Cytoplasmic suppression of Ogura cytoplasmic male sterility in European natural populations of Raphanus raphanistrum

Theor Appl Genet (2007) 114:1333–1343

ORIGINAL PAPER

Cytoplasmic suppression of Ogura cytoplasmic male sterility in European natural populations of Raphanus raphanistrum

Sandra Giancola · Yong Rao · Sophie Chaillou · Sophie Hiard · Alfred Martin-Canadell · Georges Pelletier · Françoise Budar

Received: 1 December 2006 / Accepted: 2 February 2007 / Published online: 23 February 2007© Springer-Verlag 2007

Abstract The Ogura cytoplasmic male sterility(CMS) of radish has been used for hybrid seed produc-tion in radish and Brassica crops. It is the only CMSsystem occurring in wild populations for which thegene responsible for sterility and a restorer gene havebeen formally identiWed. In Japan, gynodioecious pop-ulations of radish carrying Ogura or an Ogura-relatedcytoplasm have been described. The occurrence ofrestorer genes for the Ogura CMS in wild radish(Raphanus raphanistrum) in France led us to search forthe corresponding male sterility gene (orf138) in sev-eral natural populations in France, England and Leba-non. We detected the orf138 gene, by PCR, at lowfrequency, in three populations from France and onefrom Southern England. Further molecular character-ization showed that these plants carried a cytoplasm

closely related to the original Ogura cytoplasm, with avariant orf138 coding sequence, previously reported tobe ancestral. We performed crosses with sterile andmaintainer radish lines, to test the ability of this wildOgura-related cytoplasm to induce sterility. Surpris-ingly, the European Ogura-related cytoplasm did notcause sterility. Northern blots and circular RT-PCRanalyses showed that orf138 gene expression wasimpaired in these plants because of a novel cytoplasm-dependent transcript-processing site.

Introduction

Cytoplasmic male sterility (CMS) systems determinethe reproductive biology of plants through the interac-tion of a male sterility-inducing cytoplasm and nuclearrestorer genes. They have been widely used for theproduction of hybrid crop seeds (Havey 2004). CMSsystems have attracted considerable attention in twodiVerent scientiWc communities. Population geneticistshave studied CMS in natural populations, in which it isexpressed as gynodioecy. Gynodioecy is deWned as theco-occurrence of hermaphrodites and females in a pop-ulation or species (Darwin 1877). The studied cases ofgynodioecy result from naturally occurring CMS sys-tems. Population geneticists have studied principallythe maintenance of genetic and phenotypic polymor-phisms, and their mode of evolution (Charlesworth2002; Bailey et al. 2003; Saur Jacobs and Wade 2003).According to the most widely accepted models, nuclearrestorer genes are selected in the presence of the steril-ity-inducing cytoplasm because, by allowing the pro-duction of pollen, they improve their own transmissionto the progeny. Conversely, if the sterility-inducing

Communicated by C. F. Quiros.

Sandra Giancola and Yong Rao contributed equally to this work.

Electronic supplementary material The online version of this article (doi:10.1007/s00122-007-0520-6) contains supplementary material, which is available to authorized users.

S. Giancola · Y. Rao · S. Chaillou · S. Hiard · A. Martin-Canadell · G. Pelletier · F. Budar (&)Station de Génétique et d’Amélioration des Plantes, Institut Jean-Pierre Bourgin, INRA UR254, Versailles, Francee-mail: [email protected]

Present Address:Y. RaoInstitute of Oil Crops, Guizhou Academy of Agricultural Sciences, Xiaohe District, Guiyang, Guizhou Province, People’s Republic of China

