r iews

T I G - - September 1985

Unlike most of the mobile genetic elements which have been defined in Drosophila melanogaster, P elements are not ubiquitous. There are strains with P elements and strains without. Reciprocal crosses between flies with P Kevin O'Hare elements and flies lacking them produce strikingly different results (Fig. 1). When P elements are a class of mobile elements in Drosophila melanogaster whose rate of male flies with P elements are trar~sposition is under genetic control. While the basis for this control is poorly crossed with female flies underst(rrl, transposition can be s~qtched on and off by simple genetic crosses. Thus, it is lacking them, the germline of lx~ssible to investigate the mechanism and regulatitm of transposition. the progeny develops abnormally. This can be manifest as reduced fertility, elevated rates of mutation and chromosomal rearrangement, for transposition. The most likely hypothesis is that male recombination and segregation distortion. the elements themselves code for a regulator of transThese traits, collectively known as hybrid dysgenesis, position, although the physical nature of these regulaare not seen in the reciprocal cross nor in inter-se tory factors remains unknown and their pattern of incrosses of the parental strains. The very close cor- heritance is complex, showing characteristics of both relation between the various traits suggests that they chromosomal and extrachromosomal inheritance. are causally related, although this has not been A typical P cytotype strain has about 30-50 P definitively established. (For reviews, see Refs 1,2.) elements distributed throughout the major chromoMost of the mutations generated in dysgenic some arms s. Genetic studies suggested that some P crosses are caused by insertion of P elements. In fact, elements were fully functional while others were P elements were isolated by analysing mutations of defective and could produce neither a protein involved the white locus which arose in such crosses a. These in transposition (transposase) nor any regulator of mutations are stable, but revert at high rates during transposition6. P elements isolated from a P cytotype hybrid dysgenesis. Thus, transposition and/or strain were shown to be heterogeneous in size7. There excision can be triggered by simple crosses. P ele- are perhaps ten copies of a 2.9 kb element and some 30 ments are therefore found under two conditions. They smaller elements, each of which appears to have are normally quiescent in strains which have P arisen from the larger elements by an internal deleelements and such strains are said to have the P cyto- tion. All elements are bounded by inverse repeats of type4. P elements are mobilized when they are intro- 31 bp. The sequence of the 2.9 kb element has a duced into what is called the M cytotype. Strains number of large, open reading frames on one strand which lack P elements have the M cytotype. The non- which could contribute to the production of P specific reciprocal nature of the cross suggests that P cytotype proteins, and indeed RNAs of this strand have been represents either the presence of factors which sup- detected a. press transposition or the absence of factors required

The mechanism and control of P element transposition in

Drosophila melanogaster

Assays for P element functions

M9
x Po"
Dysgenic cross Temperature-sensitive sterility germline with elevated rates of mutation and chromosomal rearrangements

M9 P 9
P~

x x x

Mo" Md' Po"

Non-oysgenic crosses
No temperature-dependent stedliW

2~geneszs.

Fig. 1. The non-reciprocal nature of crosses in P - M hybrid dys-

The assays extensively used in the analysis of P element function (see Fig. 2) are based upon either the ability to induce dysgenic effects or to protect against them. If the progeny of a dysgenic cross are raised at 29C, the gonads fail to develop. This is known as gonadal dysgenesis (GD) sterility and is easily assayed by dissection of females. Thus the presence of active P elements can be determined by crossing a male of the strain in question with a standard M cytotype female at 29C and assaying for the induction of GD sterility in the female progeny. (A similar analysis can be made on the male progeny, but dissection of females is clearer because the reproductive organs are larger.) Conversely, by crossing a female of the strain in question with a standard P cytotype male, the ability to regulate P element activity can be determined by the same assay. The molecular basis of this assay is unclear, but as it only takes one generation it is widely used to characterize strains. A second set of assays is based upon the instability during hybrid dysgenesis of a mutation affecting bristle morphology, the singed-weak (sn=') mutations. This mutation arose in a dysgenic cross and has two deleted P elements inserted head-to-head at the
1985. Elaevier Science Pubiiah~ B.V., Amsterdam 0168 - 952585/$02 00

