Proc. Nati. Acad. Sci. USA Vol. 83, pp. 7386-7390, October 1986 Genetics Two mechanisms for directional gene conversion (heteroduplex DNA/mismatch repair/meiotic recombination/Ascobolus immersus) HANAFY HAMZA*, ANGELOS KALOGEROPOULOS, ALAIN NICOLASt, AND JEAN-LUC ROSSIGNOL Interactions Moldculaires Gdnomiques, Universitd Paris-Sud, 91405F Orsay-Cedex, France Communicated by Franklin W. Stahl, June 12, 1986

ABSTRACT G234 is a large silent deletion located in the (the longer strand) whether it was that of the donor or middle of gene b2, which controls spore pigmentation in recipient strand. We conclude that the mechanism that Ascobolus immersus. Its gene conversion directionality was produces gene conversion events from hDNA containing studied in asci, which show evidence of heteroduplex DNA at single-base mismatches differs from that which operates on flanking markers, and was compared to the behavior of closely hDNA containing large nonhomologies. We propose that linked single-base-pair insertions or deletions. We found that mismatch correction arises via single-strand excision and via with the G234 deletion, the genotype of the donor strand in the double-strand-break repair, respectively. heteroduplex is preferentially recovered, irrespective of its G234 or wild-type nature, whereas with single-base-pair inser- MATERIALS AND METHODS tions or deletions, the direction of conversion favors one genotype, whether it was the donor or the recipient strand. We Principle for Gene Conversion Studies in Ascoboius. Each conclude that there exists two mechanisms for directional gene meiosis gives rise to an ascus containing four pairs of sister conversion, the "donor-directed" conversion mechanism be- spores-the products of two meiotic divisions followed by a ing epistatic to the "genotype-directed" one. We discuss these single postmeiotic mitosis. This means that for any gene the data with regard to models for mismatch repair. eight strands of DNA that comprised the four DNA duplexes at meiosis I are recovered in separate spores. Genetic studies in fungi have provided a description of the Gene conversion events lead to non-Mendelian segrega- mechanism of genetic recombination, in particular the proc- tions (NMS). NMS is defined as departures from the 4:4 (four ess of gene conversion (1-5). In this paper, we examine the wild type:four mutant) Mendelian segregation of the meiotic number of mechanisms that produce gene-conversion events products in mutant x wild type crosses. NMS corresponds to and affect its directionality. postmeiotic segregation (PMS)-i.e., 5:3 and aberrant 4:4- Gene conversion of some Ascobolus mutants is directional and to conversion asci (6:2). In aberrant 4:4 asci, two meiotic (6, 7). In the progeny ofwild-type x mutant crosses, putative products segregate at the first postmeiotic division. In 5:3 single-base-pair (1 bp) insertion mutations give an excess of segregations (5+:3m or 3+:5m), only one meiotic product conversion towards the mutant [(two wild type:six mutant) > segregates at the first postmeiotic division; 6:2 segregations (six wild type:two mutant), or 2+:6m > 6+:2m], whereas are of two kinds: 6+:2m or 2+:6m, depending on the 1-bp deletion mutations give an excess towards wild type direction of the conversion event towards wild type or (6+:2m > 2+:6m). Studies in the b2 ascospore color locus mutant, respectively. Spore color pigmentation and morphol- suggested that this disparity (i.e., excess ofconversion in one ogy markers can be scored by visual inspection of asci. This direction) results from heteroduplex formation, followed by allows us to analyze large numbers of meiotic events. an obligatory mismatch correction involving preferential Crosses. Media and culture methods have been described excision of the shortest strand (8, 9). By comparison, the (15, 16). All pairs of crosses but one (experiment II) were conversion