123

1334 Theor Appl Genet (2007) 114:1333–1343

cytoplasm is not present, they are obviously notselected, and may even be disadvantaged. This situa-tion, called “cost of restoration”, was predicted byrecent models (Bailey et al. 2003) and leads to theselection against restorers in the absence of the steril-ity-inducing cytoplasm. Most, if not all, of the studiedexamples of gynodioecy in natural populations haveinvolved CMS systems with unknown genetic determi-nants. Our understanding of the molecular evolution ofthese systems therefore remains largely theoretical.Molecular geneticists have tried to identify the genesinvolved, mainly in crops or crop-related species, inwhich such systems can be exploited for hybrid seedproduction. A number of mitochondrial genes havebeen correlated with, or formally implicated in malesterility (for reviews, Hanson and Bentolila 2004;Budar et al. 2006). Nuclear genes for the restoration offertility (Rf) have been identiWed (Cui et al. 1996; Ben-tolila et al. 2002). All except the maize Texas-CMSrestorer Rf2 encode proteins of the PPR family,involved in the posttranscriptional control of mito-chondrial and chloroplast gene expression (Lurin et al.2004). The assumed molecular function of PPR pro-teins is consistent with the general observation that fer-tility is restored by impairing sterility gene expressionthrough posttranscriptional mechanisms aVecting ste-rility gene mRNA maturation, editing, or accumulation(Wang et al. 2006). The structure of the Wrst identiWedRf loci in Petunia, radish and rice led to comparisonswith disease resistance loci, as selection for Rf locioccurs in response to the establishment of a sterility-inducing cytoplasm in a population (Touzet and Budar2004).

The Ogura CMS system is the only naturally occur-ring CMS system in which the genes responsible forsterility and restoration have been formally identiWed.The Ogura CMS, originally identiWed in a Japaneseradish cultivar (Ogura 1968), was introduced into Bras-sica crops for hybrid seed production (Budar et al.2004). The mitochondrial gene responsible for Oguramale sterility, orf138, was shown to be constitutivelyexpressed in sterile plants (Bonhomme et al. 1991,1992; Krishnasamy and MakaroV 1993; Grelon et al.1994). In the original Ogura cytoplasm and in malesterile Brassica cybrids, the orf138 gene is expressed asa bicistronic transcript, together with the orfB gene(Bonhomme et al. 1992; Krishnasamy and MakaroV1993). We have shown, in Brassica cybrids, that thecoexpression of orf138 with orfB is involved in stabil-ization of the bicistronic transcript and expression ofthe sterility trait (Bellaoui et al. 1997). The OguraCMS has been detected in wild and cultivated AsianRaphanus genotypes (Yamagishi and Terachi 1994a, b,

1996, 1997). It is present in natural gynodioecious pop-ulations of wild radish in Japan (Murayama et al.2004). An analysis of the orf138 gene sequence of vari-ous cytoplasms from Asian wild and cultivated radishesled to the identiWcation of several intragenic variationsin the coding sequence of the gene, making it possibleto infer phylogenic relationships (Yamagishi and Ter-achi 2001). A nuclear locus restoring fertility for OguraCMS was recently identiWed and described in radish(Brown et al. 2003; Desloire et al. 2003; Koizuka et al.2003).

Restorers for the Ogura CMS were discovered innatural French populations of Raphanus raphanistrumafter interspeciWc crosses between Brassica napusmale-sterile plants and wild radish (Eber et al. 1994).We report here the identiWcation of a wild radish cyto-plasm carrying the orf138 gene in European popula-tions that also carry restorer genes for the Ogura CMS.We show that this cytoplasm cannot induce sterility,despite its strong similarity to the original Ogura cyto-plasm. We also present an analysis of the expression ofthe sterility gene in this newly identiWed cytoplasm.Our results demonstrate the existence of an unprece-dented and unexpected situation within a CMS system.

Materials and methods

Plant materials

The cultivated radish (Raphanus sativus) maintainergenotype used as the reference genotype was the “L7”line created by Bonnet (1977). It is available as fertileplants with normal radish cytoplasm (L7F plants) or asmale sterile plants with the Ogura cytoplasm (L7Splants). Plants were grown and crosses were performedin the greenhouse.

