The Wayback Machine - https://web.archive.org/web/20060111010249/http://engels.genetics.wisc.edu:80/Pelements/Pt.html
Modified from a chapter in Transposable Elements, edited by H.
Saedler and A. Gierl. Springer-Verlag ,Berlin.
pp. 103-123
The Drosophila genome has many families of transposable elements (Flybase), some of which have been studied in detail, and others are known only superficially (Berg and Howe 1989). Particular attention has been given to the P family (reviewed by Engels 1989), which has been the subject of intensive research for nearly two decades . There are two reasons for this special interest. First, the population biology and recent evolutionary history of P elements suggests a remarkable scenario of horizontal transfer from another species into D. melanogaster followed by rapid spread through the global population. The other reason is the wide array of technical applications that have made P elements an indispensable tool for manipulating the Drosophila genome.
The structure of an autonomous P element is shown in Fig. 1 . The 2907-bp sequence (available from Genbank ) features a perfect 31-bp terminal inverted repeat and an 11-bp subterminal inverted repeat (O'Hare and Rubin 1983). These repeats are needed in cis for efficient transposition, but they are not sufficient for it (Mullins, Rio and Rubin 1989). Internally, there are other repeat units of unknown function plus a transposase gene composed of four exons. This gene is required in trans for transposition, and part of the gene is also involved in regulation of P mobility (Rio 1990, Rio and Rubin 1988).
Nonautonomous P elements also exist. Some occur naturally through internal deletions of the autonomous elements, as shown in Fig. 1 . Such elements lack the transposase gene but retain the parts of the sequence required in cis for transposition. Mobilization of nonautonomous P elements occurs only if there is at least one autonomous P element present to supply transposase. Many artificial nonautonomous P elements have also been created in which the transposase gene has been replaced by another gene of interest, often functioning as a marker or reporter. Such elements are discussed further below.
When P elements are mobilized they produce a syndrome of traits known collectively as hybrid dysgenesis (Kidwell, Kidwell and Sved 1977). These traits include temperature-dependent sterility, elevated rates of mutation, chromosome rearrangement, and recombination. The syndrome is usually seen only in the progeny of males with autonomous P elements and females that lack P elements. These two kinds of strains are called "P" and "M" because they contribute paternally and maternally respectively to hybrid dysgenesis. The reciprocal cross, P(female) x M(male), yields hybrids in which the dysgenic traits are much reduced, due to the maternal component of P element regulation by cytotype, as will be discussed below.
The dysgenic traits can be explained largely by genomic changes due to P element transposition and excision in developing germ cells. The sterility is due to loss of germ cells early in development (Engels and Preston 1979, Kidwell and Novy 1979, Niki 1986, Niki and Chigusa 1986, Wei, Oliver and Mahowald 1991). It is more pronounced in females, where there are fewer germ cells to spare than in males, and at temperatures above 25°C. The mutations come about through several mechanisms, but are primarily P insertions into genes and imprecise excision of P elements near genes (Rubin, Kidwell and Bingham 1982, Salz, Cline and Schedl 1987, Tsubota, Ashburner and Schedl 1985). Chromosome rearrangements usually result from breakage at the sites of two or more P element insertions, followed by rejoining of the chromosome segments in a different order (Engels and Preston 1981, Engels and Preston 1984, Roiha, Rubin and O'Hare 1988). P-induced recombination occurs preferentially in the genetic intervals containing mobile P elements (Sved et al. 1991, Sved, Eggleston and Engels 1990), and usually within 2 kb of the insertion site (Preston and Engels 1996, Preston, Sved and Engels 1996).
P element mobilization happens throughout development of the germline. Most mutations, rearrangements and recombination events occur prior to meiosis (Engels 1979, Hiraizumi 1979), but some meiotic events have also been detected (Daniels and Chovnick 1993). Premeiotic events tend to be recovered in clusters of two or more aberrant individuals among the progeny of a single dysgenic parent. The premeiotic timing of these events places a limitation on the genomic changes that can be recovered in the next generation, since the product must be cell-viable in the germline. Thus, mutations or rearrangements that do not yield viable germ cells in the parent will not result in functional gametes, and therefore will not be recovered in the next generation, regardless of whether the hypothetical progeny bearing these changes would have been viable. In addition, premeiotic events result in frequency data that cannot be analyzed reliably by standard statistical methods based on Poisson or binomial distributions, and more robust alternatives must be employed (Engels 1979).
