Copyright 0 1996 by the Genetics Society of America

A Mendelian Mutation Affecting Mating-Type DeterminationAlso Affects Developmental Genomic Rearrangements in Paramecium tetraurelitz Eric Meyer * and Anne-Marie Kellert

* Ecole Normale Supirieure, Paris,

France and tCentre de Ginitique Moliculaire, CNRS, Gif-sur-Yvette, France

Manuscript received September 20, 1995 Accepted for publication February 2 , 1996

ABSTRACT In Paramecium tetraurelzu, mating type is determined during thedifferentiation of the somatic macronucleus from a zygotic nucleus genetically competent for both types, 0 and E. Determination of the developing macronucleus is controlled by the parental macronucleus through an unknown mechanism m t p affects maresulting inthe maternal inheritance of mating types. The pleiotropicmutation cronuclear differentiation. Determination for E is constitutive in mutant homozygotes; a number of unrelated mutant characters are also acquired during development. We have examined the possibility that the mutation causes a defect in the developmental rearrangements of the germ-line genome. We show that the excision of an IES (internal eliminated sequence) interrupting the coding sequence of a surface antigen gene is impaired in the mutant, resulting in an alternative macronuclear version of the gene. Once established, the excision defect is indefinitely transmitted across sexual generations in the cytoplasmic lineage, even in a wild-type genetic context. Thus, the processes of mating-type determination and excision of this IES, in addition to their common sensitivity to the m t p mutation, show a similar maternal inheritance of developmental alternatives in wild-type cells, suggesting a molecular model for mating-type determination.

C

ILIATES have evolved a unique system of nuclear differentiation.In these unicellular eukaryotes, germinal and somatic functions are devolved to different typesof nuclei coexisting in the same cytoplasm (two micronuclei and onemacronucleus in Paramecium tetraurelia) . Micronuclei are small, diploid nuclei in which no transcription can be detected during vegetative growth. During the sexual processes of conjugation and autogamy, they undergo meiosis to produce the gametic nuclei that will transmit the germ-line genome to the following generation. The macronucleus is a large, highly polyploid nucleus that divides amitotically and where allvegetative transcription takes place. It governs the phenotype of the vegetative cell but is not transmitted across sexual generations. During meiosis of the micronuclei, it breaks up into many fragments in which DNA synthesis is progressively inhibited, although they remain transcriptionally active ( BERCER 1973) . After fertilization, new macronuclei and micronuclei differentiate frommitotic products of the zygotic nucleus. The fragments of the old (prezygotic) macronucleus areeventually diluted out during the first vegetative divisions of the zygote. Thus, all nuclei in a vegetative cell derive from the micronucleiof the previous sexual generation (for a complete description, see SONNEBORN 1974). The differentiation of a macronucleus from a diploid Correspondingauthor:Eric Meyer, Laboratoire de GenCtique MolCculaire"URAl302, Ecole Normale Superieure, 46, rue d'Ulm, 75005 Paris, France. E-mail: [email protected] Genetics 143 191-202 (May, 1996)

nucleus involves extensive and reproducible rearrangements of the germ-line genome, as well as amplification to the final ploidy level ( 1000 n ) . The rearrangements include chromosome fragmentation and formation of new telomeres in specific regions, and precise excision of numerous elements called IESs (internal eliminated sequences) from coding and noncodingsequences (for reviews, see KL.OBUTCHER and JAHN 1991; PRESCOTT1994). IESs are AT-rich sequences that are unique in the genome;they can be as short as 28 bp and appear to be noncoding. No strictlyconserved sequence element has been identified among them, except for two 5 '-TA-3' direct repeats marking their boundaries, one of which is retained after excision (Scorn et al. 1994a; STEELE et al. 1994). Little more is known about cis-acting sequence elements involved in genomic rearrangements in Paramecium, and transacting factors have yet to be identified. The mechanism of mating-type determination is a longstanding problem in Paramecium genetics. In P. tetraurelia, the two complementary types ( 0 and E ) are not determined by a Mendelian polymorphism: cells of both types can be produced in strains with entirely homozygous micronuclear genomes. The macronucleus is irreversibly determined at thetime of its differentiation for the capacity to express 0 or E, although mating type is only expressed after induction of sexual reactivity in the mature vegetative clone. Most intriguing is the maternal inheritance of mating types: after conjugation and reciprocal fertilization, the progeny of the 0 parent is almost always determined for 0, and