Wild radish (Raphanus raphanistrum) plants werecollected from natural populations (see Table 1), eitheras leaf samples, or as seeds of maternal descent. Theleaf samples were dried in an oven (50–60°C) andstored at room temperature before genomic DNAextraction. Seeds were sown in the greenhouse andfresh leaf material was collected from one individualper mother plant for genomic DNA extraction.

Isolation of nucleic acids

Genomic DNA for PCR analysis was extracted in 96-wellplates, using a modiWed version of a previouslydescribed procedure (Loudet et al. 2002). Fresh oroven-dried tissues and a 4 mm glass bead were placed intubes held in a 96-well polypropylene storage plate. The

123

Theor Appl Genet (2007) 114:1333–1343 1335

tissues were frozen in liquid nitrogen and ground withan MM300 vibrator (Retsch GmbH). Extraction buVer(200 mM tris–HCl pH7.5, 250 mM NaCl, 25 mMEDTA, 0.5% SDS, 1.7 mg/ml proteinase K) was added(200 �l per tube) to the ground tissue and the mixturewas incubated for 20 min at room temperature. Aftercentrifugation for 15 min at 3,000g, 100 �l of superna-tant was added to 100 �l of isopropanol in one of thewells of a 96-well plate. The plate was incubated for10 min at room temperature, the DNA was pelleted bycentrifugation for 15 min at 3,000g, and the supernatantwas discarded. The pellet was air-dried and the DNAresuspended in 50–100 �l of water. Typically, we used2 �l of the DNA solution for PCR. The quantity ofDNA solution used was adjusted when the DNA yieldwas insuYcient (especially for oven-dried samples).

Genomic DNA was extracted for DNA hybridiza-tion analysis as previously described (Dellaporta et al.1983), using fresh samples harvested in the greenhouse.It was not possible to obtain genomic DNA suitable forhybridization analysis from dried leaf samples.

Total RNA was extracted by grinding fresh samplesharvested in the greenhouse in liquid nitrogen andusing Trizol reagent (Invitrogen) according to the man-ufacturer’s instructions. It was not possible to obtainRNA from dried leaf samples.

AmpliWcation analyses

Most ampliWcation reactions were performed in 25 �lof reaction mixture containing 75 mM tris–HCl pH 8.8,20 mM (NH4)2SO4, 0.01% tween, 2.5 mM MgSO4,0.3 mM of each dNTP, 0.1 �M of each primer, 1–2 units

of Taq DNA polymerase prepared in the laboratory.Typical cycling reactions consisted of 1–5 min at 95°Cfollowed by 25–30 cycles [30 s at 94°C, 30 s at annealingtemperature (see Table 2), 1–3 min at 72°C (dependingon the length of expected ampliWcation product)] and aWnal 10 min at 72°C. AmpliWcation products were ana-lyzed by 1% agarose gel electrophoresis.

Sequencing analyses

One region of the mitochondrial genome, correspond-ing to the orf138 locus, and two regions of the plastidgenome, corresponding to the intron of the K-tRNAgene and the intergenic region between the L- and F-tRNA genes (Grivet et al. 2001), were sequenced inseveral individuals (see “Results”). All sequencing wascarried out by Genoscreen. The orf138 locus wasdirectly sequenced from the product ampliWed withprimers orf138-F2 and orfB-R1 (Table 2), using thesame primers. The plastid genome regions were ampli-Wed with the primers K1-M13F and MatK1-M13R, andwith the primers trnL-M13F and trnF-M13R. AmpliW-cation products were then directly sequenced withM13F and M13R primers (Table 2).

RNA and DNA hybridization analyses

Total DNA was digested with the chosen restrictionenzymes, in the buVer recommended by the enzymesupplier (Fermentas), supplemented with 4 mM spermi-dine. The digestion products were subjected to electro-phoresis in 0.6 or 0.8% (depending on the size of theexpected hybrizing fragments, see Wgure legend) agarosegels in TBE buVer (Ausubel et al. 1990). The bandswere then transferred to a nylon membrane (Gene-screen), according to the manufacturer’s instructions.

