Biochimie 83 (2001) 1009−1022 © 2001 Société française de biochimie et biologie moléculaire / Éditions scientifiques et médicales Elsevier SAS. All rights reserved. S0300908401013499/REV

Developmentally programmed excision of internal DNA sequences in Paramecium aurelia Ariane Gratias, Mireille Bétermier* Laboratoire de Génétique Moléculaire, CNRS UMR 8541, École Normale Supérieure, 46, rue d’Ulm, 75005 Paris, France (Received 1 October 2001; accepted 6 November 2001) Abstract — The development of a new somatic nucleus (macronucleus) during sexual reproduction of the ciliate Paramecium aurelia involves reproducible chromosomal rearrangements that affect the entire germline genome. Macronuclear development can be induced experimentally, which makes P. aurelia an attractive model for the study of the mechanism and the regulation of DNA rearrangements. Two major types of rearrangements have been identified: the fragmentation of the germline chromosomes, followed by the formation of the new macronuclear chromosome ends in association with imprecise DNA elimination, and the precise excision of internal eliminated sequences (IESs). All IESs identified so far are short, A/T rich and non-coding elements. They are flanked by a direct repeat of a 5’-TA-3’ dinucleotide, a single copy of which remains at the macronuclear junction after excision. The number of these single-copy sequences has been estimated to be around 60 000 per haploid genome. This review focuses on the current knowledge about the genetic and epigenetic determinants of IES elimination in P. aurelia, the analysis of excision products, and the tightly regulated timing of excision throughout macronuclear development. Several models for the molecular mechanism of IES excision will be discussed in relation to those proposed for DNA elimination in other ciliates. © 2001 Société française de biochimie et biologie moléculaire / Éditions scientifiques et médicales Elsevier SAS. All rights reserved. site-specific recombination / DNA deletion / circularisation / transposition / ciliates

1. Introduction 1.1. Nuclear dimorphism in ciliates: separation of germline and somatic functions in a single-cell organism A common characteristic shared by all unicellular eukaryotes belonging to the monophyletic group of ciliates is the presence, within the same cytoplasm, of two types of nuclei, which play distinct roles throughout the cell life cycle (figure 1; see [1]). The diploid micronucleus divides mitotically but remains transcriptionally silent during vegetative growth. It can be viewed as the germline nucleus since it undergoes meiosis during sexual reproduction, and provides the gametic nuclei which contribute to the formation of the zygotic nucleus (figure 1, stages I–III). The macronucleus is highly polyploid, although various ploidy levels have been reported in different ciliates (45n in Tetrahymena thermophila, around 1000n in Paramecium aurelia or Euplotes crassus). It divides amitotically and is actively transcribed during vegetative *Correspondence and reprints. E-mail addresses: [email protected] (M. Bétermier), [email protected] (A. Gratias). Abbreviations: IES, internal eliminated sequence; bp, basepair; Mbp, megabase pairs; kbp, kilobase pairs; nt, nucleotides; Pddp, programmed DNA degradation protein.

growth, but is destroyed at each sexual cycle. The macronucleus can therefore be considered as the ciliate somatic nucleus, since it governs the cell phenotype but does not transmit its genome to sexual progeny. The precise number of vegetative macro- and micronuclei is variable among ciliates: P. aurelia carries one macronucleus and two micronuclei, while T. thermophila and E. crassus harbour one macronucleus and a single micronucleus. However, the general outline of the sexual processes is largely similar in all ciliates and each new sexual generation is faced with the problem of deriving a new macronucleus from a mitotic product of the zygotic nucleus. 1.2. Macronuclear development in P. aurelia Two modes of sexual reproduction have been identified in P. aurelia and can easily be induced experimentally. Mixing reactive cells of complementary mating types leads to conjugation, during which karyogamy takes place after a reciprocal exchange of gametic nuclei between two sexual partners. During the self-fertilisation process called autogamy, which can be obtained following extensive starvation of cells belonging to a single mating type, the two gametic nuclei from a single cell fuse to give the zygotic nucleus (see figure 1, stage III for details). In each case, the diploid zygotic nucleus undergoes two successive mitotic divisions (figure 1, stage IV): depending on

1010

Figure 1. The sexual cycle in P. aurelia. In the vegetative cell, the germline micronuclei are drawn as white circles and the somatic macronucleus is shown in black. The different steps of the sexual cycle are depicted as follows: Stage I: sexual processes are initiated by micronuclear meiosis, which generates eight haploid nuclei; Stage II: seven haploid nuclei quickly degenerate (grey circles), while a single one migrates to a specialised cell compartment and undergoes one mitotic division to produce two identical haploid gametic nuclei (white circles). In the meantime, the parental macronucleus becomes fragmented into approximately 30 fragments (shown in black); Stage III: two alternative fertilisation pathways, the occurrence of which depends on the presence or absence of a sexual partner, lead to the formation of the zygotic nucleus (white circle). During conjugation, reciprocal fertilisation results from the migration of one gametic nucleus from each mating cell into its sexual partner (not represented on the figure): fusion of the incoming and resident nuclei in each exconjugant produces genetically identical zygotic nuclei in both cells. During autogamy, a fully homozygous zygotic nucleus is formed by the fusion of the two identical gametic nuclei of a single cell; Stage IV: the diploid zygotic nucleus divides twice mitotically to produce four identical nuclei harbouring the same germline genome (white circles); Stage V: macronuclear development extends over two cell cycles. Va: two mitotic products of the zygotic nucleus become the new micronuclei (white circles) and the other two differentiate into the new macronuclear anlagen (shown in grey), in the presence of the fragments of the parental macronucleus (black dots). During this period, intense DNA replication and massive genome rearrangements take place within the developing anlagen. Vb: at the first cell division (or karyonidal division), the two developing macronuclei become separated into each daughter cell. Active DNA synthesis during the second cell cycle accounts for the final ploidy level reached in the mature macronucleus. For simplicity, only one daughter cell issued from each cell division is represented. Vc: macronuclear development is completed at the end of the second cell cycle and vegetative growth resumes.

Gratias et Bétermier their cellular localisation, two of the resulting nuclei become the new micronuclei while the other two differentiate into new macronuclei (figure 1, stage Va, and [2]). The whole process of macronuclear development is accompanied by intense DNA synthesis to reach a final ploidy level of 800-1000n and extends over two cell cycles following the formation of the zygotic nucleus: at the first cell division, also called karyonidal division, one developing macronucleus, or anlage, is distributed to each daughter cell (figure 1, stage Vb), and mature macronuclei are obtained at the end of the second cycle (figure 1, stage Vc). It should be emphasised that progressive degradation of the parental macronucleus starts shortly after meiosis of the germline nuclei. The parental macronucleus becomes fragmented and DNA replication rapidly stops within the resulting fragments, which persist within the cytoplasm and contribute to about 80% of total RNA synthesis throughout the whole period of formation of the new macronucleus [3]. Macronuclear fragments are eventually diluted out during the subsequent vegetative cell divisions, and can be more rapidly degraded when cells are maintained under severe starvation conditions [4]. 1.3. Developmental DNA rearrangements in P. aurelia: a genome-wide affair Not only do both types of ciliate nuclei differ in their cellular functions, their genomes also exhibit striking differences. A comparison of their respective DNA content has revealed that extensive and developmentally programmed DNA rearrangements participate in the formation of the macronuclear genome, in a highly reproducible manner from one sexual generation to the next [1, 5–7]. In the P. aurelia group of species, the germline genome is composed of 30 to 63 chromosome pairs, depending on the species or strain, and its haploid DNA content has been estimated to be around 100–200 Mbp, which would give an average chromosome size of 1–7 Mbp [1, 8]. In contrast, the acentromeric macronuclear ‘chromosomes’ are shorter molecules of 300–800 kb in length [9]. Thus, chromosomal fragmentation within reproducible regions, followed by de novo addition of telomeric repeats, is involved in the formation of the somatic genome (figure 2). Alternative fragmentation regions separated by 2–20 kbp can be used, and for each of those, the exact point of telomere addition varies within a 0.2–2 kbp range. Chromosomal fragmentation is associated with the imprecise loss of germline repetitive sequences, but, in contrast to the situation observed in T. thermophila or E. crassus, it does not appear to be determined by any specific consensus nucleotide sequence [10]. Therefore, the molecular mechanism of chromosome fragmentation in P. primaurelia remains largely unknown.