-

192

Keller

A.-M.

and

E. Meyer

that of the E parent almost always determined for E ( NANNEY1957). Similarly, autogamy of a single cell yields progeny of the same mating type. Numerous experiments have established that it is the old macronucleus that causes newmacronuclei differentiating inthe same cytoplasm to be determined for the same mating typeasits own (for a review, see SONNEBORN 1977). The molecular basis for the determination process is unknown. However, a similar maternal inheritance of developmental alternatives has been observed for a number of other macronucleus-determined pheno1977; SONNEBORN and typic characters ( SONNEBORN SCHNELLER 1979; BRYGOOand KELLER 1981b; EPSTEIN and FORNEY 1984; NYBERG 1986). One of these characters, the incapacity of the d48 cell line to express the A surface antigen, was shown to be determined by an alternative fragmentation pattern of the germ-line chromosome bearing the A gene, leading to the deletion of thegene from themacronucleargenome ( FORNEY and BLACKBURN 1988). This occurs in the absence of any germ-line mutation, the differentiating macronucleus simply reproducing the fragmentation pattern observed in the old macronucleus ( KOIZUMI and KOBAYASHI1989). More generally, there is growing evidence that the processing of many germ-line genomic regions is sensitive to the structure andcopy number of homologous sequences in the old macronucleus, which can result in the maternal inheritance of alternative rearrangements ( MEYER1992; KIM et al. 1994; SCOTTet nl. 1994b; YOU et al. 1994; DUHARCOURT et al. 1995). These observations raise the possibility that mating-type determination is also achieved by alternative rearrangements of the germ-line genome. To evaluate this hypothesis, we decided to study the Mendelian mutation m t p ( BRYGOOand KELLER 1981a). This recessive mutation is the only known mating-type mutation that affects the developmental process of determination itself: m t p homozygotes will express E at sexual reactivity because the mutationmakes determination for E constitutive during macronuclear differentiation, regardless of the maternal mating type. The m t p induced E determination is normally transmitted to sexual progeny in the cytoplasmic lineage, even when the wild-type m t p allele is reintroduced by conjugation. All other mutations restrict mutant homozygotes to 0, presumably by interfering with metabolic steps involved in the production of an &specific molecule from an 0 specific precursorduring sexual reactivity ( BYRNE 1973). Incontrast to m t p , they do notaffect the determination process. Indeed, mutant homozygotes formed in an E cytoplasmic lineage, although expressing 0, still carry and transmit E determination, as shownby the fact that sexual progeny will switch back to E upon reintrcduction of the wild-type allele ( BYRNE 1973) . The m t p mutation shows other unusual features. It is a pleiotropic mutation,affecting such diverse characters as growth rate, trichocyst discharge ( BRYGOOand

1981b), cell shape,pigmentation, vegetative macronuclear division, vegetative life span, and viability of postautogamous progeny (our unpublished observations) . Furthermore, maternal effects are observed in a cross with the wild-type strain. First-generation m t p homozygotes derived from the 0 wild-type parent, although usually (butnot always) determinedfor E, maintain a wild-type phenotype throughout vegetative growth. The full pleiotropy only appears in the following generation, after an additional autogamy. Similarly, those clones that were stilldetermined for0 only switch to E in the second generation ( BRYGOOand KELLER 1981a). While the usual one-generation lag between mating-type change and expression of the other phenotypic characters is not understood, it is clear that the mutation does not directly impair any vegetative function and specifically affectsthe process of macronuclear differentiation. This suggests that mtF encodes a trans acting factor involved in developmental genomic rearrangements, thereby affecting many different genes. KELLER