Total RNA was loaded onto a 1.2% agarose gel con-taining 8% formaldehyde in 0.5£ MOPS buVer (Ausu-bel et al. 1990). Electrophoresis was conducted in1£ MOPS buVer at 50V overnight (ca. 18 h). Thebands were then transferred to a nylon membrane(Genescreen), according to the manufacturer’s instruc-tions.

The fragments used as probes were obtained byampliWcation from plant total DNA or from clonedfragments of mitochondrial DNA. Their positions onthe orf138 locus are shown in Fig. 1 (hatched boxes).Probe A (orf138) was obtained by PCR with primersorf138-F1 and orf138-R on sterile plant DNA; probe B(orfB) was obtained by PCR with primers orfB-F andorfB-R2 on plant DNA; probe C (fMtRNA) wasobtained by PCR with primers TRNAFM-F andTRNAFM-R on plant DNA; probe D (5� region of

Table 1 Names and locations of the natural populations studied

Population name

Location (country) Materials collected

POP2 Rennes (France) SeedsPOP9 Boullay-les-Troux (France) SeedsPOP12A Poilly-lez-Gien (France) SeedsPOP13 Saint-Martin/Ocre (France) SeedsPOP14 Pont-à-Marcq (France) SeedsPOPN199 Montreuil-sur-mer (France) SeedsAZB Azay-le-Rideau (France) SeedsCR Le Croisic (France) SeedsSG Saint Gildas (France) LeavesPOPE Etacq, Jersey (Channel Islands) SeedsCHOA Chausey (Channel Islands) LeavesRB Rocquaine Bay, Guernsey

(Channel Islands)Seeds

LIZ Lizard Point (England) LeavesLBA Bater ech Chouf (Lebanon) LeavesLBB Marj el zhour (Lebanon) LeavesLBC Dahr el Souan (Lebanon) LeavesLBD Qartaba (Lebanon) Leaves

123

1336 Theor Appl Genet (2007) 114:1333–1343

orf138, 390 bp) was obtained by PCR with primers UFand UR (“universal” primers for pBluescript) on aplasmid derived from pBluescript carrying the Wrst 57codons of orf138 (Duroc et al. 2005). When necessary,the ampliWcation products were puriWed after agaroseelectrophoresis, before labeling with a random primingkit (Promega) and �32P dCTP, according to the manu-facturer’s instructions.

Hybridizations were carried out as previouslydescribed (Ausubel et al. 1990).

Circular RT-PCR (cRT-PCR) analyses (Kuhn and Binder 2002)

Circular RT-PCR was performed as described else-where (Perrin et al. 2004), using the primers describedin Table 2. Their positions on the orf138 locus aregiven in Figs. 1 and 4. Cycling reactions were carriedout as follows: 5 min at 95°C, followed by 10 cycles(30 s at 95°C, 45 s at 62°C minus 1°C each cycle,1.5 min at 72°C), 25 cycles (30 s at 95°C, 45 s at 52°C,1.5 min at 72°C), and a Wnal 10 min at 72°C. Ampliconswere inserted into pTOPO (Invitrogen), and the sizesof the inserts were estimated by PCR ampliWcation,using UF and UR primers and agarose gel electropho-resis. Genome Express sequenced certain selectedclones, using the UF and UR primers.