TIG-

September 1 9 8 5

review

thus t~avethe sequencesrequired in cLsfor transposase to act, but must lack sequences encoding the transposase. The most obvious sequence features of P elements likely to be involved in the interaction with transposase are the terminal inverse repeats. Indeed, the deletion of the final 22 bp from the 2.9 kb element abolishes transposition (see below and Ref. 8). However, it seems likely that more extensive sequences than the 31 bp repeats are required in cis. The insertion of DNA sequencesat position 41 apparently produces an element which no longer responds to transposaseTM. The smallest naturally occurring P elements have conserved some 200 bp from either end7, suggesting that this is the maximum sequence length required in cLsfor transposition. The extent of the sequences required to produce transposase have been investigated by Karess and Rubin8. Disruption of the 3' terminal repeat (with respect to transcription) does not affect the production of transposase.This 'wings-clipped' element has been Signals and sequences required for used to ensure that elements carrying frameshift transposition The genetics of hybrid dysgenesis suggest that the mutations in the open reading frames of the 2.9 kb P element will encode a function required for trans- element can be introduced into flies, irrespective of position and perhaps a function to regulate transposi- their own ability to produce transposase. Once estabtion. That the element does indeed encode a transpos- lished in vioo, the mutated elements were assayedfor ase (in the broadest possible sense of the term) has the production transposase.The results suggest that been most definitively shown by the microinjection all the large open reading frames contribute to experiments of Spradling and Rubin~'H. They transposase production. If the open reading frames showed that when cloned 2.9 kb elements were produce independent polypeptides which combine to injected into developing M cytotype embryos they form a multi-subunit transposase, complementation transposed from the plasmid into the chromosomes of might be expected when the mutated elements are some of the germ cells. The smaller elements could combined in trans. As no such complementation is not catalyse their own transposition, but could observed, it is likely that a single polypeptide chain is transpose when transposase was provided in trans by produced. This is consistent with the observation that co-injected 2.9 kb elements. These smaller elements sequencesfrom all the open reading frames contribute
singed (sn) locus (H. Roiha, K. O'Hare and G. M. Rubin, unpublished). During dysgenic crosses the mutation is destabilized, so that the germline of the progeny is mosaic for the sn locus and flies with new bristle phenotypes are produced in the next generation. The excision of one of the two P elements results in a more extreme phenotype while excision of the other element leads to a wild-type phenotype. As many as 50% of the germ cells of the progeny of dysgenic crosses may show changes in the sn phenotype. The potential to produce transposase can be measured by crossing a test male with females of an M cytotype strain carrying the, sn" mutation and assaying for destabilization. The ability to suppress transposition can be measured by crossing a test female with a male from a P cytotype strain carrying the sn ~"mutation. These assays have the disadvantage of taking two generations, but the molecular basis is known and is clearly related to P element activity.

(a)
F1

? /
9C

PC/j Dissect ~3~ and scoreovaries

sn
w

29C l
8n
w

(b)

(~ Ii
W t

113 3 I pc//
_

M~
sn-

V'//A sn

sn

sn

sn

W"

? .~-~
SBW

SB V/ /A

W~

sn -

sn W"

sn w*

sn

r-"--t
sn -

and f~l

0,4

~ZZ2
sn-

1"

sn w *

sn w*

v.,/,(~ and ~ d
sn -

Scorefor s n and s n "

Fig. 2. Assays far P element functions: only the X (r"'l) and Y (IS:I) chromosomes are stu~um ]or simplicity; I l l = strain to be assayed, I"'1 = standard P cytotype strain (e.g. Harwich, he), C~l = standard M cytotype strain (e.g. Canton S, Oregan I0, ~ = standard s n strain (e.g. sn3). The left panels show assaysfar the potential to suppress transpositions, while the right panels show assays for functional P elements. (a) Gonadal dysgermsis ((319)sterility. The crosses are made at 29C and the female progeny dissected and scared far the absence of ovaries. (b) Destabilization ofsnw. The crosses"are as indicated. In many cases the strain to be a~ayed t*qll be sn ", so onlysn~progeny can be scored in the assay. I f attached X females are mated with the F z males in the right panel, then both sn and sneprogeny c a n be scared, snw"repr~senL~a destabilized snW allele pr~wnt in the germline of the Ft flies and ret~aled~ .wmalically in the Fe flies. ..."X