behavior of the large deletion mutation b2,G234 performed in isogenic conditions. We used the protocol of was found unusual on two respects: (i) the G234 deletion in Hastings et al. (17) in experiments I, III, IV, and V and that gene b2 exhibits a strict parity in the direction of conversion of Rossignol and Paquette (18) in experiments VI-VIII. (6+:2m = 2+:6m), whereas closely linked 1-bp insertion or Mutations. The strains used belong to the stock 28 of deletion mutations exhibit a 10-fold disparity, and (ii) G234 Ascobolus immersus (19). The white-spore mutations 17 and imposes its own behavior on all tightly linked point mutations A4 map in the left and right part of the gene b2, respectively including 1-bp insertions or deletions (10). This suggested (20). The G234 deletion mutation has a wild-type phenotype that the conversion of point and nonpoint mutations may (brown spores). It lies in the middle part of b2, close to the arise via distinct pathways. E group of intragenic suppression (10). The double mutations In order to discriminate between various recombination EOEI and E2EJ show a pink-spore phenotype distinguishable models (11-14) presented in the Discussion section, we have either from wild type and from the white-spored single-site studied the formation ofheteroduplex DNA (hDNA) flanking mutations EO, El, and E2. These mutations are assumed to be these various mutations and found that asymmetric hDNA 1-bp additions (El) and 1-bp deletions (EO and E2). Recom- can form with similar frequencies on either side of them. bination between EO, El, E2, and G234 is extremely rare (10). Then we investigated the directionality ofgene conversion by Rndl.2 and vag4 mutations were used as additional mark- examining the segregation ofG234 or point mutations in these ers for the detection of aberrant 4:4 asci (21) and the individual asci that show evidence of flanking hDNA. We found that with G234, the genotype ofthe donor strand (in the Abbreviations: hDNA, heteroduplex DNA; NMS, non-Mendelian heteroduplex) was preferentially recovered, irrespective of segregation; PMS, postmeiotic segregation; 1-bp, single base pair; its G234 or wild-type nature, whereas with 1-bp insertion or x+:ym, x wild type:y mutant progeny; xC:yW, x colored-spore:y deletion, the direction of conversion favored one genotype white-spore segregation; xB:yP:zW, x brown-spore:y pink-spore:z white-spore segregation. *Present address: Faculty of Agriculture, Department of Genetics, The publication costs of this article were defrayed in part by page charge Monofiya University, Shebin el Kom, Egypt. payment. This article must therefore be hereby marked "advertisement" tPresent address: Department of Molecular Biology, Massachusetts in accordance with 18 U.S.C. §1734 solely to indicate this fact. General Hospital, Boston, MA 02114. 7386 Downloaded by guest on September 25, 2021 Genetics: Harnza et al. Proc. Natl. Acad. Sci. USA 83 (1986) 7387

distinction of sister spore pairs. RndJ, vag4, and b2 genes are a) genetically unlinked (22). A C. Accuracy of the Spore Color System. Colored spores (195; a) .1-i 17 m A4 0 brown or pink) randomly isolated from mixed pairs in 0 o IV were and strains were experiment germinated, analyzed 0 O0 m by backcrosses for their genotype. Among 121 brown spores, 17 * A4 118 were confirmed as carrying the mutation G234, and 69 a)0) among 74 pink spores were confirmed as carrying the double SE'a) mutation E2EJ. Therefore, the fraction ofphenocopies is low L LI + + + 0 S enough (4%) to infer an EE genotype safely from a pink spore c .a, and a G genotype from a brown spore. 0 + + + Genotype of Mixed Spore Pairs. The genotype of the white S 4.'- spores at 17 and A4 was determined by backcrossing the S + + + white-spore strains with 17 and A4. The white spore should .IJ r- have the same genotype as its colored partner because the 0 middle-marker configurations studied never give PMS in their conversion pattern (10, 17). This assumption was o o , 7 + AA verified by backcrossing with wild-type and El strains. 