It is now widely believed that P elements have existed in the D. melanogaster genome for less than 100 years. According to this view, the elements were introduced through a rare horizontal transmission event in which one or more autonomous P element copies were acquired by D. melanogaster from another Drosophila species. The elements then spread strictly by heredity and transposition to become ubiquitous in natural populations within a few decades. This startling scenario was proposed by Kidwell (1979, 1983) and recently reviewed (see Engels 1992, Kidwell 1993) to explain the observation that the only true M strains were old laboratory stocks dating back to the early days of Drosophila genetics. Such strains would be reproductively isolated from natural populations, and thus escaped the P element invasion.
An alternative way to explain M strains was to postulate that the elements have existed in D. melanogaster over evolutionary time, but some aspect of laboratory culture conditions, such as small population size, acted to remove the P elements from the genome over several thousand generations (Engels 1981). The question was resolved when P elements from other Drosophila species were examined (Fig. 2 ). Some species closely related to melanogaster lacked P elements, but several much more distant relatives had them. In particular, the DNA sequence of a P element from D. willistoni was nearly identical to the melanogaster sequence, differing by only one base pair among 2907. Such conservation would be impossible over the 60 million years that the two species have diverged. It implies that P elements in the two species had a common ancestor in recent historical times.
D. melanogaster is now a cosmopolitan species, but it is thought to have evolved in western Africa (Lachaise et al. 1988). The species became established elsewhere only when human commercial shipping provided a means for long distance migration (Johnson 1913, Sturtevant 1921). Meanwhile, D. willistoni and related species evolved primarily in Central and South America, and are still endemic to these regions (Ashburner 1989). Therefore, they had no contact with melanogaster until the latter species arrived in the Americas. Johnson (1913), by examining antique insect collections, estimates that the first appearance of D. melanogaster in the New World occurred in the early 1800's, and they became widespread by the end of the century.
The horizontal transfer event could have occurred at any time since melanogaster and willistoni became sympatric, but the spread of P elements through melanogaster was presumably not yet complete by the 1930's when the last laboratory M populations were established (Kidwell 1983).
Currently, natural populations of D. melanogaster all appear to have P elements, including populations in such remote sites as the mountainous regions of central Asia (S. Nuzhdin and W. Engels, unpublished). However, the type and number of P elements show geographical differences (Anxolabéhère et al. 1984, Anxolabéhère, Kidwell and Périquet 1988, Anxolabéhère et al. 1985, Boussy et al. 1988, Kidwell, Frydryk and Novy 1983). For example, a North American population had 30-50 P elements in scattered chromosomal locations, with approximately two-thirds of them being nonautonomous (O'Hare et al. 1992, O'Hare and Rubin 1983). Samples from other parts of the world, however, especially near the Mediterranean, show fewer P elements in the genome and a higher proportion of nonautonomous ones (Anxolabéhère et al. 1988, Anxolabéhère et al. 1985, Black et al. 1987). It is not known how such population differences are related to the invasion history of P elements.
The mechanism of the horizontal transfer between D. willistoni and D. melanogaster is also unknown. Speculation has focused on vector organisms, such as viruses and mites (Engels 1992, Houck et al. 1991). One parasitic mite species, Proctolaelaps regalis, has been studied as a possible DNA vector, and shows various features that make it compatible with this role (Houck et al. 1991). Once an autonomous P element had been introduced into D. melanogaster by whatever means, its ability to spread through the species was undoubtedly facilitated by a transposition mechanism in which DNA gap repair acts to increase the P element copy number, as discussed below.