r iews

T I G - - September 1985

Like essentially all transposable elements, insertion of P elements leads to the duplication of a target site such that the inserted element is flanked by a tandem duplication7. This is thought to result from the introduction of a staggered break at the target site, insertion of the element and repair of the singlestranded gaps. The size and possibly the sequence specificity in the target sites are characteristic for The mechanism of transposition different transposable elements and are thought to be As for prokaryotic transposable elements, it is likely properties of the transposase machinery. The that an in vitro system for P element transposition will observation that excision of P elements can be precise be needed to answer detailed questions about the at the nucleotide level, suggests that the same mechanism of transposition. There is little prospect staggered cut is ma.de at both ends of the element and for purifying P transposase from embryos produced in the two serf-complementary sticky ends are ligated a dysgenic cross. The abundance of P-specific together to restore the original sequence and perhaps transcripts is very lows, so the abundance of the produce a circular P element. While such circles protein is likely to be equally low. Indeed, there is remain hypothetical, they could be transposition genetic evidence that the abundance of transposase intermediates or have a role in the regulation of may limit the overall rate of transposition (see below). transposition (see below). A more promising approach is to express the Chromosomal rearrangements produced in hybrid sequences for transposase in a bacterial system. To dysgenesis are almost exclusively between sites this end, it should be possible to isolate complete where P elements reside TM. The types of complex cDNAs for the P transcripts and clone them into an multi-point rearrangements produced cannot be appropriate vector for expression in E. coli. explained by homologous recombination, nor by any However, some information about the mechanism simple mechanism involving co-integrate structures of transposition of P elements can be obtained as proposed for some prokaryotic transposable eleindirectly. There are several arguments against a ments. Engels and Preston TM have shown that the direct role for RNA intermediates in transposition. model which fits the data best is one where there are a Firstly, the structures of P elements are unlike those number of chromosomal breaks at the sites where P of transposable elements where an RNA intermediate elements reside followed by a random reassortment of is thought to be involved. Furthermore, since the the fragments produced. major P-specific transcripts lack the first 87 nucleoThere is an overall tendency to lose P elements from tides they are unlikely to be transposition inter- the rearrangement endpointsTM, suggesting that some mediates. Although very long RNAs with homology to of the chromosome breaks are produced by excision of P elements have been detected, they vary from one P the P element. The precision of this excision has not strain to another and probably represent readthrough yet been determined at the molecular level. However, transcripts initiated in sequences flanking the in most cases P element sequences remain at some if elements. not all of the rearrangement endpoints. One inversion More indirect information comes from a considera- between two P elements which has been analysed at tion of the sorts of reactions known to occur in hybrid the molecular level (H. Roiha, K. O'Hare, and G. M. dysgenesis (see Fig. 3). These include the insertion of Rubin, unpublished) suggests that the interaction of P elements3, both precise 7 and imprecise excisionsu transposase with an integrated P element may cause a of integrated P elements, and complex chromosomal chromosome break precisely at one end of an element. rearrangements TM. All these reactions only occur The sequences flanking both sides of the P elements at when there is active transposase, suggesting intimate both ends of the inversion are entirely conserved. The involvement of transposase itself. Some of these elements are now flanked on either side by sequences reactions could represent abortive or failed transposi- from opposite ends of the inversion. One element is tions, but they can all be explained by a simple model unchanged in structure, but one end of the second in which transposase may interact independentlywith element has been lost. This suggests that there was a each end of a P element. chromosome break at one end of both elements followed by inversion of the intervening segment of the chromosome and religation, but that during this 1InsertiOn I PreciSeexcision ~ Imprecise process there was some loss of sequences from one of the elements. o< 1 ,,o This sort of reaction could also be responsible for Chromosomal the imprecise excision of P elements where variable rearrangements amounts of P element sequences are left behind u. It predicts that the sequences flanking the insertion will be conserved and abutted to internal P element sequences. One very basic question which is difficult to resolve O1 ~ __ ~ : ~ .... is whether transposition is a replicative process or not. Fig. 3. DNA rearrangements catalysed by P ehqnents. The Transpositions only occur in the germline, and it is ek,rnent~ are stu)u,n with a marker restriction en4vme site to impossible to guarantee recovery of all products of indicate orientation. The 8 bp duplicati~ms made ulxm inserti~m meiosis in D. melanogaster. Thus, the observation that two copies of an element are present in the progeny of / ~ _ an, shmen as a bin:. S('c text fi~r details.
. . . .