0 17 + A4 o Altogether, 1080 -mixed pairs isolated in 5 colored-spore:3 C; white-spore segregation (5C:3W) and aberrant 4C:4W asci from experiments I, IV, and VII were analyzed. All but 2 5c showed the same middle-marker genotype for the two sister 0 + spores. Thus, the genotype ofthe middle marker in the mixed a) *0+:* + S r pair can be inferred from the phenotype ofthe colored spore. 0 0 In experiment III, the G234 or wild-type middle-marker genotype of the colored spore was determined by crossing to o + * + a strain carrying GJ: the absence of PMS in the progeny indicates a G234 genotype, whereas its presence indicates the wild-type genotype (23). B RESULTS FIG. 1. Protocol for studying gene conversion at a middle marker by selecting for flanking hDNA formed on one duplex. Asymmetrical Data reported in this paper are based on eight experiments hDNA formed at the two flanking mutations 17and A4 during meiosis (I-VIII). Each experiment comprised a pair ofcrosses (a and was detected in crosses 17A4 x + + among 5C:3W spore asci (24). b) listed in Table 1. All crosses involved two flanking markers In these asci, the mixed pair of spores reflects hDNA, with the and a middle marker. The flanking markers 17 and A4, colored spore reflecting the invading strand and the white spore in were used to reflecting the recipient one. The drawings correspond to cases where coupled cis configuration, detect meiosis in the white spore in the mixed pairs is 17 A4. (A) Middle-marker which hDNA was formed on either one (asymmetric hDNA) mutation (m) on the recipient molecule. (B) Middle marker mutation or both (symmetric hDNA) ofthe interacting DNA duplexes. (m) on the donor molecule. 5C:3W asci indicated asymmetric heteroduplex formation (Fig. 1), and aberrant 4C:4W asci indicated symmetric heteroduplex formation (Fig. 2). Between the 17 and A4 sites, experiments VI-VIII showed no effect of the E2E1/+ all crosses contained one of three markers, the double-point heterozygosity as compared with E2EJ homozygosity but mutations EOEI and E2EJ and the deletion mutation G234. showed a significant increase of5C:3W in the presence ofthe Crosses a and b in experiments I-V used the same genetic E2EJ/G234 heterozygosity as compared with E2EJ homozy- markers but differed in their relative arrangement. This gosity and E2EJ/+ heterozygosity. experimental scheme placed the mutant allele of the central This suggests that hDNA covering both flanking markers is marker on the recipient strand ofthe hDNA in one cross and on the donor in the reciprocal cross (Fig. 1 A versus B). Conversions Restorations

Experiments VI-VIJI explored the gene-conversion behavior ---.+ ---.oU of different middle markers when homozygous or heterozy- 17 + A4 gous and their effect on heteroduplex formation at flanking o o o o markers. All experiments reported here are based on o 0 17 + A4 o ascospore color observations of mixed pairs of spores found in 5C:3W and aberrant 4C:4W asci. 17 1A4 Asymmetric Heteroduplexes Can Form on Either Side of the or Large Deletion Heterozygosity. The frequency of 5C:3W asci 0 was determined in all experiments (Table 2). The results of 0 00 + + 0 0 Table 1. Pairs of crosses performed in experiments I-VIII 0 To + + 0 Exp. Cross a Cross b FIG. 2. Protocol for studying gene conversion at a middle marker I 17 E2 A4 x + + + 17 + A4 x + E2E + by selecting for flanking hDNA formed on the two interacting II 17EOEI A4 x + + + 17 + A4 X + EOEI + duplexes. Symmetrical hDNA involving both 17 and A4 sites during III 17 G234 A4 x + + + 17 + A4 X + G234 + meiosis was detected among aberrant 4C:4W asci (3:3:1:1 segrega- IV 17 E2E A4 x + G234 + 17 G234 A4 x + E2EJ + tion for the spore-color-spore-shape phenotypes) (24). The drawings V 17 EOEJA4x + G234 + 17 G234 A4 x + EOEJ + correspond to cases in which the white spores in the two mixed pairs VI 17E2E A4 x + G234 + 17E2E A4 x + + + are 17 A4. Those asci with aberrant 4:4 segregation at both flanking were marker VII 17 E2E A4 x + G234 + 17 E2E A4 x + E2E + markers analyzed for the segregation of the middle (mi). The donor-versus-recipient origin of strands could not be deter- VIII 17E2E A4 x + + + 17E2E A4x + E2E + mined. Downloaded by guest on September 25, 2021 7388 Genetics: Harnza et al. Proc. Natl. Acad. Sci. USA 83 (1986)