P elements have a long evolutionary history in diptera prior to the invasion of D. melanogaster (Clark, Maddison and Kidwell 1994, Hagemann, Miller and Pinsker 1994, Lansman et al. 1987, Lansman et al. 1985). In the D. willistoni genome, for example, there are many "dead" P elements whose DNA sequences have accumulated numerous frameshifts and substitutions preventing them from either making transposase or serving as its substrate (Daniels et al. 1990). One species group was found to have a single genomic site where a portion of the P sequence was tandemly repeated (Miller et al. 1992, Paricio et al. 1991). This sequence lacked both P element termini and was incapable of encoding the P transposase, but it did encode a truncated transposase protein, which, as discussed below, acts as a repressor of P mobility. P-like elements have also been identified in several species of other genera and even outside the Drosophilidae family (Anxolabéhère, Nouaud and Périquet 1985, Anxolabéhère and Périquet 1987, Perkins and Howells 1992, Simonelig and Anxolabehere 1991). There is preliminary evidence that distant relatives of the P element are common among diptera (H. Robertson, personal communication).
Horizontal transfer and genomic invasion are probably not unusual in the world of transposable elements. The best example is that of the mariner element, which seems to have spread throughout the animal kingdom (Robertson 1993). However, horizontal transfer events involving mariner, though frequent on an evolutionary time scale, are typically separated by millions of years. It is notable, therefore, that P elements invaded D. melanogaster within a few decades after the opportunity arose, and spread throughout the species in less than 200 years.
What are the consequences to a species when a new transposable element invades its genome? Despite some arguments to the contrary (McDonald 1993, Syvanen 1984), most evidence suggests that the harmful mutations and chromosome rearrangements produced by transposition far outweigh any beneficial mutations that might also arise (Charlesworth and Langley 1989). In one series of experiments, P element invasion and rapid expansion in inbred laboratory M strains led to extinction of the lines within 20 generations (Preston and Engels 1989). The only exception was a case in which the population was expanded sufficiently to allow natural selection to eliminate deleterious insertions more efficiently. The ability of P elements to produce a negative regulator of their own mobility (discussed below) undoubtedly gave D. melanogaster a better chance of surviving its recent P element invasion by reducing the equilibrium copy number (Brookfield 1991, Charlesworth and Langley 1989).
Several lines of evidence show that P elements transpose non replicatively and without an RNA intermediate (Engels et al. 1990, Kaufman and Rio 1992). The donor element is excised and reinserted into a recipient site creating a direct duplication of 8 bp at the site of insertion (O'Hare and Rubin 1983). The transposition reaction can be carried out in a cell-free system with partially purified transposase (Kaufman and Rio 1992), but host-encoded factors might also facilitate the reaction in vivo (Kaufman, Doll and Rio 1989).
P element insertions have been found at thousands of genomic positions, but not all sites are equally likely to be hit. The mechanism of insertion site selection is not known, but several generalizations can be made: (i) Euchromatic sites are hit more often than the heterochromatin (Berg and Spradling 1991, Engels 1989); (ii) Some euchromatic loci are much more susceptible to P mutagenesis than others. For example, the singed gene is hit at frequencies approaching 10-2 (Green 1977, Robertson et al. 1988, Simmons et al. 1984) whereas the vestigial gene has a rate of less than 10-6 (Williams and Bell 1988). Despite this variability, there is no evidence that any loci are immune from P element mutagenesis given a sufficiently large sample size; (iii) Within genes there is a preference for insertion in the non coding upstream sequences (Kelley et al. 1987); (iv) Target sites with close matches to the consensus octamer GGCCAGAC are more likely to receive P element insertions (O'Hare et al. 1992, O'Hare and Rubin 1983); (v) P elements tend to insert into or near other P elements, with a particular preference for base pairs 19-26 of the target P element (Eggleston 1990); (vi) Some P elements have been observed to jump preferentially to sites closely linked to the donor site (Golic 1994, Tower et al. 1993).
The P element transposase is an 87 kiloDalton protein encoded by autonomous P elements (Fig. 1 ). It binds to subterminal regions at both ends of the element and represses transcription (Kaufman et al. 1989, Kaufman and Rio 1991). GTP is also bound by the transposase, and is required for transposition in vitro (Kaufman and Rio 1992).