to the production of the RNAs detected in embryos arising from a dysgenic cross. Although this P-encoded polypeptide is necessary for transposition, it is probably not sufficient. Host factors for normal DNA repair, DNA replication and recombination are likely to be involved in transposition.

TIG -- September 1985

review

the physical nature of these factors and is discussed - ~ " below. It was recently proposed that one aspect of this complex regulation is the competition between elements for transposase ~8. Naturally-occuring strains have been described which appear to have very few, if any, copies of P elements which can produce transposase, but do have many, presumably defective, P elements5. Such strains have the M cytotype and are called M' strains. The effect of these defective P elements on the destabilization of sn ~ was measured ~s. The results show that the presence of the defective elements reduces the rate of mobilizationof the target elements at s n ' . A completely different set of experiments also indicated that the chromosomal P elements themselves can affect the action of transposase. Compared with M cytotype strains lacking P elements, M' strains do suffer less from the deleterious effects of P element mobilization, and so produce less GD sterility when crossed with P cytotype males ~9. The genes responsible for this effect map to all the major chromo-' somes, and again, they are presumed to be the P elements themselves. It is unlikely that the defective elements code for the regulators, as there is no evidence The inheritance of transpositional for maternally inherited factors in M' strains. The regulation Transposition of P elements is suppressed in P cyto- female progeny of the two reciprocal crosses between type strains but derepressed when they are introduced M strains completely lacking P elements and M' into the M cytotype. This is similar to the zygotic strains show identical behaviour when mated to P induction of ~. prophage which can occur during bac- cytotype males. This shows that the reduction in GD terial conjugation, and suggests that regulatory mole- sterility is entirely due to the M' chromosomes '9. If transposase does interact independently with the cules in the P cytotype repress transposition. These regulators would be present in eggs produced by ends of the element, these results may be rationalized strains with the P cytotype, but absent in those pro. as follows: at low ratios of transposase to P termini, duced by strains with the M cytotype. Thus, only one most elements would interact at only one end and the of the two P x M crosses would mobilize P elements, rate of excision might be lower than at high ratios explaining the non-reciprocal nature of the dysgenic when transposase would interact at both ends. Conversely, the rates of chromosome breakage and imprecross. The genes for these regulators apparently are dis- cise excision might be higher at low ratios. This has tributed throughout the chromosomes of P strain flies, not been tested. Thus chromosomal elements can be considered as as each P chromosome contributes more or less independently to the ability to protect from the effects of P both substrates for and genomic modifiers of the element mobilization~6. In fact, the simplest hypo- activity of transposase, which is encoded by a subset thesis is that the P elements themselves code for a of the same chromosomal elements, the 2.9 kb eleregulator of transposition. If so, strains into which P ments. It is hardly surprising that the genetics of elements have been introduced by micro-injection hybrid dysgenesis is so complex. should be able to produce the P cytotype. This has been observed (W. R. Engels, pers. commun.), The nature of the regulatory factor although the kinetics with which P cytotype is estabGiven the apparent limited capacity for self-replilished remain unclear. Lines containing a single copy cation, it was first suggested that episomal P elements of an intact 2.9 kb element have the M cytotype8, but are transmitted through the maternal line and trancan be surprisingly stable for P element copy number scribed and translated to produce a protein reguand position. The frequency of transposition is not as latorz. While there is no evidence for this, circular high as one would naively expect, suggesting that copies of other transposable elements have been much remains to be learnt about the control of P found in cultured cells ofD. melanogaster 2~. element transposition. Indeed, a description of control A second suggestion was that the properties of a by the elements alone would ignore the evidence that protein regulator encoded by the 2.9 kb P element extrinsic factors such as the age of the flies and the might be sufficient to account for the persistence of temperature at which they were raised have profound suppression of transposition through the maternal effects on hybrid dysgenesis ~z. line. If the regulator positively affected its own The regulatory factors themselves are maternally activity while negatively affecting transposase inherited, and this maternal inheritance can persist activity, this feedback loop could ensure that represthrough several generations. This has led to the sug- sion was maternally maintained7. gestion that the regulatory factors may be capable of The pattern of transcription of the 2.9 kb P element some limited replication independentlyof the chromo. does not shed much light on this problem. The pattern somes. How this might be accomplished depends on of transcription in natural P strains is very complex, ~ - x a fly where the parent had only one copy does not establish that transposition is replicative. It remains possible that the second copy arose by non-replicative transposition from a sister chromatid after DNA replication. While most circumstantial evidence suggests that transposition is replicative, the question is far from being resolved. The best picture of the transposition mechanism at present is perhaps that of a 'cut and paste' mechanism where transposase introduces double-stranded cuts in the target DNA and at the end(s) of an element. The sequences of the element are tiien either copied into or physically moved into the new site, depending on whether transposition is replicative or not. The copying required by a replicative version of this mechanism appears to be prone to slippage of the replication fork, resulting in deletions. This has been suggested to produce the small P elements7, and could also explain the loss of element sequences in imprecise excisions and in some chromosomal rearrangements. It might also account for the element internal deletions described by Daniels et al. ~r,