MATERIALS AND METHODS Cell lines and cultivation: P. tetraurelia wild-typestrains 1974) arewell-Characterized stocks d4.2 and 51 ( SONNEBORN that have been used extensively in genetic studiesof the surface antigen system. The d48 cell line was kindly provided by Dr. LINDAMARTIN. The m t p mutant has been described ( BRYGOOand JSELLER 1981a). Cells were grown in a wheat grass powder (Pines InternationalCo.) infusion medium bacterized the day before usewith Klebsiella pneumoniae and s u p plemented with 0.8 mg/liter of beta-sitosterol (Merck, Darmstadt, Germany), at 20 or 27". Basic methods of cell culture have been described ( SONNEBORN1970). Autogamy and conjugation: Autogamy was induced by starvingthecellsaftertheyhadreachedtheappropriate clonal age(30 vegetative divisions), and assessed for fragmentation of the old macronucleusby staining with a 15:l ( v / v ) mix of carmine red (0.5% in 45% aceticacid) and fast green ( 1% in ethanol). Exautogamous cells were isolated from depressions showing 100% autogamous cells, and after the first cellular division, the two caryonides were isolated and cultivatedseparately. For massautogamies, a wholedepression (-1000 autogamous cells)was transferred to bacterized mediumandgrowncollectively.Conjugation was induced by starving two clones with complementary mating types. Pairs firmly engaged in conjugation were transferred to individual depressions. The two exconjugants from each pair were is@ lated after their separation; after the first division,thetwo caryonides from each exconjugant were isolated and grown separately. Their cytoplasmic origins were determined by phenotypic differences and/or mating-type testing, using standard d4.2 mating-type tester strains. Dot-blot analyses: For each sample, 50 cells were micropipetted from depression cultures under the binocular microscope and transferredto 400 p1 of 0.4 N NaOH, 50 mM EDTA. The lysates were incubated for 30 min at 68" and loaded on a Hybond N + membrane (Arnersham, UK) using a dot-blot apparatus. The membrane was kept wet with 0.4 N NaOH for 15 min, washedin 2X SSC (SSC is 0.15 M NaCI, 0.015 M sodium citrate), and further treated as a Southern blot. Genomic DNA extraction: Cultures (400 ml) of exponentially growing cells at 1000 cells/ml were centrifuged. After being washed in Volvic mineral water (Volvic, France), the

Paramecium Chemic Rearrangements pellet was rrsrtsprntlrd i n 1 volume of mineral water, and quickly adtlrtl t o 4 volumcs oflysis solution [0.44 %I EDTA pH 9.0. 1% SIX. 0.5% S-lau~lsarcosine (Sigma), antl 1 mg/ m l pmteinasr K (Mrrck) 1 at 5.5'. The Ivsatc was incrlhatrd a t 33" fc)r 2.5 hr, gcntly extracted once with phenol, and dialysetl twicc against TI.: ( I O m M Tris-HCI, 1 m%cEDTA, pl-l 8.0) containing 20'7) ctllanol, and oncc against TE. Southern hybridization: DNA restriction antl elcctrophorc sis \vcr(*carrirtl o ~ t ;recording t to standard procedures (SAMRROOK rt nl. I ! W ) ) . DNA WRS transfrrrctl from ararose gcls to I-Iyhond N mcmhranes ( Amersham, UK) in 0.4 N NaOH aftcr dcpurination in 0.2.5 S HCI. Prehvbridimtion and hybridization were carried out i n 7% SDS, 0.5 x4 sodium phosphatc, I 'X RSA and I m%fEDTA (pH 7.2) at 6 1 " ( C H V R C I I and C.II.RERT 1984). Probeswerelabeledusing a random priming kit (Rorhringrr-lu(annhcim. Germany) t o a spccific activityof 3.10" cpm/pg. Membranes were then washed for 30 min in 0.2x SSC and 05?h SDS at 58" (for the 214-bp IES-spccific probe)or 60" (other probes) before autordtliography. PCR amplification:Amplificationswerc carried out in capped 0.5-1nl Sigma polypropvlenc tubcs, using a Perkin Elmer' (:CIIIS thermocvcler.Reactions (2.5 pl) contained the indicated amounts of genomic DNA, 1X PCR buFfer supplicd by the manrtfilcturer (Epiccntre). 200 p~ of each dSTP. 2 p~ of each oligonucleotide (:i"CA25 autogamies showed no evidence of excision ( DUHARCOURT et al. 1995) . Thus the excision defect of the Ggene IES is transmitted maternally in a wild-type genetic context, like mating type E and the d48 alternative rearrange-