IES excision in Paramecium

1011 2. General features of IESs in P. aurelia 2.1. Sequence analysis

Figure 2. DNA rearrangements during the formation of the somatic genome in P. aurelia. The germline regions imprecisely eliminated in association with chromosomal fragmentation are shown as a grey box on the representation of the micronuclear genome (top), and precisely excised IESs are drawn as black boxes between two TA dinucleotides. The white arrow represents an open reading frame. Only a few corresponding somatic chromosomes are shown at the bottom of the figure. Telomeric repeats are drawn as grey boxes at macronuclear chromosome ends, and the 0.2–2 kbp heterogeneous regions that define their point of addition are hatched. The precise macronuclear junctions formed after IES excision are represented by TA dinucleotides.

The second type of DNA rearrangements involved in macronuclear development is the precise deletion of interstitial DNA segments specifically found in the germline genome (figure 2). In P. aurelia, these internal eliminated sequences (IESs) can be found in non-coding regions, including introns, but most of them interrupt open reading frames: therefore, IES elimination must be efficient and precise at the nucleotide level to allow the reconstitution of an active somatic genome. An extrapolation of the available data has led to an estimated number of 50 000–60 000 IESs per haploid genome (i.e., one IES every 1–2 kbp), each element being present as a single copy [11]. Thus, IES elimination in P. aurelia is not restricted to a few specific loci, but is a genome-wide phenomenon. Kinetic analyses have allowed the determination of the relative chronology of both types of DNA rearrangements in several ciliates, and have pointed to the diversity of the developmental programs involved in different organisms. In T. thermophila, all DNA rearrangements take place within the same time window [12], while precise IES elimination is completed prior to chromosome fragmentation in E. crassus [13]. This type of study has long been delayed in Paramecium, because of experimental limitations in obtaining large amounts of synchronous cells undergoing macronuclear development, but a link between IES deletion and chromosome fragmentation has been suggested in P. primaurelia [14]. Comparison of the timing of both reactions during macronuclear development should provide a better understanding of the relationships that may exist between the molecular mechanisms involved in the two types of DNA rearrangements.

The nucleotide sequence of 78 IESs of P. primaurelia and P. tetraurelia was determined by different laboratories ([11, 14–26] and S. Duharcourt, O. Garnier, A. Le Mouël and K.Y. Ling, personal communications). A striking feature of P. aurelia IESs is their extremely high A/T content (80% compared to 70% in their flanking macronuclear-destined DNA regions), similar to that of non coding sequences. All IESs are flanked by an absolutely conserved 5’-TA-3’ dinucleotide present as a direct repeat on each side, a single copy of which is retained on the chromosome after their precise deletion (figure 3A). P. aurelia IESs are short, ranging from 26 to 882 bp, but are not randomly distributed within this size range (figure 3B): 76% are shorter than 100 bp, and nearly one third are 26–30 bp long, which, interestingly, is within the size range of Paramecium introns (18–35 nt: see [27, 28]). These features relate P. aurelia IESs to the family of short ‘TA’ IESs also found in E. crassus [29]. A statistical analysis of the nucleotide sequence of 20 IESs of P. aurelia has allowed to propose a degenerate consensus sequence present as an inverted repeat at both ends [18]. This loosely conserved 8-bp sequence (5’ TAYAGYNR 3’) includes the flanking TA and is very similar to the ends of Tc1-related transposons (5’ TACAGTKS 3’), which duplicate a target TA dinucleotide upon insertion (figure 3A and [30]). Extension of this analysis to the updated database of 78 IESs confirms the general consensus and reveals an intriguing bias in the nature of the pyrimidine present at the third position: most IESs of 26-30 bp carry a T residue, while a C residue is more statistically significant for longer IESs (figure 3C and legend). Although the biological significance of this sequence bias is unclear, it could reflect differences in the evolution of the two classes of IESs or in the molecular mechanism of their elimination. The similarity between IES ends and the terminal inverted repeats of transposable elements was extended to the ‘TA’ IESs of E. crassus, in which the Tec elements, transposon-like elements belonging to the Tc1/mariner family, are massively eliminated from the developing macronuclear genome [6]. This led to the hypothesis that ‘TA’ IESs in ciliates may have evolved from ancestral transposons by losing their coding capacity while being kept under selective pressure for sequence features allowing their precise elimination during macronuclear development. 2.2. Sequence determinants required for deletion Genetic evidence points to a functional role of Paramecium IES ends in the elimination process. Mutations of the TA dinucleotide or other positions in the consensus

1012

Gratias et Bétermier

Figure 3. Sequence analysis of P. aurelia IESs. A. IES structure and general consensus for the ends. The IES is drawn as a black box flanked by two TA repeats and white triangles indicate the additional six nucleotides that define the consensus for the terminal inverted repeats. The consensus for P. aurelia IES ends is compared to the terminal inverted repeats of Tc1/mariner elements and to the consensus for E. crassus short IESs and Tec transposon-like elements. Y = C or T, R = A or G, K = G or T, S = C or G (adapted from [29]). B. Size repartition of IESs in P. aurelia. The histogram is based on the available sequences of 78 IESs from the micronuclear versions of the genes encoding the surface antigens A51, B51, α-51D, ε-51D and G51, and of the ICL1d, PAK1, PAK11 and pwB genes of P. tetraurelia strain 51, the A29 and sm19 genes of P. tetraurelia strain d4.2, the 156G and 156φG loci and one chromosome fragmentation region of P. primaurelia strain 156 (see section 2.1 for references). When alternative ends were reported for a given IES, the size of every alternative element was recorded. For the two 370-bp IESs of the A51 surface antigen gene reported to carry internal shorter IESs, the deleted forms of 341 (for IES 51A6649) and 342 bp (for 51A2591) that result from internal IES excision were considered as alternative IESs. One 882-bp IES was omitted in this representation. The size of IESs under epigenetic control (77, 222, 229 bp and two IESs of 370 bp) is indicated by asterisks. C. Variation of the consensus according to IES size. For the two groups of IESs (26–30 bp and > 30 bp), the frequency of each nucleotide for the first five positions is displayed (starting from the conserved TA). The two variant consensus sequences were determined using an adaptation of the ConsTrans computer program (S. Graziani, personal communication), by normalising these values relatively to the expected frequencies calculated from the overall nucleotide composition of the 78 IESs of our database (41% for A, 7% for C, 12% for G and 40% for T: these frequencies do not vary significantly if the calculation is restricted to the 25 IESs shorter than 30 bp).

result in the constitutive maintenance of the mutant IES in the macronuclear genome [17, 20, 21, 23]. Moreover, a comparison of excision patterns of two allelic versions of the same IESs has indicated that, when one TA boundary is mutated, a closely located TA dinucleotide can be recruited for IES elimination. This suggests that the

boundaries of a deleted DNA segment can also result from an adaptative convergent evolution of their sequence, leading to a better match to the consensus for IES ends [15]. This is further supported by the observation that, in rare cases, alternative excision boundaries are used in the same macronuclear anlage for a given IES [15, 17].