There is now considerable evidence that P element transposition leaves behind a double-strand DNA break. Sequences homologous to the flanking DNA are then copied in to repair the break (Engels et al. 1990, Gloor et al. 1991). The relative frequencies of transposition and excision suggest that approximately 85% of these repair events utilize a sister chromatid for the template (Engels et al. 1990). Thus the donor element is replaced by an identical P element copied in from the sister chromatid (Fig. 3 ). In such cases, the end result of transposition is a net gain of one P element copy. This net gain is probably responsible for the ability of P elements to increase their copy number in nature and in experimental populations (Engels 1992, Good et al. 1989, Kiyasu and Kidwell 1984, Meister and Grigliatti 1993, Preston and Engels 1989).
In the remaining 15% of repair events in which the sister chromatid is not used, the template can be either the homologous chromosome or an ectopic sequence, such as a transgene (Engels et al. 1990, Gloor et al. 1991). The homolog is used preferentially in such cases, especially if it contains a P element or fragment of a P element at the site corresponding to the break (Johnson-Schlitz and Engels 1993). The tendency to copy from a P-bearing template might account for some of the preferential use of the sister chromatid, which necessarily contains a P element at the site. There is high sensitivity to mismatches between the sequence flanking the break and the template. Even 0.5% mismatches is sufficient to decrease the rate of repair three fold (Nassif and Engels 1993). Templates located on the same chromosome as the break are also used preferentially, even if the template lies on the opposite end of the chromosome (Engels, Preston and Johnson-Schlitz 1994).
Various types of imprecise excision can be attributed to aberrant repair. The most frequent such events are internal deletions resulting in structures similar to the nonautonomous P elements shown in Fig. 1 . Breakpoints commonly occur at direct repeats of three or more base pairs, resulting in loss of one copy and the intervening sequence (Eggleston 1990, Engels 1989, O'Hare and Rubin 1983). When longer direct repeats are present, the frequency is greatly increased (Paques and Wegnez 1993). The most frequent kind of internal deletion leaves only 10-20 base pairs from each terminus, resulting in a non mobile "footprint" (Johnson-Schlitz and Engels 1993, Searles et al. 1982, Takasu-Ishikawa, Yoshihara and Hotta 1992). The term "internal deletion" might be a misnomer, since the events are probably due to a complete deletion of the P element followed by incomplete gap filling from the sister chromatid (Gloor et al. 1991). This interpretation is strengthened by the observation that when the P element resides on an extrachromosomal plasmid, and therefore lacks a sister chromatid for a template, the resulting footprints rarely contain more than four base pairs from each terminus (O'Brochta, Gomez and Handler 1991). The four or fewer bases that remain from each end could be explained if excision occurs by a staggered cut. Finally, imprecise P excisions that remove flanking DNA (Salz et al. 1987) can also be interpreted as aberrant repair events by assuming the gap left by a P element excision can be subsequently widened to varying degrees.
P elements are not normally mobile in somatic cells, and their germline mobility does not occur within P strains. These two restrictions come about by different mechanisms.
Repression of P element transposition in somatic cells occurs on the level of RNA processing (Laski, Rio and Rubin 1986). The 2-3 intron (Fig. 1 ) is spliced only in the germ cells, resulting in the absence of transposase in somatic cells. Splicing of this intron is prevented in the somatic cells by a 97-kD protein that binds to a site in exon 2 located 12 to 31 bases from the 5´ splice site (Chain et al. 1991, Siebel and Rio 1990, Tseng et al. 1991). When the 2-3 intron is removed artificially, the resulting transposase gene, designated Æ2-3, produces functional transposase in both somatic and germline cells, and P elements are mobile in all tissues (Laski et al. 1986). This mobility results in pupal lethality if several mobile P elements are also present (Engels et al. 1987), especially in a DNA repair-deficient background (Banga, Velazquez and Boyd 1991).
The quiescent state of P elements within established P strains is best attributed to element-encoded repressor products. These repressors fall into at least two discrete categories, designated Type I and Type II.