Results

The orf138 gene is present at low frequency in non-gynodioecious European wild populations of Raphanus raphanistrum

The known occurrence of fertility restorers for OguraCMS in natural populations of Raphanus raphanistrum(wild radish) in France (Eber et al. 1994) led us tosearch for the Ogura CMS mitochondrial gene in

Table 2 Primers used in this study

Primer 5�–3�sequence Annealing temperature for PCR(°C)

Amplicon size(bp)

cobU TCTTCTCTCGGGGTCATCCT 53 700cobL CCCCCTTCAACATCTCTCATorf138-F1 GCATCACTCTCCCTGTCGTTATCG 512orf138-R ATTATTTTCTCGGTCCATTTTCCA

orf138-F2 GAAACGGGAAGTGACAATAC 51 788orfB-R1 GTACTCCATCTCCATCATTGC

cDNA-priming TGGGGTCCTTGCTCTGGATGGTCTcRT-F GCTCTAGAGACTTATTGGGAAAAAGGAGG 52 variablecRT-R GCATTATTTTCTCGGTCCAT

orfB-F TCAACAACCAACCACAACTTT 52°C 520orfB-R2 TACAAGTGATCCACCTTCCAG

TRNAFM-F ACGTGTAGCCCTGTATGGACT 54 398TRNAFM-R GGTATTGTCACTTCCCGTTTC

UF GTAAAACGACGGCCAGT 53 variableUR GGAAACAGCTATGACCATG

K1-M3F CACGACGTTGTAAAACGACGTTGCCCGGGATTCGAA 50 706MatK1-M13R GGATAACAATTTCACACAGGATTAGGGCATCCCATTAGTA

trnL-M13F CACGACGTTGTAAAACGACGGTTCAAGTCCCTCTATCCC 52 450trnF-M13R GGATAACAATTTCACACAGGATTTGAACTGGTGACACGAG

M13F CACGACGTTGTAAAACGACM13R GGATAACAATTTCACACAGG

Fig. 1 The orf138 locus. The orf138 locus is presented as de-scribed for the original Ogura cytoplasm (Bonhomme et al. 1992;Krishnasamy and MakaroV 1993; Bellaoui et al. 1997). The boxeson the horizontal line represent the coding sequences of genes:white box, gene for the formylmethionine tRNA (tRNAfM);black box, orf138; gray box, orfB (or atp8). Arrowheads below theline indicate the positions of the oligonucleotide primers used inthis study (see also Table 2): black arrowheads, primers for PCRdetection of the orf138 gene in wild plants (orf138-F1 & orf138-R); white arrowheads primers for ampliWcation and sequencing oforf138 in wild plants (orf138-F2 & orfB-R1); gray arrowheadsprimers for cRT-PCR, the internal primer was used for cDNApriming (cDNA priming), the external primers for ampliWcation(cRT-F & cRT-R). The hatched boxes above indicate the regionsused as probes in the DNA and RNA hybridization experimentsshown in Figs. 2 and 3

123

Theor Appl Genet (2007) 114:1333–1343 1337

populations of wild radish from Europe and the MiddleEast. We screened a total of 17 populations of R. raph-anistrum, from various locations in France, the ChannelIslands, England, and Lebanon (Table 1). None of thepopulations studied was considered gynodioecious onthe basis of in situ observations. We ampliWed the mito-chondrial cob gene and orf138 ampliWcation in a singlereaction, to distinguish between failed ampliWcation dueto poor quality DNA and failed ampliWcation due to theabsence of the orf138 gene sequence. Only samples giv-ing a cob ampliWcation product and no orf138 ampliWca-tion product were considered negative for the orf138gene. The orf138 gene was detected in three popula-tions from France and one population from southernEngland, at a low frequency (see Table 3). We investi-gated whether orf138 was associated with orfB in thesecytoplasms, as in the Ogura cytoplasm, by ampliWcationwith the orf138-F2 and orfB-R1 primers (see Fig. 1). Inall cases, the orf138 gene present in these Europeanwild radishes was associated with orfB, as in the Oguracytoplasm. The 788 bp region of the orf138 locusbetween the orf138-F2 and orfB-R1 primers wassequenced for two individuals in each population (withthe exception of the LIZ population, in which only onepositive individual was detected, and sequenced). Allthe sequences were identical and corresponded to theB-type sequence identiWed as the probable ancestralsequence for the orf138 gene: they carried the previ-ously described silent A99G substitution in the orf138coding region (Yamagishi and Terachi 2001). The cyto-plasm detected in European wild radishes is thereforenot identical to the Ogura cytoplasm and is hereafterreferred to as the Ogura-related cytoplasm.