IES excision in Paramecium

1013

Although the 8-bp consensus sequence is involved in specifying the ends of an IES, it is probably not the sole determinant for elimination since it can be found in many non-IES sequences of the Paramecium genome [18]. The overall A/T-richness of IESs might favour the formation of internal secondary DNA structures involved in elimination. In addition, as was demonstrated for T. thermophila IESs (see [7]), important information may be contained within the flanking DNA of P. aurelia IESs: indeed, it has been reported that the removal of DNA sequences flanking a particular 28-bp IES inhibits its developmental elimination [31]. 2.3. Homology-dependent epigenetic control of IES elimination The existence of an epigenetic regulation of IES elimination was first suggested by the study of the mtFE mutation, inherited in a classical Mendelian manner, which was shown to affect the elimination of a 222-bp IES during macronuclear development in P. tetraurelia [22]. Surprisingly, when the wild-type mtF+ allele is reintroduced by conjugation into the micronucleus of the mutant cell line, elimination of the IES from the macronuclear genome is not restored in its sexual progeny. Instead, a new cell line is obtained, which is homozygous for the mtF+ allele but still maintains the 222-bp IES in its macronucleus, even after multiple successive rounds of autogamy. In this cell line, the alternative macronuclear DNA rearrangement pattern, initially observed in the mtFE mutant, is transmitted from one sexual generation to the next, even though the macronuclear genome of each new sexual generation is derived from a fully wild-type germline genome. The demonstration that this epigenetic control of IES elimination is exerted by the parental macronucleus was obtained by transforming the vegetative macronucleus of a wild-type cell with a plasmid carrying the cloned IES sequence (figure 4, see [32]). These experiments clearly showed that the presence of the 222-bp IES in the parental macronucleus inhibits the elimination of the same IES from the genome of the developing macronucleus of the next sexual generation. Inhibition depends on the copy number of the plasmid injected in the parental macronucleus, and also on sequence homology, since no other IES was found to be affected. In addition, efficiency of inhibition increases with the length of the flanking sequences present on the injected construct. Thus, in each cell line, although the genome of the parental macronucleus is not transmitted to the sexual progeny, it can influence the way the new macronuclear genome is formed from the inherited germline genome. In Paramecium, by analogy to sexual reproduction in other eukaryotic systems, this type of ‘macronuclear’ heredity is also referred to as ‘maternal’ heredity (reviewed in [10]). An extension of this study revealed that five out of the 13

Figure 4. Homology-dependent epigenetic control of IES excision by the parental macronucleus (adapted from [10]). A. Wild-type cell line. B. Cell line injected with a plasmid carrying a cloned copy of an IES under epigenetic control. After injection into the macronucleus, the plasmid DNA is linearised by the cell and forms multimers to the ends of which terminal telomeric repeats are added. The injected plasmid is maintained, throughout vegetative growth, as an autonomous macronuclear minichromosome. Like the rest of the macronuclear genome, it is lost at each sexual cycle. A simplified view of the germline (mic) and somatic (MAC) genomes is displayed. The chromosome ends are hatched, and the telomeres added to the ends of the linearised injected plasmid are drawn as open boxes. IESs are represented by black boxes.

IESs tested are submitted to the same kind of maternal effect [11]. No obvious size difference distinguishes the regulated IESs from those which are not (see figure 3B). Interestingly, however, two short 28- and 29-bp IESs do escape the effect, even though they are inserted within larger IESs showing inhibition of their excision. The only noticeable feature of four out of the five IESs under epigenetic control is that the nucleotide sequence of their ends (5’ TATT 3’) differs at the fourth position from the general consensus (5’ TAYA… 3’). Homology-dependent maternal inhibition of IES elimination was also described in T. thermophila [33], which suggests that a general epigenetic regulatory mechanism

1014 of DNA rearrangements may exist in ciliates [10]. Two models have been proposed to explain this phenomenon. In the first one, a cytoplasmic factor could be titrated out by the IES copies present in the parental macronucleus. However, to account for the sequence specificity, one has to imagine that one specific factor exists for each IES showing inhibition, which, given the probable high number of these sequences in the germline genome, makes it unlikely to be a protein. In the second model, sequences originating from the parental macronucleus could pair with the homologous germline sequence in the developing nucleus. This pairing would interfere with the elimination process either by directly blocking access to a putative enzymatic machinery (steric inhibition) or by inducing an epigenetic modification of the rearranged sequence, such as DNA methylation for instance, that would prevent it from being recognised for elimination. It has been proposed that an IES-specific RNA, transcribed from the parental macronucleus and imported into the anlage, might be mediating this trans-nuclear sequence specificity. However, the presence of specific transcription promoters, either within the regulated IESs or in their flanking sequences, has not been directly assayed, and the existence and transport of this putative RNA molecule still have to be demonstrated. 3. Precise elimination of IESs : a site-specific excision reaction 3.1. Models for IES elimination in P. aurelia Formally, the elimination of ‘TA’ IESs can be viewed as the precise deletion of a DNA sequence located between two short direct TA repeats. Three models can be proposed for the molecular mechanism of this particular type of DNA rearrangements. The first one relies on DNA polymerase slippage during replication [34]: for Paramecium IESs, this would involve polymerase pausing at the first TA, or immediately downstream of it, followed by reannealing of the nascent DNA strand to the second TA repeat. After several rounds of replication, the IES would be under-amplified or completely absent from the new somatic chromosomes. In a second model, an internal double strand break would be introduced within the IES, perhaps at the level of an internal secondary structure. Double strand break repair through a pathway related to single strand annealing (see [35] for a review) would produce 3’-overhangs able to anneal at the level of the TA repeats. Trimming of the unpaired 3’ extensions followed by gap-filling and ligation would produce the macronuclear chromosome sequence. The last type of mechanism is related to the recombination reactions which participate in DNA site-specific excision/integration or transposition [36]. In this case, endonucleolytic cleavages would be introduced at IES ends, in the region of the TA

Gratias et Bétermier repeats. This would release the excised IES while the flanking macronuclear DNA ends would be precisely rejoined by direct strand transfer or DNA repair. 3.2. Extrachromosomal forms of excised IESs Of the three models presented above, only the third one would allow the liberation of extrachromosomal forms of the excised sequence. The analysis of IES excision products in P. tetraurelia has provided strong experimental support for this model [37]. Southern blots of total genomic DNA from large-scale cultures of autogamous cells revealed the accumulation, during macronuclear development, of extrachromosomal forms for all IESs tested larger than 200 bp (figure 5A). Further analysis of the electrophoretic behaviour of the excised molecules obtained for two IESs of 222 bp and 370 bp, and their treatment by restriction enzymes and exonuclease III, indicated that the major excision product in each case is a covalently closed, double stranded DNA circle. Interestingly, the exact number of extrachromosomal bands observed on the electrophoresis gel increases with IES size and their relative abundance varies during macronuclear development (figure 5A). These bands have tentatively been assigned to topoisomers of the excised IES circles, namely nicked or relaxed forms for the slowest migrating species, and differentially supercoiled circles for the faster migrating ones. However, the exact nature of these forms and the biological significance of their different timing of appearance still have to be elucidated. Sequence analysis of the circles obtained for three different IESs showed that their ends are precisely joined on these molecules and separated by one copy of the flanking TA repeat ([37] and Gratias, unpublished). Thus, the same precision is observed in the formation of the circular and chromosomal junctions following IES excision (figure 5B). This is in favour of a model where excision involves DNA cleavage near the TA dinucleotide at each IES end, and results in the release of an excised form which, at least for IESs larger than 200 bp, can be mainly detected as a circular molecule. The homogeneity of the macronuclear junction and the accurate formation of both chromosomal and circular junctions characterise IES excision in P. aurelia. In E. crassus, excision of ‘TA’ IESs also results in the formation of a precise TA macronuclear junction and produces abundant circular molecules [38]. However, the circular junction is composed of two copies of the TA repeat, separated by a 10-bp partial heteroduplex originating from the flanking DNA (figure 5B, see [39, 40]). Moreover, although sequence similarities have been observed between the ends of ‘TA’ IESs and Tc1/mariner transposons, the products of IES developmental excision differ significantly from those produced during the ‘cut-and-paste’ transposition of Tc1 elements (figure 5B): the latter are excised as linear molecules and, when the donor sequence is repaired by