As noted earlier, P elements are repressed not only within P strains, but also in the hybrids of P(female) x M(male) crosses. They are not repressed in hybrids from the reciprocal cross (Kidwell et al. 1977), suggesting that repressor products produced in the P strain germline can be inherited maternally. Interestingly, this effect goes beyond simple maternal inheritance. Progeny from the cross MP(female) x P(male) show more P mobility than those from the cross PM(female) x P(male) (Engels 1979), where "MP" and "PM" represent the two reciprocal F1 hybrids with the female component shown first. Thus, the repressed state, called the P cytotype, is jointly determined by chromosomal and maternal components. One explanation proposed for this unusual inheritance was that the repressor-making P elements in the MP hybrids are more likely to be excised than those of the PM hybrids (Misra et al. 1993, Misra and Rio 1990). However, that explanation was ruled out by the finding that a single repressor-producing P element was sufficient for this mode of inheritance (Ronsseray, Lemaitre and Coen 1993). An alternative model (Lemaitre, Ronsseray and Coen 1993) involving differential splicing of the same intron that is involved in tissue-specific regulation, provides an adequate explanation for available data, and will be discussed further below.
Various lines of evidence indicated that the 66 kD truncated transposase protein that is made when the 2-3 intron is unspliced functions as a repressor of P mobility (Gloor et al. 1993, Handler, Gomez and O'Brochta 1993, Misra et al. 1993, Misra and Rio 1990, Rio, Laski and Rubin 1986). Several other truncated transposase molecules with breakpoints in slightly different places can also function in this way (Gloor et al. 1993, Robertson and Engels 1989). Fine structure deletion mapping revealed that the minimal 3´ boundary for this kind of repressor was between nucleotides 1950 and 1956 of the P element sequence (Gloor et al. 1993). These repressors are designated Type I to distinguish them from a class of much smaller truncated transposase proteins called Type II, discussed below.
The use of a reporter gene fused to the P element's transposase promoter showed that this repression acted on the level of transcription (Lemaitre and Coen 1991, Lemaitre et al. 1993). This effect also provided an explanation for several observations that P repressors affected expression of genes neighboring P element insertion sites (Coen 1990, Engels 1979, Gloor et al. 1993, Robertson and Engels 1989, Williams, Pappu and Bell 1988). Finally, the transcriptional regulation revealed by the reporter gene showed a maternal effect in the germline but not in somatic cells (Lemaitre et al. 1993).
These observations led Lemaitre et al. (1993) to propose a model in which the 97-kD protein previously shown to prevent splicing of the 2-3 intron in somatic cells was also present germinally, but in much reduced quantities. Thus, in the P cytotype, transcription of the transposase message is relatively low, and the available germline splice blocker is sufficient to ensure that only 66-kD product is made. This 66-kD product serves as a transcriptional repressor in the germline to perpetuate the P cytotype through the female lineage as long as there are sufficiently many repressor-producing P elements on the chromosomes. In the M cytotype, the level of expression is increased sufficiently to overwhelm the splice-blocking agent in the germline, and functional 87-kD transposase is made. This model requires a nonlinear relationship between transcript level and the ratio of 66 to 87 kD products in order to explain the maintenance of M cytotype.
The number of distinct nonautonomous P elements in nature is so large that few have been observed in more than one population (O'Hare et al. 1992, O'Hare and Rubin 1983). The first and most conspicuous exception to this rule is the KP element, which is very common worldwide (Black et al. 1987). It was therefore suggested that KP elements might function as P element repressors, and thus be favored by natural selection (Black et al. 1987, Jackson, Black and Dover 1988). This possibility was verified when KP elements were isolated genetically and tested for repressor (Rasmusson, Raymond and Simmons 1993). KP elements did not fit the paradigm of Type I elements for two reasons: First, they had a deletion for nucleotides 808-2560 (Fig. 1 ), and thus lacked the minimal sequence previously shown to be required for Type I repressors (Gloor et al. 1993). Second, the repression in KP lines showed none of the maternal inheritance associated with cytotype (Rasmusson et al. 1993, Raymond et al. 1991). This lack of maternal inheritance is expected under the splice-blocking model of cytotype described above because the 2-3 intron is not present in KP elements.