The Ogura-related cytoplasm of European wild radish does not induce male sterility

The B-type sequence of orf138 described by Yamagi-shi and Terachi (2001) induces male sterility in Japa-nese wild radishes (Yamagishi and Terachi 1994a,1997). At least one of the European populations withindividuals carrying the Ogura-related cytoplasm con-tains restorers of fertility (Eber et al. 1994). Thus, pro-vided that fertility restorers are segregated out, wewould expect to observe male sterility caused by theOgura-related cytoplasm in the progeny of wildmother plants. We crossed individuals carrying theOgura-related cytoplasm from the initially prospectedpopulations (POP2 and POPN199) with radish testergenotypes. The same plants (grown in the greenhousefrom seeds harvested from natural populations, mater-nal descent) were used in both crosses. We Wrst usedthe wild plants as male parents for the pollination of

male-sterile radish plants (L7S). In the progenies ofthese crosses, we obtained male fertile (restored) andmale sterile plants. The small numbers of seeds col-lected from the crosses precluded a precise geneticanalysis of restoration. However, among the crossesperformed with 13 diVerent wild plants used as maleparents, and which gave between one and nine seeds,Wve gave at least one sterile progeny, showing that res-toration was not Wxed in the population considered.We also used the wild plants as female parents andpollinated their emasculated Xowers with pollen fromthe maintainer radish line (L7F). Based on the resultsobtained for the Wrst cross, we sowed only seedsobtained from plants giving male sterile progeny whenused as paternal parents—i.e., those not homozygousfor fertility restorers. Similar numbers of seeds wereobtained from both crosses (between 3 and 11 seeds).No male sterile plant was observed in the maternalprogeny of wild plants, in contrast to what would beexpected if these plants carried a sterility-inducingcytoplasm. The cross was repeated with one of thematernal plants and 30 new progenies were all fertile.Several plants from three backcross progenies werebackcrossed again with the radish maintainer line

Table 3 Detection of orf138 in natural populations of Raphanusraphanistrum

a Only samples giving ampliWcation with the cob primer pair wereconsideredb Individuals giving positive ampliWcation with the orf138 primerpairc Seeds were harvested in bulk from the natural population thisparticular year, therefore, diVerent individuals tested may haveoriginated from the same female parent

Population Year of sampling

Number of tested individualsa

Number of individuals with orf138b

POP2 1998 223c 151999 83 02001 150 2

POP9 1999 66 0POP12A 1999 43 0POP13 1999 70 0POP14 1999 62 0POPN199 1999 55 8AZB 2000 15 0CR 2004 24 0SG 2004 23 2

2005 17 1POPE 2000 24 0CHOA 2004 46 0RB 2000 12 0LIZ 2003 56 1LBA 2005 42 0LBB 2005 36 0LBC 2005 40 0LBD 2005 30 0

123

1338 Theor Appl Genet (2007) 114:1333–1343

(L7F) and gave between 10 and 50 BC2 progenies.Once again, no male sterile plant was obtained, dem-onstrating that despite the apparently normal malesterility gene sequence, the cytoplasm of wild plantscould not induce male sterility.

The European Ogura-related cytoplasm is very similar to the Ogura cytoplasm

The inability of the Ogura-related cytoplasm to causesterility may be linked to a rearrangement of mito-chondrial DNA at the orf138 locus undetectable byPCR. We therefore carried out DNA hybridizationanalyses on several plants from the POP2 andPOPN199 populations, using probes detecting theorf138 coding sequence, the orfB gene, and the fMet-tRNA gene. Wild plants carrying the Ogura-relatedcytoplasm gave similar hybridizing fragments to theOgura cytoplasm (Bonhomme et al. 1991, 1999). Theorf138 probe detected the Ogura-speciWc NcoI 2.5 kband BglI 9 kb fragments. The orfB probe also detectedonly these fragments, demonstrating that the only orfBgene present in the Ogura-related cytoplasm is linked