Table 2. Distribution of 5C:3W asci in different phenotypes Frequency B:P:W phenotypes of 5C:3W ascii Total of 5C:3W ______Exp. Genotype* asci (per 1000) 5:0:3 4:1:3 1:4:3 0:5:3 (3:2 + 2:3):3W Ia E2EJ/+ 4,376 17 18 38 0 0 3 lb +/E2E1 23,912 19 0 0 95 327 6 Ha EOEJ/+ 12,000 14 86 67 0 0 9 HIb +/EOE- 3,936 16 0 0 25 36 2 MIla G/+ 3,000 19 58 0 0 0 0 IlIb +IG 2,000 22 44 0 0 0 0 IVa E2EI/G 7,000 15 (130) (42) (0) (0) (1) IVb G/E2EJ 6,000 16 (0) (0) (48) (126) (0) Va EOEI/G 2,000 19 34 4 0 0 0 Vb G/EOEI 2,000 17 (0) (0) (6) (51) (0) VIa E2EI/G 4,000 20 60 18 0 0 0 VIb E2EI/+ 4,000 15 18 40 0 0 1 VIIat E2EI/G 6,000 24 - - VIIbf E2E1 5,000 14 VIIIat E2EI/+ 5,000 10 VIIIbt E2E1 5,000 8 - *The first genotype is that associated with 17 and A4, the second is that associated with wild type. tThe numbers in parentheses were obtained from samples of total asci larger than those given in column 2 that were used to determine the frequency of 5C:3W asci. fIn experiments VII and VIII, the 5C:3W asci were counted without distinguishing the brown spores from pink spores. slightly more frequent when G234 is involved in the middle- "Allele Genotype"-Directed and "Donor"-Directed Gene marker heterozygosity. Conversion. 5C:3W asci described above were analyzed for No statistical difference was found between reciprocal the middle-marker genotype (Table 3). In each ofthe crosses, crosses of experiments I-V. This shows that none of the both alternate middle-marker genotypes were recovered in reciprocal configuration of the five middle-marker heterozy- the mixed pairs: E2EJ and + in experiment I, G234 and + in gosities bias hDNA formation in either parental duplex DNA experiment III, and E2EJ and G234 in experiment IV. This (i.e., one duplex acts as a donor, and one acts as the shows that conversion events of these various heterozygosi- recipient). This will allow the comparison of the middle- ties occurred in meioses with flanking heteroduplexes. marker segregation between reciprocal crosses. The middle marker, which is the most frequent among the In order to test the assumption that the mixed pairs in a double PMS, is also the most frequent among the other SC:3W asci arose from a large proportion ofthe heteroduplex classes of flanking-marker segregation (PMS at one site with DNA flanking the middle marker, we analyzed a random conversion at the other). Thus, the segregation of the middle sample of mixed pairs isolated from experiments I, III, and marker observed within total 5C:3W asci accurately reflects IV for the segregation of flanking markers 17 and A4 (Table that occurring among the fraction of 5C:3W asci, where 3). The fraction of double PMS ranged from 62% to 78% as double PMS unambiguously shows the presence of flanking expected, since 17 and A4 yielded >90% PMS when individ- heteroduplex. On the basis of this result and the accuracy of ually crossed with wild type (15). The other 5C:3W asci the ascospore color system (Materials and Methods), we resulted from a PMS event at 17 (5+:3m segregation in the argue that we can determine whether heteroduplex DNA at octad) and a conversion towards wild type at A4 (6+:2m these mutant sites has been processed to yield conversion or segregation in the octad) or vice versa. The similar fraction restoration events by observation of ascospore color segre- of events in gations. double PMS crosses involving G234 to that in The various brown:pink spore segregations within colored crosses with EIE2 shows that the efficiency of formation of spores in 5C:3W asci (xB:yP:zW) are given in Table 2. In flanking heteroduplex DNA is not affected by this large experiments I, II, and IV-VI involving the EE mutations, the nonhomology. 