It soon became clear that KP was not the only other element of this kind. The D50 element had similar repressor properties but slightly different deletion endpoints (Rasmusson et al. 1993). To date, five such elements, designated Type II repressor-makers, have been identified, at least four of which are geographically widespread (C. Preston, G. Gloor and W. Engels, unpublished). All have large deletions whose endpoints are similar to (within 300 bp) the KP deletion endpoints. A base substitution at either nucleotide position 32 or 33 is also present in most Type II repressors.
It is not known whether Type II regulation works by the same mechanism as Type I, but they do share several features in common. Type II regulation probably also acts on the level of transcription, since the elements can reduce the expression of a reporter gene fused to the P promoter (Lemaitre et al. 1993), and they display the secondary effects on expression of genes closely linked to P insertions (C. Preston and W. Engels, unpublished). For both types, there is a pronounced position effect such that the ability of the element to function as a repressor is highly sensitive to its genomic insertion site (Gloor et al. 1993, Higuet, Anxolabéhère and Nouaud 1992, Misra et al. 1993, Robertson and Engels 1989, Ronsseray, Lehmann and Anxolabéhère 1991).
Drosophila has long been a favorite organism for genetic and developmental research, but it was largely through the use of P elements that the powerful tools of molecular biology were fully employed. P elements are used for identifying genes of interest, cloning them, and for placing them back into the genome. There are several key features of P element biology that make them especially well suited for these roles. The existence of M strains allows experimenters to create stocks containing only selected P elements. Transposase can easily be added or removed genetically. The high mobility of P elements and their retention of this mobility despite drastic modifications to their internal sequences are also essential features. Most recently, the double-strand DNA breaks created by P element excision have been used to effect gene replacement and to study the repair process.
A basic problem in genetics has long been how to obtain molecular information for a gene known only by the phenotype of its mutations. The reverse problem, obtaining mutations in a gene known only by its DNA sequence is also becoming increasingly common. Oddly enough, both problems often have the same solution for Drosophila geneticists: obtain a P element insertion into the gene. In the former case, the P insertion allows cloning by transposon tagging (Bingham, Kidwell and Rubin 1982, Searles et al. 1982), and, in the latter, P element insertion mutations can be selected by PCR-based methods without knowledge of the phenotype (see below). For these reasons, the search for P element insertion mutations engages much of the effort of Drosophila workers regardless of their particular biological focus (Kidwell 1986).
The most efficient approach to P element mutagenesis utilizes an immobile copy of the transposase gene combined with one or more mobile, but nonautonomous, P elements to serve as "ammunition". The P{ry+ 2-3}(99B) element is an example of an immobile transposase source that has been widely used for P mutagenesis (Robertson et al. 1988). Its transposase gene lacks the 2-3 intron, thus precluding the production of any 66 kD protein, which, as discussed above, functions as a repressor. This element cannot transpose due to a deletion of one of its termini (H. Robertson, personal communication). A cross with the transposase source coming from one parent and the ammunition elements from the other yields progeny in whose germ cells mutagenesis occurs. The transposase source is then eliminated by segregation in the next generation to stabilize any mutations obtained.
There are two general strategies in the selection of ammunition elements. One is to use a chromosome carrying as many highly mobile elements as possible, such as the Birm2 chromosome (Engels et al. 1987, Robertson et al. 1988) which has 17 small nonautonomous P elements from nature. This approach maximizes the likelihood of obtaining the desired mutation, but it is often laborious to isolate the mutation from the rest of the P elements in the genome. Alternatively, one can use a smaller number of artificially constructed P elements (Cooley, Kelley and Spradling 1988). This method usually requires a larger screen, but any mutation obtained is easier to isolate, especially if the ammunition element(s) carries a bacterial origin of replication and selectable marker to permit cloning by plasmid rescue.