IES excision in Paramecium

1015

Figure 5. Detection of extrachromosomal products of IES excision in P. tetraurelia. A. Southern blot analysis of genomic DNA from vegetative (V) and autogamous cells at different stages of macronuclear development, till the karyonidal division. The blot was successively hybridised with four radioactively labelled probes corresponding to IESs 51A4578, 51A2591, 51A6649 and 51G4404, as described in [37]. For each IES, an arrowhead indicates the minor product migrating at the size expected for a putative linear excised form (respectively 882 bp, 370 bp, 341 bp and 222 bp). B. Schematic representation of the excision products obtained for E. crassus and P. aurelia IESs, and comparison with Tc1/mariner transposons. The excised sequences are drawn as a double continuous line and dotted lines represent the flanking chromosomal DNA. The structure of the linear excised forms of Paramecium IESs is unknown, as indicated by the ? at their ends. The double-strand cleavages at Tc1 ends are shown: although circular versions of excised Tc1 have been detected, these are most likely by-products of transposition and are therefore not represented.

cellular enzymes, the two flanking TA dinucleotides are generally retained at the chromosomal junction, separated by a ∼ 2-bp footprint originating from the transposon sequence [30]. More than 70% of P. aurelia IESs are shorter than 80 bp (figure 3B), which is about the minimal length compatible with the formation of double stranded DNA circles in vivo [41]. All attempts to detect circles for 77-bp and even 150-bp IESs have been unsuccessful so far (Gratias, unpublished): thus, circular molecules may not be produced at all for the shortest IESs and may be only secondary products of excision for the longest ones. In support of this hypothesis, a minor product exhibiting the electrophoretic behaviour expected for a linear molecule has been detected at early time points of macronuclear development for the four IESs examined so far (figure 5A). This suggests that, as proposed for T. thermophila IESs

[42], which do not belong to the ‘TA’ IES family, IES excision in P. aurelia may result in the early release of linear molecules. These would then be rapidly degraded, or, although this remains to be assayed directly, could be converted into circles, as demonstrated for the signal joints produced by V(D)J recombination during the assembly of immunoglobulin genes in mammals (reviewed in [43]). 4. IES excision is regulated during macronuclear development 4.1. Timing of IES excision Various aspects of macronuclear development have been studied using similar techniques for the synchroni

1016 sation of small-scale cultures of well-fed exconjugants of P. aurelia. Microscopic analysis of radioactively pulselabelled cells led to the determination of DNA synthesis rates within the developing macronucleus and to the detection of anlage-specific transcription as early as 3 to 4 hours after exconjugant separation [3], which takes place between the first and second divisions of the zygotic nucleus [44]. Semi-quantitative PCR using convergent pairs of primers was used to specifically amplify the chromosomal (unexcised) copies of several IESs [37]. Shortly after exconjugant separation, IESs start to be amplified in the developing macronuclear genome, in association with their flanking chromosomal sequences (figure 6A). Halfway through the first cell cycle, which extends over a 10- to 12-hour period, a rapid decrease in the chromosomal signal for the three IESs tested reflects their massive elimination, which is essentially completed when karyonidal division takes place. Massive IES excision is reproducibly detected within a time window corresponding to the middle of the first cell cycle, independently of the P. tetraurelia strain used or of variations in the extent of this cycle (Gratias, unpublished). However, excision could be initiated earlier: indeed, for two particular IESs of 28 bp and 29 bp, it has been possible to directly monitor the transient appearance of the chromosomal junctions, which are not present in the parental macronucleus since these sequences are located within two larger IESs of 370 bp (51A2591 and 51A6649, respectively). Excision of these short IESs is detected approximately 2 h before the decrease in the chromosomal signal observed for the larger ones (figure 6A). Analysis of the excision products of the larger IESs revealed striking differences in the amounts of full-sized circles compared to those from which the internal IESs have been deleted: the circles observed for IES 51A6649 are mostly 349-bp molecules deleted from the internal element, while most of IES 51A2591 circles are 370 bp long and retain the short IES (figure 5A). This indicates that some excision events are initiated early, at least for short IESs, with an efficiency that may vary from one element to the other. Massive IES excision takes place during a period of intense DNA synthesis within the macronuclear anlage (figures 6A, B). Such a correlation in the timing of IES elimination and DNA replication has also been reported in other ciliates [45–47], but no general scheme has emerged from these studies for a participation of DNA replication in excision. In T. thermophila, inhibition of DNA polymerase α by aphidicolin treatment just after the beginning of macronuclear development does not prevent or delay excision [48]. In contrast, addition of hydroxyurea decreases the efficiency of ‘TA’ IES excision in E. crassus [46]. Similar studies have not been carried out in P. aurelia.

Gratias et Bétermier 4.2. Detection and fate of circular products In P. aurelia, massive elimination of IESs larger than 200 bp is associated with the formation of abundant circular products that can be amplified by PCR with divergent primers [37]. These molecules transiently accumulate during macronuclear development till the end of the first cell cycle, at which time they quickly disappear (figure 6C). This suggests that circles are degraded by an active mechanism, rather than simply being diluted out during successive cell divisions. Similarly, although IES circles are hardly detected in T. thermophila, time-course studies have indicated that they do persist in this ciliate until karyonidal division [49]. During the second cell cycle following meiosis in P. aurelia, a second round of production of DNA junctions characteristic of circularised IESs has been detected, in two independent experiments and for two different IESs, while no previous increase in the chromosomal IES signal can be observed (figure 6C, and Gratias, unpublished). This may result from the excision of the few remaining chromosomal IES copies that, for some reason, would have escaped the massive round of elimination during the first cell cycle and would be deleted concomitantly with their replication. In support of this hypothesis is the observation that, in some cases, cloned IESs injected into the developing macronucleus after karyonidal division can still be excised, which indicates that the excision machinery is still present and active at this stage [31]. 5. Towards the molecular mechanism of IES excision in P. aurelia 5.1. DNA transactions leading to excision The available data suggest that excision of P. aurelia IESs involves DNA cleavage at their ends, near the flanking TA repeats, but no information has been obtained on the precise number of initial cleavage events or on the existence of a concerted cleavage at both ends. Several mechanisms have been proposed for the developmental deletion of germline sequences in other ciliates (figure 7A). In E. crassus, the unusual structure of the circular DNA junctions formed by ‘TA’ IESs has led to an excision model involving the simultaneous introduction of staggered, double-stranded DNA breaks at each end: one strand would be cleaved adjacent to the TA repeat and the other one within the flanking DNA, to generate 5’overhangs. The macronuclear junction would be formed by the pairing of the TA dinucleotides carried by the flanking DNA ends, followed by gap filling and ligation, while the circle heteroduplex junctions would result from the alignment of the non-complementary overhangs generated at IES ends [29]. Direct experimental evidence was obtained in T. thermophila for the existence of double