5B:OP:3W and OB:5P:3W asci were conversion events for the Table 3. Genotype of mixed pairs of spores from 5C:3W asci middle marker: the mixed pair had the central-marker gen- otype of the invading or donor DNA duplex. In contrast, the Flanking marker genotypes, % of 4B:1P:3W and 1B:4P:3W asci showed Mendelian segregation Middle-marker mixed pairs analyzed for the middle marker (restoration events): the mixed pair had Exp. genotypes 17A4/++ + A4/++ 17+/++ n the genotype ofthe recipient duplex. The proportion ofdonor and recipient middle-marker genotypes among mixed pairs of Ia EE 54 7 9 79 5C:3W asci for the five heterozygosities tested are given in ++ 24 5 1 Table 4. lb EE 59 11 5 158 In experiment I, an excess ofthe E2EJ over the + genotype ++ 17 1 7 was observed in both reciprocal crosses (2.1- and 3.3-fold MIIa G 18 0 0 excess in crosses a and b of experiment I, respectively), + 55 9 18 regardless of whether the double mutant was on the donor or IIIb G 76 0 8 12 recipient strand. This relative frequency of the E2EJ geno- + 8 0 8 type versus wild type reflects its gene conversion pattern (in IVa G 48 23 9 67 a cross without flanking markers): E2EJ crossed to + EE 14 5 1 genotype gave a 3-fold excess of conversion to E2EJ (2+:6- IVb G 19 2 5 84 E2EJ) relative to conversion to + (6+:2E2E1) (17). In EOEJ EE 47 15 12 X + crosses, EOEI showed parity: 6+2EOEI and 2+:6EOEJ n, Number of mixed pairs analyzed. were recovered with equal frequencies (9). In asci with Downloaded by guest on September 25, 2021 Genetics: Harnza et al. Proc. Natl. Acad. Sci. USA 83 (1986) 7389

Table 4. Distribution of recipient and donor genotypes for the Table 5. Absolute frequencies of aberrant 4C:4W asci and middle marker in mixed pairs from SC:3W asci double aberrant 4:4 at 17 and A4 with various Genotype distribution, middle-marker configurations Genotype % Absolute frequencies per 103 Exp. Recipient Donor Recipient Donor Double Ia E2EJ + 67* 33 Aberrant aberrant lb + E2EJ 23 77* Exp. Genotype* nt 4C:4Wt 4:40 Samplet Ila EOEI + 44 56 Ta E2EJ/+ 3000 14 3.7 177 Ilb + EOEI 41 59 Tb +/E2EJ 2000 15 2.6 153 IIIa G234 + 18 82* Ila EOEI /+ 4000 18 3.4 127 ITlb + G234 16 84* VIIa E2EM/G 6000 9 0.6 63 IVa E2EJ G234 25 75* VTIb E2EJ 5000 27 7.2 62 IVb G234 E2EJ 26 74* VIIIa E2EI/+ 5000 9 2.5 32 Va EOEI G234 11 89* VIIb E2EJ 5000 23 8 28 Vb G234 EOEI 11 89* *The first genotype is that associated with 17 and A4, and the second VIa E2EJ G234 24 76* is that associated with wild type for the flanking mutations. VIb E2EJ + 68* 32 tThe absolute frequency of aberrant 4C:4W was calculated from a *The number of this genotype is statistically higher than that of the sample of n asci. other genotype within the analyzed sample (P < 0.05). tThe absolute frequency of double aberrant 4:4 (involving both 17 and A4) was calculated after analyzing the genotype of a sample of aberrant 4C:4W asci (last column) that was isolated from a large flanking heteroduplex DNA, it gave as many EOEI as + sample of total asci. genotypes (experiment II, Table 4). We conclude that the directionality of gene conversion is not affected by the 4:4 segregation. These represent 25 of 48, 19 of 26, and 16 of presence of flanking heteroduplex DNA. 24 events in experiments Ia, Ib, and Ila, respectively. Asci The behavior of the G234 heterozygosity contrasts with where both mixed spore pairs are EE or + are conversion that of the EE mutants. In experiment III, the favored events towards EE or + (6:2 segregation). In crosses Ia and middle-marker genotype within the mixed pair was