PCR can be employed to screen for P insertions if no phenotypic screen is available but the target gene has been cloned. DNA is extracted from a pool of potential mutants and amplified with one primer in the P element sequence and the other in the targeted DNA. A P element insertion close to the targeted site is required to bring sites for these two primers together, and yield amplification (Ballinger and Benzer 1989, Kaiser and Goodwin 1990). The target size of this approach is necessarily small because PCR cannot amplify more than a few kb. An alternative approach that permits a much larger target size is provided by inverse PCR (Sentry and Kaiser 1994). Here the two primers are both within the P element sequence, but directed away from each other. Amplification can occur when the newly inserted P element, along with some of the flanking DNA is circularized following digestion with a restriction enzyme and ligation (Ochman, Gerber and Hartl 1988). DNA amplified in this way from a pool of potential mutants can then be probed with target DNA of arbitrary length, and insertion mutations can be identified.
Once a P insertion has been obtained in or near the gene of interest, additional genetic variability can be generated readily by the reintroduction of transposase. Internal deletions or flanking deletions can be selected (Salz et al. 1987, Tsubota and Schedl 1986). In some cases, transposase can catalyze an event in which one P element in the genome is substituted for another P element elsewhere in the genome by an unknown mechanism (Staveley et al. 1994). This process can be useful for putting a reporter gene into a specific site.
The most important use of P elements is undoubtedly that of making transgenic flies (Rubin and Spradling 1982, Spradling and Rubin 1982). The gene of interest is placed between P element ends, usually within a plasmid, and injected into pre-blastoderm embryos in the presence of transposase. This P element, with the gene as cargo, then transposes from the plasmid to a random chromosomal site. Technical aspects of the method have been described elsewhere (Ashburner 1989, Spradling 1986). In a typical experiment, 10-20% of the fertile injected flies produce transformant progeny.
The P element may also carry a second gene used to identify transformants. The frequency of transformation is usually sufficiently great that a visible marker, such as an eye color gene (Pirrotta 1988, Rubin and Spradling 1982) is more efficient than a selectable marker, such as neomycin resistance. The size of the inserted sequence can exceed 40 kb (Haenlin et al. 1985), but such large vectors come at a cost of decreased transformation frequency. In some cases, the sequence carried by the P element can influence the transformation rate (Spradling 1986) or the insertion site specificity (Kassis et al. 1992).
There are several options for providing transposase to the injected DNA. One way is to bind purified transposase protein to the element prior to injection (Kaufman and Rio 1992). However, the difficulty in obtaining transposase in sufficient quantities usually makes this method impractical. Alternatively, one can coinject a transposase-making "helper" plasmid, preferably one that is unable to integrate into the chromosomes itself (Karess and Rubin 1984). A third approach is to inject directly into embryos that have an endogenous transposase source, such as the P{ry+ 2-3}(99B)element mentioned previously (Robertson et al. 1988). The transposase-bearing chromosome can be marked with a dominant mutation, and stable transformants lacking the transposase gene are then selected among the progeny. This procedure is probably more efficient than coinjection, since it does not require the embryonic nuclei to take up two independent plasmids.
P element mobility also provides a way to sample the genome for loci whose expression matches a particular pattern (O'Kane and Gehring 1987). A lacZ reporter gene is fused to a weak promoter and mobilized within a P element to produce a collection of Drosophila lines, each with a single insertion of the "enhancer trap" element at a random site. The expression pattern of lacZ in each line tends to reflect the expression of nearby genes. Thus, one can identify genes that are active in specific tissues and developmental periods. The power of this technique increases rapidly with time, as large collections of enhancer trap lines become available (eg., Hartenstein and Jan 1992), thus eliminating the need for each worker to produce a new collection.
Several P element vectors are available to facilitate expression of a given gene in a particular tissue through fusion of the gene to a specific promoter. For example, one set of vectors includes a promoter for strong expression in the developing egg and early embryo (Serano et al. 1994). A particularly versitile system employs a two-element combination to allow a given gene to be expressed in any of a wide variety of patterns (Brand and Perrimon 1993). One element in this combination is similar to the enhancer trap construct discussed above except that the reporter gene is the yeast transcriptional activator, GAL4. The second element carries the gene of interest driven by a promoter containing GAL4 binding sites. The gene is then activated only in the cells where GAL4 is expressed. Thus, the expression of the gene of interest depends on the insertion site of the P{GAL4} element.