IES excision in Paramecium

1017

Figure 6. Time-course analysis of IES excision in well-fed exconjugants of P. tetraurelia. Semi-quantitative PCR reactions with convergent or divergent primers were performed on total lysates from cells issued from 5 synchronised mating pairs (as described in [37]), to amplify chromosomal and circular IES forms, respectively. A. Amplification of IES chromosomal forms as a function of time, following exconjugant separation until the end of the second cell cycle. Elimination of IESs 51A–712 (77 bp) and 51G4404 (222 bp) is displayed relative to the left axis, and the right axis refers to 51A6649 (370 bp). The arrow indicates the first excision events detected for the 29-bp IES within 51A6649. B. Peaks of DNA synthesis during the first cell cycle, as measured by radioactive pulse-labelling of exconjugants (from [3]). Since the extent of the first cell cycle was longer in this experiment than in A, the scale of the horizontal axis was adapted to superimpose exconjugant separation and karyonidal division in both experiments. C. Semiquantitative PCR detection of the chromosomal and circular forms of IES 51G4404 during macronuclear development, from exconjugant separation till the end of the second cell cycle. The black box on the horizontal axis represents karyonidal division.

strand breaks generating 5’-overhangs of 4 nucleotides [50, 51]. However, recent results have suggested that, in this ciliate, the initiating break occurs at a single end, and cleavage of the other end is mediated by a nucleophilic attack by the 3’OH group from the flanking macronuclear DNA (figure 7A and [42]). This first attack would give rise to a branched intermediate, from which a linear form of

the excised IES could be released after a second cleavage step. A similar transesterification mechanism was proposed for the precise excision of TBE-1 elements in Oxytricha fallax (figure 7A and [52]). In this model, however, the 3’OH groups liberated on both strands by the initial double strand break on one end would attack the phosphodiester backbone of the two DNA strands at the

1018

Gratias et Bétermier

Figure 7. Molecular models for DNA transactions involved in IES excision. A. Models proposed for E. crassus, T. thermophila and O. trifallax. IESs are drawn as thick grey lines and the excision process is described in three steps: 1 , the initial DNA cut; 2 , the nucleophilic attack involved in cleavage of the other end for T. thermophila and O. trifallax elements (such an intermediate step has not been proposed for E. crassus); and 3 the obtention of the final products. Detailed experimental data were obtained for IES excision in T. thermophila [42, 51]. The initial 4-bp staggered double-strand break is introduced at a single end, 3’ to an adenine residue in the consensus sequence 5’-ANNNNT-3’: either end can be cleaved first, and alternative cleavage sites have been demonstrated for a given end. Nucleophilic attack of the same strand at the other end by the free 3’OH group also takes place at alternative positions, which explains the heterogeneity of the macronuclear junctions in this ciliate. The excised IES is released as a linear molecule by an additional cleavage at variable positions on the other strand: the linear IES exhibits a typical structure, with a 5’-overhang at one end, corresponding to the initial cleavage, and a 3’ overhang at the other, corresponding to the transesterification step. The macronuclear junction is repaired by degradation of the unpaired overhangs and gap filling (dotted line in step 3 ). The reactive 3’OH groups involved in transesterification reactions are drawn as full circles. B. Proposed models for IES excision in Paramecium, that would generate circular (left) or linear molecules (right) as primary excision products (see text for discussion). DNA cleavage events are symbolised by scissors, and DNA strand-transfers by arrows, or by a cross within the nucleoprotein complex involving both ends (left part of B).

other end: this would directly result in the circularisation of the excised element and the formation of the chromosomal junction. Although the exact positions of the initial cleavages need to be determined for P. aurelia IESs, two excision models can tentatively be proposed (figure 7B), which mainly differ by the nature of the excised molecule produced by the reaction. Circular molecules could directly be generated through pathways involving DNA strand transfers between both ends, such as those proposed

for TBE-1 or various other transposition and site-specific recombination systems [36]: the linear molecules would correspond to excision by-products or degradation intermediates of the circles. This type of model implies that IES ends are able to be physically brought together, which can be facilitated by DNA structure, such as intrinsic bending, or the binding of an architectural protein, but seems hard to reconcile with the short size of most P. aurelia IESs and the early detection of linear excised molecules. Alternatively, DNA double strand breaks intro

IES excision in Paramecium duced simultaneously at both ends, or a single initiating break followed by a transesterification step analogous to the one proposed for T. thermophila IESs, would release a linear form of the excised sequence. For IESs larger than 200 bp, this linear molecule could subsequently be circularised. With respect to this hypothesis, the biological significance of the circles should be addressed, since these are the major excision products observed for large IESs and accumulate in significant amounts prior to their degradation. IES excision could potentially generate one double strand break every 1–2 kb in the developing macronuclear genome. Consequently, P. aurelia may have evolved a highly efficient DNA repair system, and the great precision observed for the circular junctions could reflect the fidelity of this double strand break repair pathway. It has also been proposed that circle formation could prevent the ‘reactive’ ends of an excised IES from reintegrating into macronuclear chromosomes [52]. 5.2. Protein factors involved in DNA cleavage The identity of the enzymatic machinery responsible for the initial cleavages of IES ends remains an open question in ciliates. Given the lack of coding capacity of ‘TA’ IESs of P. aurelia, it can be postulated that the reaction is carried out by functions encoded elsewhere in the genome. One mutation, mtFE, has been reported to inhibit IES excision in P. tetraurelia, in addition to causing pleiotropic effects during development [22]. However, since only a few IESs appear to be affected in this mutant, the mtFE gene probably corresponds to an accessory factor. An interesting hypothesis for a family of putative excision factors has emerged from the discovery of a ‘transposon link’ for ciliate IESs [6]. Indeed, Tc1-related transposons encode a transposase carrying a characteristic DDE catalytic domain [30] and the Tec transposon-like elements of E. crassus harbour several open reading frames, one of which encodes a putative DDE transposase [53]. Their precise excision produces the same type of circular junctions as those reported for E. crassus short ‘TA’ IESs [54]: this led to the suggestion that IESs and Tecs could be excised by the Tec transposase. However, the very low levels of Tec transcripts detected during macronuclear development do not support the sole participation of the transposase to the massive developmental excision of the Tecs and of the ‘TA’ IESs [55]. Instead, it has been proposed that these transcripts may be sufficient for low-level Tec transposition within the germline genome, although such transposition events have not been described. Transposon-like elements belonging to the Tc1/mariner family have also been discovered in the germline genome of P. aurelia [10], but further work is needed to evaluate the contribution of their putative proteins to IES excision. In other systems, in vitro assays have revealed that

