transposition; retroelements are also passive participants in homology-dependent transactions of many types, including the formation of deletions, inversions, and translocations, as well as more complex rearrangements. Many of these are mediated by ectopic (i.e., nonallelic) recombination events involving transposon copies at distinct loci on the same or different chromosomes. These transposon-punctuated chromosome aberrations have been observed in fungi, insects, and mammals (Copeland et al. 1983; Davis et al. 1987; Kupiec and Petes 1988a; Petes 1991; Sutton and Liebman 1992).

Although transposon-mediated genome fluidity is often perceived as dangerous and deleterious, it is also likely to be one of the major mechanisms by which gross chromosome structural evolution occurs. These major changes in chromosome structure and even chromosome number are likely to be important genetic factors driving speciation.

Studies in yeast indicate that ectopic recombination events mediated by artificial sequence duplications occur at surprisingly high frequencies. However, rates of ectopic recombination involving Ty element copies appear to be considerably lower, both in mitotic cells and in meiotic cells (Kupiec and Petes 1988b). Thus, there may be factors that limit the rates of these events such that within a species, gross genome structure is maintained.

Recent studies indicate that gene conversion events involving Ty elements can be mediated directly by chromosomal DNA or, in some cases, may be mediated by reverse transcripts themselves (Melamed et al. 1992). Even non-Ty1 sequences can be demonstrated to undergo reverse-transcription-mediated gene conversion under special conditions (Derr and Strathern 1993).

The argument can be made that these mechanisms cannot apply to mammalian genomes, which contain hundreds of thousands of L1 and Alu-like repeats, but in which such rearrangements are rarely observed. In fact, such events do occur, and a variety of rearrangements, both homology-dependent and homology-independent, have been associated with disease states and other polymorphisms (Pousi et al. 1994; Labuda et al. 1995; Rüdiger et al. 1995). On the other hand, the extreme sequence heterogeneity of these elements (as well as the short length of the RNA polymerase III retrotranscripts) may explain why even more rearrangements of this type are not observed. It remains possible that such rearrangements do occur frequently among pairs of very closely related element copies.

Copyright © 1997, Cold Spring Harbor Laboratory Press.
Bookshelf ID: NBK19447