different Ib, conversion to E2EJ was more frequent than conversion to in the reciprocal crosses: G234 genotype was in excess in + (24 and 6, respectively), as expected from the E2EJ NMS experiment lIla, but + was in excess in experiment IIIb. This pattern and from the excess of E2EJ over + among mixed corresponds to a 4-fold excess of the donor genotype in both pairs in 5C:3W asci. In cross Ila, we did not expect a crosses. A similar situation was encountered in experiments difference between conversion to + and to EOEJ because IV and V: the donor genotype was always found in excess, EOEI gives parity in its NMS pattern: the results (5 conver- irrespective of its G234 or EE nature. This excess was 3-fold sions to +, 3 conversions to EOEJ) agree with this expecta- in experiment IV and 9-fold in experiment V. This indicates tion. that G imposes its conversion pattern on E2EJ and EOEI, as The extremely low frequency ofdouble aberrant 4:4 asci in confirmed in experiment VI, where the G234/E2EJ and the presence of the G234 middle marker prevented us from +/E2EJ heterozygosities are compared with that ofE2E1 on analyzing the behavior of this marker. the recipient duplex in both cases. E2EJ genotype was found to be in excess with the +/E2EJ heterozygosity but in deficit DISCUSSION with the G234/E2EJ one, where the donor was again favored. We conclude that the directionality of conversion of G234 is The key result in these experiments is that the same donor-directed, in contrast to that of point mutations, which heterozygosity, G234/+, exhibits opposite directionalities of are genotype-directed, and that donor-directed conversion is gene conversion in a pair ofreciprocal crosses. The formation epistatic to genotype-directed conversion. of flanking hDNA is accompanied by the preferential con- Conversion Events on Symmetric hDNA. To examine the version to G234 genotype when G234 is on the donor strand pattern of gene conversion on symmetric hDNA, we exam- and the preferential conversion to + genotype when G234 is ined aberrant 4C:4W asci. The frequencies ofaberrant 4C:4W on the recipient one. This results in an excess of conversion, were determined in experiments I, II, VII, and VIII, and a establishing the donor genotype in the heteroduplex over sample of them was analyzed to detect double aberrant 4:4 Mendelian segregation, which is obtained when the recipient segregations at both flanking markers 17 and A4 (Table 5), genotype is restored (donor-directed conversion). The anal- unambiguously demonstrating symmetric hDNA flanking the ysis of SC:3W asci in the two reciprocal crosses mimics an middle marker (Fig. 2). The highest frequency of double individual cross in which hDNA is assumed to form on either aberrant 4:4 segregations was found in the absence of one or the other chromatid, the middle-marker being alter- middle-marker heterozygosities for EE and G234 (experi- natively in the donor or the recipient position. In cross to ments VIIb and VIIIb), while there was a decrease of double wild-type without flanking markers, G234 gives parity aberrant 4:4 segregation by a factor of2 in the presence ofthe (6+:2G = 2+:6G) (10). In such a cross, the opposite EOE1/+ and E2E1/+ heterozygosities (experiments I, II, directionality favors conversion to G234 and to + to an equal and VIIIa) and by a factor of 10 with E2EJ/G234 as middle extent when the hDNA is formed on the original wild-type or marker (experiment Vlla). This result is consistent with the G234 chromatid, respectively. previous observations that the EE/+ and G/+ heterozygosi- With EOEJ, no directionality is observed in the two ties reduce the aberrant 4:4 segregation of downstream reciprocal crosses, corresponding to the absence of direc- markers (including A4) by factors of about 2 and 10, respec- tionality in its gene-conversion pattern (12). In conclusion, tively (23, 25). We determined the