1019 transposition-related DNA excision is always initiated by the liberation of a reactive 3’OH group through a single strand cleavage reaction catalysed by the DDE domain of the recombinase [56]. The various strategies used for the second strand cleavage leading to excision of the internal sequence are characterised by the formation of specific molecular intermediates [57]. Second strand cleavage can be performed by another protein, not necessarily carrying a DDE catalytic site, as illustrated for the bacterial Tn7 transposon [58]. Alternatively, nucleophilic attack of the facing strand by the free 3’OH group generates a hairpin structure, as demonstrated for some cut-and-paste transposons and for the coding ends generated by V(D)J recombination. In these first two situations, the internal sequence is primarily excised as a linear molecule. For some bacterial insertion sequences, such as IS911, however, a 3’OH group is liberated at one end only and nucleophile attack of the other end gives a typical ‘figureof-eight’ molecule, which is resolved as a double stranded circle by a host-encoded machinery: additional cleavage of the circular junction by the transposase generates a reactive linear transposition intermediate [59]. No clear experimental evidence has been obtained so far that would link P. aurelia IES excision to transpositional recombination. The initial DNA breaks could be produced by topoisomerases or site-specific recombinases, which generate transient covalent intermediates between the protein and the cleaved DNA [36, 60], or by any other type of endonucleases. Molecular analysis of intermediate products and extensive search for the putative developmental recombinase should contribute to a better understanding of the reaction. 5.3. Targeting the cleavage machinery to the excised sequences The TA repeats flanking Paramecium IESs, and their adjacent nucleotides corresponding to the consensus for IES ends, are most likely part of the site cleaved by the recombinase during the initiating step of excision. These sequences, however, are probably not sufficient to precisely target the recombinase and several speculations can be made about the pathways that could allow the specification a germline sequence to be excised. Although the macro- and micronucleus of ciliates are derived from identical mitotic products of the zygotic nucleus, significant differences exist in their chromatin structure, as shown in E. crassus and T. thermophila (see [61–65] and references therein). Macronuclear development should, therefore, be accompanied by profound chromatin changes within the anlage, such as the incorporation of histone variants or other chromatin-associated proteins, and the post-translational modification of histones which could account for transcriptional activation of the developing macronucleus [66]. Chromatin remodelling could also epigenetically label IES sequences for

1020 excision, by discriminating them from their macronucleardestined flanking sequences, or for subsequent degradation of their excised forms. This has been suggested on the basis of the unusually compact chromatin structures formed by the Tec elements of E. crassus during macronuclear development, which have been attributed to the specific incorporation of a developmental variant of core histone H3 [67]. In T. thermophila, three proteins, named Pdd1p, Pdd2p and Pdd3p, have been reported to specifically associate with eliminated germline DNA in electrondense structures, during the formation of the macronucleus [48, 68–70]. Expression of Pdd1p and Pdd2p is required for IES excision and, strikingly, Pdd1p and Pdd3p contain characteristic conserved protein motifs called chromodomains, originally found in the Drosophila heterochromatin associated protein HP1 and in the Polycomb protein [71]. This suggests a possible involvement of the Pddp proteins in heterochromatin formation, via protein-protein interactions with specific modified forms of histones [72–74] or protein-RNA interactions [75]. In P. aurelia, activation of transcription within the macronuclear anlage takes place during a time interval that could correspond to the first detectable IES excision events, a few hours prior to their massive excision (figure 6). Interestingly, in T. thermophila, specific transcription of the IESs, initiated from their flanking DNA sequences, was also detected prior to their elimination and suggested to be required for excision [76]. Read-through transcription could result in a local change in the germline chromatin structure at IES ends, as a consequence of nucleosome displacement or histone acetylation [77, 78]. As proposed for other systems such as V(D)J [79, 80] or the class switch recombination of immunoglobulin genes (reviewed in [81]), this could specifically make the ends accessible to a recombinase. Alternatively, a direct role can be proposed for IES specific transcripts: through pairing to their homologous DNA, they could specifically target the cutting machinery to IES ends. The existence and mode of action of such ‘guide’ RNA molecules need to be investigated, but they would provide an alternative explanation for the homology-dependent maternal inhibition of IES excision (section 2.3). During macronuclear development, putative aberrant transcripts produced from IES copies retained in the parental macronucleus could pair to the guide RNAs and induce their degradation through a pathway related to RNA interference [82]. The existence of a post-transcriptional, homology-dependent gene silencing system in P. aurelia has been demonstrated in vegetative cells ([83] and A. Galvani and L. Sperling, personal communication) and provides support to this hypothesis. 6. Conclusion Significant progress has been made, in the past few years, in the characterisation of the cis requirements, the

Gratias et Bétermier description of intermediate products and the determination of the timing of IES excision during macronuclear development. This information should be of great help in the understanding of the molecular mechanisms that participate in the recognition and excision of P. aurelia eliminated sequences, and in the regulation of these processes. This area of research will greatly benefit from the use of new powerful tools, such as homology-dependent gene silencing [83] and the development of a genomic program for the discovery of genes in P. aurelia [28], and should provide important contributions to the general knowledge of site-specific recombination and homology-dependent epigenetic effects.

Acknowledgments We wish to thank Sandra Duharcourt, Angélique Galvani, Olivier Garnier, Anne Le Mouël, Kit-Yi Ling and Linda Sperling for the communication of unpublished results and Stéphane Graziani for his help in adapting his ConsTrans computer program to the statistical analysis of IES ends. We are grateful to all former and present members of Eric Meyer’s lab for extremely rich and stimulating discussions, and to E. Meyer for critical reading of the manuscript. The work in the Ciliate Molecular Biology group was supported by the Association pour la Recherche sur le Cancer (grant no. 5733), the Centre National de la Recherche Scientifique (Programme Génome), the Ministère de l’Education Nationale, de la Recherche et de la Technologie (Programme de Recherche fondamentale en Microbiologie et Maladies infectieuses et parasitaires), and the Comité de Paris de la Ligue Nationale contre le Cancer (grant no. 75/01-RS/73). A. Gratias is the recipient of a doctoral fellowship from the French Ministère de la Recherche.

References [1] Prescott D.M., The DNA of ciliated protozoa, Microbiol. Rev. 58 (1994) 233–267. [2] Grandchamp S., Beisson J., Positional control of nuclear differentiation in Paramecium, Dev. Biol. 81 (1981) 336–341. [3] Berger J.D., Nuclear differentiation and nucleic acid synthesis in well-fed exconjugants of Paramecium aurelia, Chromosoma 42 (1973) 247–268. [4] Berger J.D., Selective autolysis of nuclei as a source of DNA precursors in Paramecium aurelia exconjugants, J. Protozool. 21 (1974) 145–152. [5] Coyne R.S., Chalker D.L., Yao M.C., Genome downsizing during ciliate development: nuclear division of labor through chromosome restructuring, Annu. Rev. Genet. 30 (1996) 557–578. [6] Klobutcher L.A., Herrick G., Developmental genome reorganization in ciliated protozoa: the transposon link, Prog. Nucleic Acid Res. Mol. Biol. 56 (1997) 1–62. [7] Yao M.C., Duharcourt S., Chalker D.L., Genome-wide rearrangements of DNA in ciliates, in: Craig N., Craigie R., Gellert M., Lambowitz A. (Eds.), Mobile DNA II, American Society for Microbiology, Washington D.C., 2001 in press. [8] Sonneborn T.M., Paramecium aurelia, in: King R.C. (Ed.), Handbook of genetics : Plants, plant viruses and protists, Plenum Press, New York, 1974, pp. 469–594.