to orf138. The fMet-tRNA probe detected the samefragments as the orf138 probe plus another fragment(approximately 18.5 kb for the BglI digestion), show-ing that this gene is duplicated in the Ogura-relatedcytoplasm, as in the original Ogura cytoplasm (Krish-nasamy and MakaroV 1993). Furthermore, none of thewild plants testing negative for orf138 in speciWc PCRassays had hybridization fragments in common withthe Ogura or Ogura-related plants. Moreover, theirhybridization proWles diVered (see for example sam-ples 4 and 5 in Fig. 2b, and samples 8 and 9 in Fig. 2c),showing that the populations studied contained variouscytoplasms lacking orf138. We also sequenced twosmall regions of the plastid genome (see “Materialsand methods”) from several individuals of each pros-pected population. For both regions, the sequencesobtained from wild plants carrying the Ogura-relatedcytoplasm were identical to those from plants with theOgura cytoplasm, and diVerent from those of wildplants lacking the orf138 gene (see Electronicsupplementary data). These analyses conWrm thatdiVerent cytoplasms lacking orf138 coexist in the wildpopulations.

Fig. 2 DNA hybridization analysis of wild plants. Two successivehybridizations of the same membrane are shown in a and b. Thesame is true for c and d. The loaded samples are indicated foreach gel in the boxes on the right. POPX#n means “plant numbern in population POPX”. L7S is sterile and carries the Ogura cyto-plasm, L7F is the fertile maintainer (see “Materials and meth-ods”). POPX#n xL7F means that the DNA was extracted fromthe maternal progeny of the wild plant, obtained by pollinationwith L7F. The signs in the column furthest to the right indicatewhether the considered plant did (+) or did not (¡) generate an

ampliWcation product in the orf138-speciWc PCR. a, b: NcoI andBglI digestions of total DNA were separated by electrophoresisin a 0.8% agarose gel and blotting onto a membrane. The mem-brane was (a) hybridized with an orf138 probe (hatched box A inFig. 1), stripped and (b) reprobed with an orfB probe (hatchedbox B in Fig. 1). c, d BglI digestions of total DNA were separatedby electrophoresis in a 0.6% agarose gel and blotting onto a mem-brane. The membrane was (c) hybridized with a tRNAfM probe(hatched box C in Fig. 1), stripped, and (d) reprobed with anorf138 probe (hatched box A in Fig. 1)

123

Theor Appl Genet (2007) 114:1333–1343 1339

The expression of orf138 is impaired in the European Ogura-related wild cytoplasm

We analyzed orf138 expression in the wild Ogura-related cytoplasm, carrying out RNA hybridizationanalyses to determine the reasons for the lack of steril-ity in maternal descent from wild mother plants. Theseanalyses were performed on plants with the EuropeanOgura-related cytoplasm, which are fertile, and sterileplants with the Ogura cytoplasm, using orf138 and orfBprobes. The orf138-orfB cotranscript accumulated asthe major RNA in plants with the Ogura cytoplasm(Fig. 3), as previously shown (Bonhomme et al. 1992).However, this cotranscript was present in extremelysmall amounts in plants with the wild Ogura-relatedcytoplasm, despite these plants being produced by one(M1, M3) or two (M2, M4, M5) back-crosses with theL7F maintainer radish line. A shorter RNA, not

detected with the probe covering the 5� part of orf138,accumulated as the major orfB RNA in these plants.As this cytoplasm contains no other orfB gene (Fig. 2)the orfB mRNA must originate from the orf138-orfBlocus.