Rearrangements Genomic WT

Paramecium

199

m tFE

and JAHN 1993; KLOBUTCHER et al. 1993; WIL al. 1993;YAO and YAO 1994; SAVELIEV and COX 1995) . The finding supports thehypothesis that macronucleus-determined mtpphenotypic characters areprimarily caused by rearrangement defects. However, the fraction of genomic rearrangementsaffected by the mutation is probably quite limited. The d48 alternative fragmentation pattern is not affected; neither is the mechanism of itsepigenetic transmission. Although the lack of excision ofvery short IESs, or a very partial impairment of excision, might have been overlooked in the procedure used, it is also clear that most IESs are excised in the m t p macronucleus, as only one excision defect was identified in a random survey of 30 kb of macronuclear DNA. Furthermore, a Southern blot analysis showed thenine IESs located inthe A5’ gene ( STEELEet al. 1994) to be excised in the mutant; so were two of the Bgene IESs ( S C O ~ T et al. 1994a), as shown by PCR amplification of the corresponding macronuclear sequences (data not shown). This should not come as a surprise: a mutationblocking the excision of a significant fraction of IESs would probably be completely lethal because IESs appear to be very frequent in germ-line genes (13 have been found in a total of 20 kb of micronuclear sequences encompassing the A and B surface antigen genes) . The specificity of effect of the mtF’: mutation on the one IES may mean that there areseveral classes of IESs using different excision mechanisms. Each of20 known IESs has a unique sequence. However, in a statistical analysis, nonrandom nucleotide frequencies can be detected in the first and last 14 bp. A degenerateconsensus sequence can thus be written for the ends of the IESs, whichis the same for both endsand shows similarity to the terminal bases of the inverted repeats of transposons of the Tcl /mariner family ( KLOBUTCHER and HERRICK 1995). Although the ends of the Ggene IES show a poor fit to that consensus, this is also true of other IESs that wereshownto be excised in the mutant. The explanation of the specificity of the effect of the mutation will thus have to await the molecular analysis of other mtp-dependentIESs and a better understanding of the excision mechanism. A possible role forthe mtF gene productin IES excision: IES excision may involve recognition of structural rather than sequence determinants. Similar structures could be formed with different IES end sequences, but with different stabilities, leading to different requirements for stabilization by the binding of an accessory protein. Members of the DNA chaperon class of proteins participate in the assembly of many site-specific recombination complexes ( TRAVERS et al. 1994) . Some of these proteins, including eukaryotic HMGbox proteins, can substitute for one another in different systems, such as TnlO transposition, phage Mu transposition, and phage A excisive recombination. They are thought toplay similar architectural roles through their CZEWSKI LIAMS et

J

conjugation

\ G G / ; t F +

F1

”e5,

FIGURE 7.-Recapitulation of the structure of the macronuclear G gene in the cross of a wild-type strain to the mtF’ mutant. The arrow represents the excised version of the gene ( G + ) ; the arrow with a hatched box represents the nonexcised version (G-) . The occurrence of excised and nonexcised forms strictly parallels the 0 and E mating types, respectively. In particular, the wild-type homozygous lineage derived from the m t p parent remains E and does not excise the Ggene IES.

ment. Figure 7 recapitulates the structure of the macronuclear G gene in all clones of the cross. It can be seen that the excised and nonexcised macronuclear versions exactly parallel the 0 and E macronuclear states, respectively. In particular, of the two m t p lineages issued from the cross, which have the same wild-type germline genome, one is stably 0 and regularly excisesthe G gene IES during macronuclear development and the other is stably E and never excises the IES. DISCUSSION

Effect of the mtFE mutation on developmental rearrangements: We haveshown that the excision of an IES interrupting the coding sequence of the G surface antigen gene is impaired in cells homozygous for the m t p mutation. This is the first example in ciliates of a Mendelian mutation in a transacting factor involved in developmental genomic rearrangements. It should open the way for the genetic and molecular dissection of the precise somatic excision of IESs and transposonlike elements, which appears to involve novel site-specific recombination mechanisms in other ciliates (JARA-