IES excision in Paramecium [9] Caron F., Meyer E., Molecular basis of surface antigen variation in paramecia, Annu. Rev. Microbiol. 43 (1989) 23–42. [10] Meyer E., Garnier O., Non-Mendelian inheritance and homologydependent effects in ciliates, Adv. Genet. (2001) in press. [11] Duharcourt S., Keller A.M., Meyer E., Homology-dependent maternal inhibition of developmental excision of internal eliminated sequences in Paramecium tetraurelia, Mol. Cell. Biol. 18 (1998) 7075–7085. [12] Yao M.C., Site-specific chromosome breakage and DNA deletion in ciliates, in: Berg D.E., Howe M.M. (Eds.), Mobile DNA, American Society for Microbiology, Washington, D.C., 1989, pp. 715–734. [13] Tausta S.L., Klobutcher L.A., Internal eliminated sequences are removed prior to chromosome fragmentation during development in Euplotes crassus, Nucleic Acids Res. 18 (1990) 845–853. [14] Amar L., Chromosome end formation and internal sequence elimination as alternative genomic rearrangements in the ciliate Paramecium, J. Mol. Biol. 236 (1994) 421–426. [15] Dubrana K., Le Mouël A., Amar L., Deletion endpoint allelespecificity in the developmentally regulated elimination of an internal sequence (IES) in Paramecium, Nucleic Acids Res. 25 (1997) 2448–2454. [16] Dubrana K., Amar L., Programmed DNA under-amplification in Paramecium primaurelia, Chromosoma 109 (2000) 460–466. [17] Haynes W.J., Ling K.Y., Preston R.R., Saimi Y., Kung C., The cloning and molecular analysis of pawn-B in Paramecium tetraurelia, Genetics 155 (2000) 1105–1117. [18] Klobutcher L.A., Herrick G., Consensus inverted terminal repeat sequence of Paramecium IESs: resemblance to termini of Tc1related and Euplotes Tec transposons, Nucleic Acids Res. 23 (1995) 2006–2013. [19] Ling K.Y., Vaillant B., Haynes W.J., Saimi Y., Kung C., A comparison of internal eliminated sequences in the genes that encode two K+-channel isoforms in Paramecium tetraurelia, J. Euk. Microbiol. 45 (1998) 459–465. [20] Mayer K.M., Mikami K., Forney J.D., A mutation in Paramecium tetraurelia reveals function and structural features of developmentally excised DNA elements, Genetics 148 (1998) 139–149. [21] Mayer K.M., Forney J.D., A mutation in the flanking 5’-TA-3’ dinucleotide prevents excision of an internal eliminated sequence from the Paramecium tetraurelia genome, Genetics 151 (1999) 597–604. [22] Meyer E., Keller A.M., A mendelian mutation affecting matingtype determination also affects developmental genomic rearrangements in Paramecium tetraurelia, Genetics 143 (1996) 191–202. [23] Ruiz F., Krzywicka A., Klotz C., Keller A., Cohen J., Koll F., Balavoine G., Beisson J., The SM19 gene, required for duplication of basal bodies in Paramecium, encodes a novel tubulin, etatubulin, Curr. Biol. 10 (2000) 1451–1454. [24] Scott J., Leeck C., Forney J., Analysis of the micronuclear B type surface protein gene in Paramecium tetraurelia, Nucleic Acids Res. 22 (1994) 5079–5084. [25] Steele C.J., Barkocy-Gallagher G.A., Preer L.B., Preer J.R., Developmentally excised sequences in micronuclear DNA of Paramecium, Proc. Natl. Acad. Sci. USA 91 (1994) 2255–2259. [26] Vayssié L., Sperling L., Madeddu L., Characterization of multigene families in the micronuclear genome of Paramecium tetraurelia reveals a germline specific sequence in an intron of a centrin gene, Nucleic Acids Res. 25 (1997) 1036–1041. [27] Dupuis P., The beta-tubulin genes of Paramecium are interrupted by two 27 bp introns, EMBO J. 11 (1992) 3713–3719. [28] Dessen P., Zagulski M., Gromadka R., Plattner H., Kissmehl R., Meyer E., Bétermier M., Schultz J.E., Linder J.U., Pearlman R.E., Kung C., Forney J., Satir B., van Houten J.L., Keller A.M., Froissard M., Sperling L., Cohen J., Paramecium genome survey: a pilot project, Trends Genet. 17 (2001) 306–308. [29] Jacobs M.E., Klobutcher L.A., The long and the short of developmental DNA deletion in Euplotes crassus, J. Euk. Microbiol. 43 (1996) 442–452.

1021 [30] Plasterk R.H., Izsvak Z., Ivics Z., Resident aliens: the Tc1/mariner superfamily of transposable elements, Trends Genet. 15 (1999) 326–332. [31] Ku M., Mayer K., Forney J.D., Developmentally regulated excision of a 28-base-pair sequence from the Paramecium genome requires flanking DNA, Mol. Cell. Biol. 20 (2000) 8390–8396. [32] Duharcourt S., Butler A., Meyer E., Epigenetic self-regulation of developmental excision of an internal eliminated sequence in Paramecium tetraurelia, Genes Dev. 9 (1995) 2065–2077. [33] Chalker D.L., Yao M.C., Non-Mendelian, heritable blocks to DNA rearrangement are induced by loading the somatic nucleus of Tetrahymena thermophila with germ line-limited DNA, Mol. Cell. Biol. 16 (1996) 3658–3667. [34] Viguera E., Canceill D., Ehrlich S.D., Replication slippage involves DNA polymerase pausing and dissociation, EMBO J. 20 (2001) 2587–2595. [35] Karran P., DNA double strand break repair in mammalian cells, Curr. Opin. Genet. Dev. 10 (2000) 144–150. [36] Hallet B., Sherratt D.J., Transposition and site-specific recombination: adapting DNA cut-and-paste mechanisms to a variety of genetic rearrangements, FEMS Microbiol. Rev. 21 (1997) 157–178. [37] Bétermier M., Duharcourt S., Seitz H., Meyer E., Timing of developmentally programmed excision and circularization of Paramecium internal eliminated sequences, Mol. Cell. Biol. 20 (2000) 1553–1561. [38] Tausta S.L., Klobutcher L.A., Detection of circular forms of eliminated DNA during macronuclear development in E. crassus, Cell 59 (1989) 1019–1026. [39] Tausta S.L., Turner L.R., Buckley L.K., Klobutcher L.A., High fidelity developmental excision of Tec1 transposons and internal eliminated sequences in Euplotes crassus, Nucleic Acids Res. 19 (1991) 3229–3236. [40] Klobutcher L.A., Turner L.R., LaPlante J., Circular forms of developmentally excised DNA in Euplotes crassus have a heteroduplex junction, Genes Dev. 7 (1993) 84–94. [41] Ringrose L., Chabanis S., Angrand P.O., Woodroofe C., Stewart A.F., Quantitative comparison of DNA looping in vitro and in vivo: chromatin increases effective DNA flexibility at short distances, EMBO J. 18 (1999) 6630–6641. [42] Saveliev S.V., Cox M.M., Product analysis illuminates the final steps of IES deletion in Tetrahymena thermophila, EMBO J. 20 (2001) 3251–3261. [43] Grawunder U., Harfst E., How to make ends meet in V(D)J recombination, Curr. Opin. Immunol. 13 (2001) 186–194. [44] Diller W.F., Nuclear reorganization processes in Paramecium aurelia, with description of autogamy and hemixis, J. Morphol. 59 (1936) 11–67. [45] Austerberry C.F., Allis C.D., Yao M.C., Specific DNA rearrangements in synchronously developing nuclei of Tetrahymena, Proc. Natl. Acad. Sci. USA 81 (1984) 7383–7387. [46] Frels J.S., Jahn C.L., DNA rearrangements in Euplotes crassus coincide with discrete periods of DNA replication during the polytene chromosome stage of macronuclear development, Mol. Cell. Biol. 15 (1995) 6488–6495. [47] Frels J.S., Tebeau C.M., Doktor S.Z., Jahn C.L., Differential replication and DNA elimination in the polytene chromosomes of Euplotes crassus, Mol. Biol. Cell 7 (1996) 755–768. [48] Nikiforov M.A., Smothers J.F., Gorovsky M.A., Allis C.D., Excision of micronuclear-specific DNA requires parental expression of Pdd2p and occurs independently from DNA replication in Tetrahymena thermophila, Genes Dev. 13 (1999) 2852–2862. [49] Saveliev S.V., Cox M.M., The fate of deleted DNA produced during programmed genomic deletion events in Tetrahymena thermophila, Nucleic Acids Res. 22 (1994) 5695–5701. [50] Saveliev S.V., Cox M.M., Transient DNA breaks associated with programmed genomic deletion events in conjugating cells of Tetrahymena thermophila, Genes Dev. 9 (1995) 248–255.