middle-marker genotype the parity in gene conversion of EOEI and G234 are reached (EE or +) of 98 asci with a double aberrant 4:4 segregation by two completely different ways. from crosses Ia, lb, and Ila. In these three crosses, all As previously observed (17), with E2EJ the same -combinations of middle-marker genotype were found within directionality is observed in the two reciprocal crosses the two mixed pairs of an individual asci. Asci where one favoring the establishment of the E2EJ genotype in the spore pair is EE and the other is + correspond to a Mendelian heteroduplex (genotype-directed conversion). The same dis- Downloaded by guest on September 25, 2021 7390 Genetics: Harnza et al. Proc. Natl. Acad. Sci. USA 83 (1986)

parity is observed in the cross with wild type without flanking involves the cutting of only one strand (14). Alternatively, markers and is presumed to result from preferential destruc- one could envisage that the double-strand break (or gap) is tion ofthe wild-type information during mismatch correction. repaired along the sister chromatid rather than along the In crosses with the G234/E2EJ or G234/EOEJ heterozy- homolog. gosities, the donor-directed conversion process is epistatic to In conclusion, the data suggest that, in Ascobolus, there the genotype-directed process, which normally acts on E2EJ. are different pathways for the gene conversion of 1-bp point This explains why G234 appears to impose parity on the mutations and one large deletion. We favor the idea that this conversion of mutations such as E2EJ when the donor and difference reflects the existence of two mechanisms of recipient strands cannot be distinguished (12). mismatch correction. Each pathway can lead to directionality When double aberrant 4:4 segregations flanking the in gene conversion. The genotype-directed pathway (which E2EJ/+ heterozygosity were selected, preferential correc- acts on 1-bp mutations) usually favors one allele, irrespective tion towards E2EJ was observed. This suggests that the same of its donor or recipient origin, and leads to a disparity in the correction mechanism is responsible for directional conver- frequency ofthe two alternative conversion events. For some sion on asymmetric and symmetric heteroduplex DNA. With mutants (e.g., double mutant EOEJ), mismatch repair favors both EOEI and E2EJ, we found that the correction of the neither wild type nor mutant, so that there is no directionality homologous mismatches may occur in opposite directions, of conversion and parity is observed. In contrast, the donor- restoring a Mendelian segregation. Further studies are need- directed pathway leads to parity because it favors the allele ed for a quantitative estimation of the disparity of conversion brought by the donor duplex, irrespective of its genetic and the ratio of conversion versus restoration events on information. symmetric heteroduplexes. Current recombination models postulate two possible or- We thank Michelle Dahuron and Julienne Delaruelle for technical igins for gene conversion: the repair of double-strand breaks assistance. We also thank Andrew Murray for helpful comments on (12) or the formation of hDNA followed by mismatch cor- the manuscript. The work was supported by Centre National de la rection (11). The similarity of frequencies of flanking asym- Recherche Scientifique (LA86) and Universitd Paris-Sud grants. metric hDNA when G234 or EE mutations are used middle as 1. Stahl, F. W. (1979) Genetic Recombination: Thinking About It markers suggests that the formation of hDNA is the primary in Phage and Fungi (Freeman, San Francisco). process occurring in the conversion of G234 as it is in the 2. Whitehouse, H. L. K. (1982) Genetic Recombination: Under- conversion of the 1-bp addition or deletion. We stress that standing the Mechanisms (Wiley, New York). hDNA involving both 17 and A4 is also formed in the absence 3. Orr-Weaver, T. L. & Szostak, J. (1985) Microbiol. Rev. 49, of middle-marker heterozygosity. It strongly suggests that 33-58. G234 conversions are not initiation events. The formation of 4. Rossignol, J.