200

E. Meyer and A.-M. Keller

common DNA bending capacities; interestingly, the efficiencyof the replacement of the sequence-specific IHF protein by the nonspecific HU or HMGl / 2 proteins in the A intasome depends on the particular IHF binding site ( SEGALL et al. 1994). Furthermore, excision has been observed in the absence of HU or IHF by engineering intrinsically bent DNA stretches in the vicinity of IHF sites (GOODMAN et al. 1992). The m t F gene could encode such an accessory protein, facilitating the organization of an excision complex, but required only for some of the IESs. The m t p mutation would subtly decrease the efficiency of each IES excision to a different extent, depending on its particular sequence. It is worth mentioning that we have recently observed the m t p mutation to be temperature-sensitive: mtF': homozygotes obtained at 13" in the 0 wildtype macronuclear lineage do notshow anyof the m t p phenotypic characters, including mating-type change, even after one additional autogamy at this temperature (our unpublished observations). The m t p mutation could be a substitution resulting in a gene product that functions normally at 13"; alternatively, it could be a null mutation in a gene whose function is onlyrequired at high temperatures. At 18", the numberof mtF" homozygotes showing phenotypic defects is intermediate between 13and 2 7 , consistent with the idea that themutation only affects excision in a quantitative manner. The mtFgene is expressed from the zygotic genome: Mutations with maternal effects areoftenfoundin genes that are expressed from the maternal genome during development of the zygote. A maternal effect is observed in the F1 progeny of a cross between m t p and m t p homozygotes: although theheterozygotes on both sides of the cross have the same zygotic genome, the G gene IES is excised in the macronuclear genome of those deriving from the m t F + parent, butit is fullymaintained in those deriving from the m t p parent (Figure 6 ) . This stronginfluence of the old macronucleus would seem to indicate that,during nuclear reorganization, the mtFgeneis expressed in theold macronucleus and not in the developing macronucleus. However, in the autogamy of the F1 heterozygote derived from the wild-type parent, the new macronuclei of both types of F2 homozygotes ( m t F / m t p and m t p - / m t p ) differentiate in the presence of the same mtF"/ m t p old macronucleus (see Figure 7 ) . Therefore the fact that m t p F4 homozygotes excise the IES, whereas m t p F2 homozygotes mostly do not, shows that the mtFgene is expressed mainly, if not exclusively, from the zygotic genome. Thus the influence of the old macronucleus on thedevelopmental excision process, observed in the F, progeny of the cross, cannot be explained by a strictly maternal expression of the gene. The small amount of excision observed in the F2 m t p macronucleus might nevertheless be due to a small contribution of the old macronucleus to the synthesis of the mtFgene product. However, if the mutation only partially affects excision,

the relative amounts of excised and nonexcised versions of the G gene in the F2 macronucleus could simply reflect the extent of the effect of the mutation on the excision of this IES at this temperature. Epigenetic regulation of developmentally controlled IES excision: We have shown that F:+clones obtained by an additional autogamy of F2mtF': homozygotes with mixed macronuclei completely failtoexcise the IES (Figures 3 and 7 ) . This difference between F2 and F3 mtF': homozygotes inthe cytoplasmic lineage of the wild-type parent, as well as the total lack of excision in the F1 heterozygote derived from the mtfi".' parent, can be explained by a different mechanism that is independent of the m t p mutation. Indeed, Figure 7 shows that no excision occurs in the entire cytoplasmic descent of the m t p parent, regardless of mtFgenotype; in particular, we have not observed any excision in the m t F homozygous lineage after >25 successive autogamies. It was recently shown that the presencein the old macronucleus of the Ggene IES sequence itself quantitatively inhibits the excision of the homologous IES in the differentiating macronucleus: transformation of a wildtypevegetative macronucleus by microinjection of a cloned fragment containing the IES sequence yields postautogamous progeny that maintain the IES (DUHARCOURT et al. 1995) . The maternal effects observed in the mtF" X mtF+ cross are readily explained by this previously unsuspectedcontrol level. It is because themacronuclear genome of the m t p parent contains the IES, and not because of its m t p homozygous genotype, that derived F1 heterozygotes donot excise the IES. The effect should also apply in the autogamy of F2 mtP* homozygotes in the cytoplasmic descent of the wild-type parent, because their macronuclei contain a sizable number of nonexcised copies. During differentiation of the F3 m t p macronucleus, this should add to the reduced excision rate due to the m t p mutation to abolish excision totally. Similar effects could account for thephenotypic difference between F2 and F3 m t p homozygotes. If the effect of the m t p mutation on the rearrangement of other genes is also partial, as suggested by the quantitative differences observed, the fraction of correctly rearranged copies in the F2 macronucleus maywell be enough to ensure near-normal expression (it was not possible to test this in the case of the G gene because the G serotype cannot be reliably induced). The appearance of the pleiotropic characters in theF:+generation would reveal a quantitative aggravation of other rearrangement defects. Maternalinheritance of mtFE-inducedphenotypic characters: If other rearrangement defects are aggravated by the same epigenetic mechanism, mtp-induced phenotypic characters should also be maternally inherited in a wild-type genetic context. Two of them, slow growth and trichocyst nondischarge, were indeed shown to be maintained in mtplderived F1 heterozygotes, and