r iews

TIG -- September 1985

presumably because of the transcription of both 2.9 kb and the smaller elements, and also read-through by transcripts initiated in flanking sequences into the P elements s. A line containing a single copy of a 2.9 kb element was also examined, and two long RNAs (2.6 and 2.9 kb) were found. They contain sequences from all the open reading frames, suggesting that they arise by differential processing of the same primary transcript 7. Presumably one of these is the mRNA for the transposase, although it remains formally possible that both are required to produce the enzyme. If one is the mRNA for a regulator, then the regulator would have to share much of the amino acid sequence with the transposase. This suggests a simple model whereby both proteins bind to the 31 bp repeats, the transposase to catalyse chromosome breaks and/or transposition, the regulator to physically interfere with the action of the transposase. Given that transcription initiates around position 87, it is likely that binding of a protein at the 5' terminal repeat would affect transcription. If the mRNAs for the transposase and the regulator do arise L ' differential processing of the same primary transcript, then the positive feedback postulated to account for the inheritance of cytotype through the maternal line would have to act at the level of splicing, to produce more regulator mRNA at the expense of the production of transposase mRNA. A radically different model has been proposed where the 2.9 kb elements code only for a transposase and regulation is achieved by binding transposase to episomal P elements TM. These extrachromosomal circles would be produced by the action of transposase on chromosomal P elements, but would not be intermediates in transposition. By competing for transposase with chromosomal P elements they would suppress transposition. There is a positive feedback loop in this model because the regulatory P elements are generated by transposase, but lead to the suppression of transposition.
Prospects While much has been learnt about the molecular basis of some traits associated with hybrid dysgenesis, even the most basic questions about the mechanism and control of transposition remain unanswered. This is partly because transposition is restricted to the germline of the progeny of a dysgenic cross. It is now possible to ask if this restriction is at the level of transcription by inducing the transcription of P elements in the somatic tissues using appropriately engineered P elements. If transposition can be induced in the soma, will it be regulated as it is in the germline? With investigators using a range of techniques combining classical genetics, recombinant DNA technology and whole organism transformation, the prospects for improved understanding of this eukaryotic transposable element are both exciting and promising. Note added in proof At the 50th Cold Spring Harbor symporium on Quantitative Biology (29 May-.5 June 1985), G. M. Rubin (University of