1022 [51] Saveliev S.V., Cox M.M., Developmentally programmed DNA deletion in Tetrahymena thermophila by a transposition-like reaction pathway, EMBO J. 15 (1996) 2858–2869. [52] Williams K., Doak T.G., Herrick G., Developmental precise excision of Oxytricha trifallax telomere-bearing elements and formation of circles closed by a copy of the flanking target duplication, EMBO J. 12 (1993) 4593–4601. [53] Doak T.G., Doerder F.P., Jahn C.L., Herrick G., A proposed superfamily of transposase genes: transposon-like elements in ciliated protozoa and a common ’D35E’ motif, Proc. Natl. Acad. Sci. USA 91 (1994) 942–946. [54] Jaraczewski J.W., Jahn C.L., Elimination of Tec elements involves a novel excision process, Genes Dev. 7 (1993) 95–105. [55] Jaraczewski J.W., Frels J.S., Jahn C.L., Developmentally regulated, low abundance Tec element transcripts in Euplotes crassusimplications for DNA elimination and transposition, Nucleic Acids Res. 22 (1994) 4535–4542. [56] van Gent D.C., Mizuuchi K., Gellert M., Similarities between initiation of V(D)J recombination and retroviral integration, Science 271 (1996) 1592–1594. [57] Turlan C., Chandler M., Playing second fiddle: second-strand processing and liberation of transposable elements from donor DNA, Trends Microbiol. 8 (2000) 268–274. [58] Hickman A.B., Li Y., Mathew S.V., May E.W., Craig N.L., Dyda F., Unexpected structural diversity in DNA recombination: the restriction endonuclease connection, Mol. Cell 5 (2000) 1025–1034. [59] Rousseau P., Normand C., Loot C., Turlan C., Alazard R., Duval-Valentin G., Chandler M., Transposition of IS911, in: Craig N., Craigie R., Gellert M., Lambowitz A. (Eds.), Mobile DNA II, American Society for Microbiology, Washington D.C., 2001 in press. [60] Champoux J.J., DNA Topoisomerases: Structure, function, and mechanism, Annu. Rev. Biochem. 70 (2001) 369–413. [61] Huang H., Wiley E.A., Lending C.R., Allis C.D., An HP1-like protein is missing from transcriptionally silent micronuclei of Tetrahymena, Proc. Natl. Acad. Sci. USA 95 (1998) 13624–13629. [62] Jahn C.L., Ling Z., Tebeau C.M., Klobutcher L.A., An unusual histone H3 specific for early macronuclear development in Euplotes crassus, Proc. Natl. Acad. Sci. USA 94 (1997) 1332–1337. [63] Stargell L.A., Bowen J., Dadd C.A., Dedon P.C., Davis M., Cook R.G., Allis C.D., Gorovsky M.A., Temporal and spatial association of histone H2A variant hv1 with transcriptionally competent chromatin during nuclear development in Tetrahymena thermophila, Genes Dev. 7 (1993) 2641–2651. [64] Strahl B.D., Ohba R., Cook R.G., Allis C.D., Methylation of histone H3 at lysine 4 is highly conserved and correlates with transcriptionally active nuclei in Tetrahymena, Proc. Natl. Acad. Sci. USA 96 (1999) 14967–14972. [65] Sweet M.T., Jones K., Allis C.D., Phosphorylation of linker histone is associated with transcriptional activation in a normally silent nucleus, J. Cell. Biol. 135 (1996) 1219–1228. [66] Jenuwein T., Allis C.D., Translating the histone code, Science 293 (2001) 1074–1080.

Gratias et Bétermier [67] Jahn C.L., Differentiation of chromatin during DNA elimination in Euplotes crassus, Mol. Biol. Cell 10 (1999) 4217–4230. [68] Coyne R.S., Nikiforov M.A., Smothers J.F., Allis C.D., Yao M.C., Parental expression of the chromodomain protein Pdd1p is required for completion of programmed DNA elimination and nuclear differentiation, Mol. Cell. 4 (1999) 865–872. [69] Madireddi M.T., Coyne R.S., Smothers J.F., Mickey K.M., Yao M.C., Allis C.D., Pdd1p, a novel chromodomain-containing protein, links heterochromatin assembly and DNA elimination in Tetrahymena, Cell 87 (1996) 75–84. [70] Nikiforov M.A., Gorovsky M.A., Allis C.D., A novel chromodomain protein, Pdd3p, associates with internal eliminated sequences during macronuclear development in Tetrahymena thermophila, Mol. Cell. Biol. 20 (2000) 4128–4134. [71] Cavalli G., Paro R., Chromo-domain proteins: linking chromatin structure to epigenetic regulation, Curr. Opin. Cell Biol. 10 (1998) 354–360. [72] Bannister A.J., Zegerman P., Partridge J.F., Miska E.A., Thomas J.O., Allshire R.C., Kouzarides T., Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo domain, Nature 410 (2001) 120–124. [73] Lachner M., O’Carroll D., Rea S., Mechtler K., Jenuwein T., Methylation of histone H3 lysine 9 creates a binding site for HP1 proteins, Nature 410 (2001) 116–120. [74] Nielsen A.L., Oulad-Abdelghani M., Ortiz J.A., Remboutsika E., Chambon P., Losson R., Heterochromatin formation in mammalian cells: interaction between histones and HP1 proteins, Mol. Cell 7 (2001) 729–739. [75] Akhtar A., Zink D., Becker P.B., Chromodomains are protein-RNA interaction modules, Nature 407 (2000) 405–409. [76] Chalker D.L., Yao M.C., Nongenic, bidirectional transcription precedes and may promote developmental DNA deletion in Tetrahymena thermophila, Genes Dev. 15 (2001) 1287–1298. [77] Flaus A., Owen-Hughes T., Mechanisms for ATP-dependent chromatin remodelling, Curr. Opin. Genet. Dev. 11 (2001) 148–154. [78] Marmorstein R., Roth S.Y., Histone acetyltransferases: function, structure, and catalysis, Curr. Opin. Genet. Dev. 11 (2001) 155–161. [79] McBlane F., Boyes J., Stimulation of V(D)J recombination by histone acetylation, Curr. Biol. 10 (2000) 483–486. [80] McMurry M.T., Krangel M.S., A role for histone acetylation in the developmental regulation of V(D)J recombination, Science 287 (2000) 495–498. [81] Kinoshita K., Honjo T., Linking class-switch recombination with somatic hypermutation, Nat. Rev. Mol. Cell. Biol. 2 (2001) 493–503. [82] Matzke M., Matzke A.J., Kooter J.M., RNA: guiding gene silencing, Science 293 (2001) 1080–1083. [83] Ruiz F., Vayssie L., Klotz C., Sperling L., Madeddu L., Homologydependent gene silencing in Paramecium, Mol. Biol. Cell 9 (1998) 931–943.