RearrangementsGenomic

Paramecium

further in m t F homozygotes obtained by autogamy of these heterozygotes ( BRYGOOand KELLER 1981b). It should be noted that thetransmission efficiency differs for these different characters. While mating-typeE and the Ggene excision defect are very stably maintained, the trichocyst nondischarge character reverses with a high frequency at autogamy and can only be maintained with regular selection; the more variable slowgrowth phenotype is bettermaintained,but also reverses easily. This also implies that the differentcharacters can be transmitted independently from each other: mating-type E and the Ggeneexcision defect are maintained in lineages that have reverted for the trichocyst or slow-growth characters. It is already known that mating-type E can be stably transmitted in lineages that excise the Ggene IES. Conversely, the maintenance of the Ggene excision defect in an E lineage is not affected in rare ( 1/ 3000) cases of spontaneous matingtype change at autogamy (S. DUHARCOURT, personal communication).Thus,each of these mtp-induced characters is self-maintained in genetically wild-type cells, implying that different specific factors transmit information to the developing macronucleus. Under the assumption that these characters are also determined by alternative rearrangements, this specificity is consistent with the hypothesis that theepigenetic mechanism underlying maternal inheritanceinvolves a direct transfer of sequence information from the old macronucleus to the developing one, as suggested in a study of the epigenetic regulation of excision of the Ggene IES ( DUHARCOURT et al. 1995). A model €or mating-type determination: Mating-type determination is thought to be achieved during macronuclear differentiation through the stabilization of one master-switch gene under one oftwo alternative forms, one in which it can be activated, leading to the expression of &specificfunctions during sexual reactivity, and one in which it cannot, resulting in the 0 default state (SONNEBORN1974, 1977; BRYCOO1977). This master-switch gene was tentatively identified as mtD, a gene showing the required characteristics: it appears to be active in E cells and inactive in 0 cells, and one ofits products was postulated to be responsible for the maternal inheritance of its expression pattern ( BRYGOO1977).Although both mating types are maternally inherited, an important asymmetry was noted. Determination for E results from the action upon the developing macronucleus of cytoplasmic &determining factors produced by the E old macronucleus, and thus is truly inherited from the cytoplasmic parent, while determination for 0 occurs in the absence of these factors through the action of nuclear Odetermining factors ( BRYGOOet al. 1980). It was further proposed that the mtFgene encodesone of these nuclear factors ( BRYGOOand KELLER 1981a). The strictly correlated appearance of mating type 0 and the excised version of the G gene on the one hand

201

and of mating type E and the nonexcised version on the other, at all stages of the m t p X mtF‘ cross, suggests a molecular model for mating-type determination. Determination for 0 would be accomplished by the developmental excision of an IES from the mtD gene, a processinvolving themtFencoded nuclear factor. The excision would be affected by the m t p mutation, resulting in constitutive determination for E in m t p homozygotes. The occurrence of some 0 or selfer clones among first-generation m t p homozygotes appearing in an 0 cytoplasmic lineage could be attributable to excision of the IES from a variable number of mtD copies in these clones, reflecting the partial impairment of excision caused by the m t p mutation. This would bear out anearly interpretation in which the macronuclei of selfer caryonides were held to be “mixed”, containing different proportions of stable 0 and E “quantities” ( NANNEY1957). In wild-type ( mtF‘ / m t F ) cells, the m t F gene product and other nuclear factors necessary for IES excision have to be present during macronuclear differentiation, because they are required for the rearrangement of other genes. The &determining factors produced by the old macronucleus of wild-type E cells would specifically prevent excision ofthe mtDgene IES. The maternal inheritance of mating types would simply result from the epigenetic self-maintenance of the mtDgene IES,similar to that of the Ggene IES. This is also consistent with NANNEY’S conclusion that the same macronuclear “quantities” that control vegetative mating types are also responsible for the effect of the old macronucleus on the determination of the differentiating macronucleus ( NANNEY1957). If this model is correct, there is a major difference between the effect of the m t p mutation on the expression of the m D gene and its effect on the expression of the G gene. Indeed the nonexcised version of the m D gene, being the activatable one, should be dominant over the excised version, whereas the nonexcised version of the G gene, being defective, should be recessive. Whilemost known IESs have to be removed before their host genes can be expressed, it is conceivable that an IES might contain promoter elements necessary for the activation of the mtD gene. This dominance of the m t p version of the mtD gene would explain why most m t p homozygotes appearing in the 0 cytoplasmic descent of the wild-type parent show a change in mating type in the F2 generation, when the macronucleus contains both the m t p and the mtpversions of the genes, whereas all other m t p phenotypic characters appear only after one additional autogamy, when allmacronuclear copies of the genes present the mtpversion. We thank LAURENCE AMAR, JANINE BEISSON,MIREILLE BETERMIER, FRANCOIS W O N , MARC DREYFUS,KARINE DUBRANA, SANDRADUHARCOURT, and ANNE LE MOUELfor critical readings of the manuscript and helpful discussions. This work was supported by grant No. 30 from the Groupement de Recherches et d’Etudes sur les Genomes, BP25, 91193 Gif-sur-Yvette cedex, France.