Another yeast system that has proven useful in Drosophila is the FLP site-specific recombinase, and its target site, FRT (Golic and Lindquist 1989). This system is particularly useful for generating mosaics. One P element carries the FLP recombinase gene driven by a heat shock promoter, and a second element has a gene with two FRT sites embedded. When heat shock is applied to such flies, FLP-mediated recombination causes somatic loss of the gene carrying FRT sites. More recently, FLP has been used to generate somatic mosaics with sectors homozygous for an entire chromosome arm (Xu and Rubin 1993). A homozygous P element near the base of a chromosome arm and bearing an FRT site undergoes mitotic recombination when FLP is expressed. The result is a somatic sector that is homozygous for all genes distal to the FRT-bearing P element. Such sectors can be identified by absence of a cell-autonomous marker present on one of the homologs. This method allows identification and analysis of genes that are lethal when homozygous in the whole organism.
With P element mediated transformation, as described above, the researcher has no control over where in the genome the construct goes. In many instances, however, what is needed is to replace genes in situ. For example, some genes are too large to manipulate in vitro and return to the genome by transformation. Others are too sensitive to position effects. In addition, some genes have no null alleles to provide a suitable background to test transgenes.
To achieve gene replacement, Drosophila geneticists can make use of P-induced double strand breaks (Gloor et al. 1991).The method requires construction of an altered version of the gene which will be used as the template for gap repair. This construct must contain the sequences flanking the P insertion site. As discussed earlier the gaps produced by P element excision are usually repaired by copying in sequences from the sister strand (Fig. 3 ). However, in approximately 15% of the cases, the homolog or an ectopic sequence can provide the template (Engels et al. 1990). This method has been tested most extensively in the white gene, where hundreds of gene replacement events have been analyzed, but it has also been used successfully at forked (Lankenau, Corces and Engels 1996) and at least two other loci (Papoulas, McCall and Bender 1994). The frequencies of gene replacement measured with white were dependent on the genomic position of the template, averaging 1% for autosomal sites (Gloor et al. 1991) and 6% for X-linked sites (Engels et al. 1994). Extrachromosomal templates have also been used (G. Gloor, personal communication, Banga and Boyd 1992, Papoulas et al. 1994). Insertions and deletions could be copied into the gap just as efficiently as single base pair changes (Johnson-Schlitz and Engels 1993, Nassif et al. 1994).
The primary limitation of this technique is its requirement for a P insertion close the site being modified. A site within 8 bp of the P insertion in white was replaced in close to 100% of the gene replacement events, but one 2 kb away was replaced less than 10% of the time. To a good approximation, the replacement frequency of a site n bp away from the P insertion was 0.99855n, expressed as a proportion of the gap repair events (Gloor et al. 1991). Therefore, the P mutagenesis techniques discussed above, especially the PCR-based screens (Ballinger and Benzer 1989, Kaiser and Goodwin 1990, Sentry and Kaiser 1994), are particularly valuable as preliminary steps toward gene replacement.
Recent work has shown that an efficient way to obtain deletions of the sequences flanking a P element is by selection for male recombination events (Preston et al. 1996). The mechanism is thought to be the "half-element insertion" (HEI) process postulated from work on end-deleted P elements (Gray, Tanaka and Sved 1996, Svoboda, Robson and Sved 1995).
If you are using Netscape Navigator 2 or later you can see how the HEI model works to produce recombination by viewing the animated diagram . In addition, HEI can explain P-induced inversions on nonrecombinant chromosomes. There is also an animated inversion diagram available for viewing.
The following points are useful in planning the use of this technique:
P elements are relative newcomers in the Drosophila melanogaster genome, probably arriving through a horizontal transfer event less than 200 years ago. Their invasion of the genome was almost certainly harmful to the species, lowering the average fitness throughout natural populations. However, P elements have undoubtedly enhanced the fitness of Drosophila geneticists, forming the basis for a variety of techniques that have become essential to most current research with this species.
Christine R. Preston provided many comments and suggestions in preparation of this review.