-L., Paquette, N. & Nicolas, A. (1979) Cold hDNA spanning the G234 heterozygosity should result in a Spring Harbor Symp. Quant. Biol. 43, 1343-1356. looping out of the wild-type strand as a single-strand loop. 5. Rossignol, J.-L., Nicolas, A., Hamza, H. & Langin, L. (1984) Radding is made Cold Spring Harbor Symp. Quant. Biol. 49, 13-21. (13) proposed that the single-strand loop 6. Leblon, G. (1972) Mol. Gen. Genet. 115, 36-48. double-stranded by DNA replication and that this double- 7. Nicolas, A. (1979) Mol. Gen. Genet. 170, 129-138. strand loop may be either integrated into the chromosome or 8. Leblon, G. & Rossignol, J.-L. (1973) Mol. Gen. Genet. 122, eliminated, depending on which pair of crossing strands are 165-182. cut at the base of the loop. This model predicts an absence of 9. Leblon, G. (1979) Genetics 92, 1093-1106. directionality in the conversion process, in contradiction with 10. Hamza, H. (1985) Dissertation, Universit6 Paris-Sud, Orsay, our observation. We also can exclude the hypothesis that France. mismatch correction occurs via a single-strand excision in 11. Holliday, R. (1964) Genet. Res. 5, 282-304. which the looped strand is excised as frequently as its shorter 12. Szostak, J. W., Orr-Weaver, T. L., Rothstein, R. J. & Stahl, partner. Hastings proposed that F. W. (1983) Cell 33, 25-35. mismatch correction could 13. Radding, C. M. (1979) Cold Spring Harbor Symp. Quant. Biol. arise via a double-strand cut at the site of nonhomology (14). 43, 1315-1316. This will initiate a double-strand break repair event using the 14. Hastings, P. J. (1984) Cold Spring Harbor Symp. Quant. Biol. chromatid that was the donor of information in heteroduplex 49, 49-53. formation as a template. This will give rise to a conversion. 15. Paquette, N. & Rossignol, J.-L. (1978) Mol. Gen. Genet. 163, This hypothesis predicts a "donor-directed conversion" and 313-326. accounts for our observations. The extension of the double- 16. Rossignol, J.-L. & Haedens, V. (1978) Heredity 40, 405-425. strand break into a double-strand gap could then explain the 17. Hastings, P. J., Kalogeropoulos, A. & Rossignol, J.-L. (1980) observation that G234 imposes its own conversion pathway Curr. Genet. 2, 169-174. to EE mutations (10). The epistasis of the G conversion 18. Rossignol, J.-L. & Paquette, N. (1979) Proc. Natl. Acad. Sci. pathway upon the EE one implies that E mismatches are not USA 76, 2871-2875. corrected at first because, in this case, the G 19. Rizet, G., Rossignol, J.-L. & Lefort, C. (1969) C. R. Hebd. nearby heterol- Seances Acad. Sci. 269, 1427-1430. ogy could be cocorrected, and this would result in an 20. Leblon, G., Haedens, V., Kalogeropoulos, A., Paquette, N. & interaction between EE and G. The epistasis of G is explained Rossignol, J.-L. (1982) Genet. Res. 39, 121-138. if the loop triggers the double-strand cut before the cutting of 21. Paquette, N. (1978) Can. J. Genet. Cytol. 20, 9-17. the shortest strand at E sites can operate, or if the presence 22. Nicolas, A., Arnaise, S., Haedens, V. & Rossignol, J.-L. of the loop prevents the detection of E mismatches by the (1981) J. Gen. Microbiol. 125, 257-272. single-strand correction apparatus. 23. Hamza, H., Haedens, V., Mekki-Berrada, A. & Rossignol, The directionality of conversion of G234 is not total. J.-L. (1981) Proc. Natl. Acad. Sci. USA 78, 7648-7651. Between 28% and 11% of the events give a restoration of the 24. Rossignol, J.-L. & Haedens, V. (1980) Curr. Genet. 1, recipient genotype, leading to a Mendelian segregation. 185-191. Hastings proposed that a portion of the mismatch corrections 25. Nicolas, A. & Rossignol, J.-L. (1983) EMBO J. 2, 2265-2270. Downloaded by guest on September 25, 2021