Californiaat Berkeley)presented evidencethat the restrictionof P element activity to the germ line is due to different splicingof P transcripts in somatic and germ cells.
References 1 Kidwell,M. G. and Bregliano,J-C.(1983) Hybriddysgenesis determinants. In Mobile Geru'tic Elements (Shapiro. J., ed.), pp. 363 410, Academic Press 2 Engels,W. R. (1983) The P familyof transpo~ble elements in Drosophila. Annu. Rev. Genet. 17,315 344 3 Rubin,G. M., Kidwell,M. G. and Bingham,P. M. (1982) The molecular basis of P M hybrid dysgenesis:the nature of induced mutations. Cell 29,987-994 1 Engels, W. R. (1979} Hybrid dysgenesis in Drosophila melanogaster, rules of inheritance of female sterility. Genet. Res 33,219 223 5 Bingham,P. M., Kidwell,M. G. and Rubin, (,. M. (1982)The molecular basis of P-M hybrid dysgenesis: the role of the P element, a P strain specific transposon family. Cell29, 995 10(14 6 Engels,W. R. (1984) A trans-acting product needed for P factor transpositionin Drosophila. Science 226, 1194-1196 7 O'Hare,K. and Rubin,G. M. (1983) Structures of P transposable elements of Drosophila melam~gaster and their sites of insertion and excision.Cell 34, 25 35 8 Karess,R. E. and Rubin, G. M. (1984) Analysis of P transposable element functionsin Drosophila. Cell 38, 135 146 9 Kidwell,M. (;. and Novy.J. B. (19791 Hybrid dysgenesis in Drosophila melanogaster: sterility resulting from gonadal dysgenesis in the P M system. Genetics 92, 1127 1140 I0 Spradling, A. C. and Rubin, G. M. (19821 Transposition of cLonedP elements intoDrosophila germ linechromosomes.Science 218,341 347 1l Rubin. G. M. and Spradling, A. C. (1982) Genetic transformation with transposable element vectors. Sci~u'e 218. 348-353 12 Rubin, G. M. and Spradling, A. C. (1983) Vectors for P element mediated gene transfer in Drosophi& Nuek'ic Acids Res. lI, 6341-6351 13 Searles,l.. l.., et al. (1983) Molecular cloning of sequences froma Drosophila RNA polymeraselocusby P elementtranspomn tagging. Cell 31,585-.592 14 Engels, W. R. and Preston, C. R. (1984l Formation of chrom(~omalrearrangements by P factors in Drosophi& Genetics 107,657 678 15 Daniels,S. B. etal. (1985} Dysgenesis-inducedinstabilityof rosy locus transformationin Drosophila melanogaster analysis of excision events and the selective recover3.'of control element deletions. Genetics 109, 95 117 16 Engels, W. R. (19811 Extrachromosomal control of nmtability in Drosophila melano~aster. Pn~'. Natl Acad. Scz. USA 76, 4011 4015 17 Ronserray, S., Anxolabehere, D. and Periquer, G. (19841 tlybrid dysgenesis in Drosophila melanogaster - influence of temperature on cytotype dcterminatinnin the P M system. MoL Gen. (;c~wt. 196, 17-23 18 Simmons,M. J. and Bucholz, L. M. Transposase titration in Drosophila melanogaster a modelof cytotype in the P M system of hybrid dysgenesis.Proc. Natl Aead .c.,ci. USA (in press) 19 Kidwe]!, M. G. Hybrid dysgenesis in Drosophila melanogaster: further studies on the nature and inheritance of P element regulation. Gem'tk~ (in press} 20 Engels,W. R. (1981) Hybrid dysgenesis in Drosophila and the stochastic loss hypothesis. Cold Spring Harbor S:vmp. Quant. Biol. 45, 561 565 21 Flavell, A. J. and Ish-Horowicz, D. (1981} Extrachromosomalcircular copiesof the eukaryotic transposable element copia in cultured Drosophila cells. Nature 292,591-595

K O'ttare is at the Department of BioctwmistO', Imperial College of Science and Technology, London S W 7 2AZ, IrK.