E. Meyer and A.-M. Keller

202 LITERATURE CITED

BERGER, D., J. 1973 Nuclear differentiation and nucleic acid synthesis in well-fed exconjugants of Paramecium aurelia. Chromosoma 42: 247-268. BRYGOO, Y.,1977 Genetic analysis of mating-type determination in Paramecium tetraurelia. Genetics 87: 633-653. BRYCOO,Y.,and A,". KELLER,1981a Genetic analysisof matingtype differentiation in Paramecium tetraurelia. 111. A mutation restricted to mating-type E and affecting the determination of mating type. Dev. Genet. 2 13-22. BRYCOO, Y., and A,". KEI.LER,1981b A mutation with pleiotropic effects on macronuclearly differentiated functionsin Paramecium tetraurelia. Dev. Genet. 2: 23-34. BRYGOO, Y., T. M. SONNEBORN, A,". KEILER, R. V. DIPPEI.and M. V. SCHNEI.I.LR, 1980 Genetic analysis of mating-type differentiation in Paramecium tetraurelia. 11. Role of the micronuclei in mating-type differentiation. Genetics 94: 951-959. BYRNE,B. C . , 1973Mutational analysis of mating-type inheritance in syngen 4 of Paramecium aurelia. Genetics 74: 63-80. CHURCH, G. M., and W. GILBERT, 1984 Genomic sequencing. Proc. Natl. Acad. Sci. USA 81: 1991-1995. DUHARCOURT, S. D., A. BUTLER and E. MEYER,1995 Epigenetic selfregulation of the excision of an internal eliminated sequence in Paramecium tetraurelia. Genes Dev. 9: 2065-2077. EPSTEIN,L. M., and J. D. FORNEY, 1984 Mendelian and non-Mendelian mutations affecting surface antigen expression in Pammecium tetraurelia. Mol. Cell. Biol. 4 1583-1590. FORNEY, J. D., and E. H. BUCKBURN,1988 Developmentally controlled telomere addition in wild-type and mutant paramecia. Mol. Cell. Biol. 8: 251-258. GOODMAN, S . D., S. C. NICHOISONand H. A. NASH,1992 Deformation of DNA during sitespecific recombination of bacteriophage X: replacement of IHF protein by HU protein or sequence-directed bends. Proc. Natl. Acad. Sci. USA 89: 11910-11914. JARACZEWSKI, J. W., and C. L. JAHN, 1993 Elimination of Tec elements involves a novel excision process. Genes Dev. 7: 95-105. KIM, C. S., J. R. PREER, JR. andB. POI.ISKY,1994 Identification of DNA segments capable of rescueing a non-Mendelian mutant in Paramecium. Genetics 136: 1325-1328. KIX)BUTC:IILR, L. A,, and G. HERRICR, 1995 Consensus inverted terminal repeat sequence of Paramecium IESs: resemblance to tcrmini of Tcl-related and Euplotrs transposons. Nucleic Acids Res. 23: 2006-2013. KI.OBLI'I'(:HEK,L,. A,, and C. L. JAHN,1991 Developmentally controlledgenomic rearrangements in ciliated protozoa. Curr. Opin. Genet. Dev. 1: 397-403. KLOBLITCHER, L. A., L. R. TURNER and J.LAPIANTE, 1993 Circular forms of developmentally excised DNA in Eup1ote.r crassus have a heteroduplexjunction. Genes Dev. 7: 84-94. KOIZLJMI,S . , and S . KOBAYASFII, 1989 Microinjection ofplasmid DNA encoding the A surface antigen of Paramecium tetraurelia restores the ability to regenerate a wild-type macronucleus. Mol. Cell. Biol. 9: 4398-4401. MEYER,E., F. CARON and A. BAROIN, 1985 Macronuclear structure of the G surface antigen gene anddirect expression of its repeated epitopes in Escherichia coh. Mol. Cell. Biol. 5: 2414-2422. MEYXR, E., 1992 Induction of specific macronuclear developmental

mutations by microinjections of a cloned telomeric gene inParamecium primaurelia. Genes Dev. 6: 21 1- 222. NANNLY, D. L., 1957 Mating-type inheritance at conjugation in variety 4 of Paramecium aurelia. J. Protozool. 4: 89-95. NIELSEN, E., Y. You and J. D. FORNEY, 1991 Cysteine residue periodicity is a conserved structural feature of variable surface proteins from Paramecium tetraurelia. J. Mol. Biol. 222: 835-841. NYBERG, D.,1986 Cytoplasmic inheritance of temperature sensitivity in a wild stock of Parameciumprimaurelia. J. Protozool. 3 3 38-41. PRAT,A,, 1990 Conserved sequences flank variable tandem repeats in two alleles of the G surface protein of Paramecium primaurelia. J . Mol. Biol. 211: 521-535. PRAT,A,, M. KATINKA, F. CARONand E. MEYER, 1986 Nucleotide sequence of the Paramecium pimaurelia G surface protein. A huge protein with a highly periodic structure. J. Mol. Biol. 189: 47-60. PRESC:OTT,D.M., 1994 The DNAof Ciliated Protozoa. Microbiol. Rev. 58: 233-267 SAMBKOOK,J., E. F. FRITCHand T. ULATIS, 1989 Molecular Cloning: A Laboratoly Manual, Ed. 2. Cold Spring Laboratory Press, Cold Spring Harbor, N Y . SAVEI.IEV, S. V., and M. M. COX,1995 Transient DNA breaks associated with programmed genomic deletion events in conjugating cells of Tetrahymena thennqbhila. Genes Dev. 9: 248-255. Scorn, J., C . LEECKand J. FORNEY,1994a Analysis of the micronuclear B type surface protein gene in Paramecium tetraurrlia. Nucleic Acids Res. 22: 5079-5084. Scorr, J. M., K. MIKAMI, C. L. LEECRand J.D. FORNEY, l994b NonMendelian inheritance of macronuclear mutations is gene-specific in Paramcium tetraurelia. Mol. Cell. Biol. 1 4 2479-2484. SEGAI.L, A. M., S . D. GOODMAN and H. A. NASH.1994 Architectural elements in nucleoprotein complexes: interchangeability of specific and non-specific DNA binding proteins. EMB0.J. 13: 45364548. SONNEBORN,M., T. 1970 Methods in Paramecium research. Methods Cell Physiol. 4: 241-339. SONNEBORN, T. M., 1974 Paramerium aurelia, pp. 469-594 in Handbook of Genetics: Plants, Plant Viruses and Protists, Vol. 2, edited by R. C. KING.Plenum Press, New York. SONNEBORN, T. M., 1977 Genetics of cellular differentiation: stable nuclear differentiation in eucaryotic unicells. Annu. Rev. Genet. 11: 349-367. SONNERORN, T. M., and M. V. SCHNEILER, 1979 A genetic system for alternative stable characteristics in genomically identical homozygous clones. Dev. Genet. 1: 21-46. STEELE, C. J., G. A. BARKOCY-GALLAGHER, L. B. PEER and J.R. PREER, JR., 1994 Developmentally excised sequences in micronuclear DNA of Paramecium. Proc. Natl. Acad. Sci. USA 91: 22.55-2259. TRAVI~RS, A. A,, S. S . NER and M. E. A. CHURCHILL, 1994 DNA c h a p erones: a solution to a persistence problem? Cell 77: 167-169. K., T. G. DOAKand G. HERRICK, 1993 Developmental WILLJAMS, precise excision of Oxytricha trifallax telomere-bearing elements and formation of circles closed by a copy of the flanking target duplication. EMBO J. 12: 4593-4601. YAO, M.-C., and C.-H. YAO, 1994 Detection of circular excised DNA deletionelements in Tetrahymena thennophila duringdevelop ment. Nucleic Acids Res. 22: 5702-5708. You, Y., J. SCOTT and J. FORNEY,1994 The role of macronuclear DNA sequences in the permanent rescue of a non-Mendelian mutation in Paramecium tetraurelia. Genetics 136: 1319-1324. Communicating editor: S. L. AILEN