We deWned the orfB mRNA ends by circularRT-PCR (cRT-PCR) analysis (Kuhn and Binder 2002)in several male fertile plants with the wild Ogura-related cytoplasm (Fig. 4). BrieXy, RNA molecules arecircularized by T4 RNA ligase before Wrst strandcDNA synthesis followed by PCR with primersdirected outwards the gene, thus amplifying in one stepboth 5� and 3� ends of the RNA. As a control, the sameexperiment was performed on male sterile plants withthe Ogura cytoplasm. The sizes of the major cRT-PCRproducts obtained were consistent with the major orfBmRNA detected in RNA hybridization analysis(Fig. 4a). After cloning, however, inserts of diVerent

Fig. 3 RNA hybridization analysis. a, b The same membrane washybridized (a) with a probe speciWc for the 5� part of the orf138gene (hatched box D in Fig. 1), stripped and then reprobed (b)with a probe speciWc for the orfB gene (hatched box B in Fig. 1).Figures on the left indicate the positions of size markers given inkb. c Description of the samples. The plants are designated as inFig. 2. (POPX#nxL7F)#mxL7F is the progeny of the second back-cross with L7F, performed on plant #m of the Wrst backcross ofplant #n from POPX by L7F. P1 and M1 were obtained from thesame wild plant, crossed either as a male parent with the L7Splant, giving P1, or as female parent with the L7F plant, givingM1. P1 therefore has the Ogura cytoplasm and is male sterile,whereas M1 has the wild Ogura-related cytoplasm and is malefertile. The asterisk indicates the position of the major RNA inplants carrying the original Ogura cytoplasm (O, P1); the blackdot indicates the position of the major RNA in plants with thewild Ogura-related cytoplasm

Fig. 4 cRT-PCR analysis. a Typical cRT-PCR result. AmpliWca-tion products, obtained as described in the “Materials and meth-ods”, were separated by electrophoresis in a 1% agarose gel. wproducts obtained with RNA from a plant with the wild Ogura-related cytoplasm (in this case POPN199#23xL7F). O productsobtained with RNA from a plant with the Ogura cytoplasm (inthis case L7Sx POPN199#23). b Representation of the orf138-orfB locus (see also legend of Fig. 1). The horizontal arrowsrepresent the RNA molecules mapped in this study. The longestarrow (asterisk) represents the orf138-orfB cotranscript present insterile plants. The shorter arrow (dot) represents the major orfBRNA of plants with the Ogura-related cytoplasm. Arrowheadsshow the position of primers used in cRT-PCR, the internal prim-er was used for cDNA priming (cDNA priming), the externalprimers for ampliWcation (cRT-F & cRT-R). c Precise mapping ofthe ends of the new orfB RNA (dot in Figs. 3, 4b). The results ofthe sequencing of cRT-PCR products are summarized: the verti-cal bars above the sequence indicate the positions of the 5� endsof RNA. The lengths of the bars are proportional to the numberof clones with the considered 5� end, given above the bars. TheXbaI site corresponding to the end of probe D (see Fig. 1) used inRNA hybridization analysis (Fig. 3) is shown in bold

123

1340 Theor Appl Genet (2007) 114:1333–1343

sizes were obtained, as estimated by ampliWcation, andthe clones could be grouped into three size classes.Class I corresponded to the major cRT-PCR product inplants with the Ogura cytoplasm. Class II corre-sponded to the major cRT-PCR product in plants withthe wild Ogura-related cytoplasm. Class III clones hadinserts shorter than those of classes I and II. Weobtained the sequences of Wve clones for class I, 15clones for class II, and 30 clones for class III. In allthree classes, the 3� end of the RNA corresponded tothe 3� end of the previously mapped orf138-orfBcotranscript (Bonhomme et al. 1992; Bellaoui et al.1997). For classes I and II, the 5� ends of the RNAsdetermined from the diVerent clones were mappedwithin a few bases of the DNA sequence. Class I corre-sponded to full-length orf138-orfB cotranscripts (aster-isk), as previously mapped (Bonhomme et al. 1992;Bellaoui et al. 1997). Class II corresponded to RNAswith a 5� end in the middle of the orf138 codingsequence (Fig. 4b), consistent with the length of themajor orfB RNA in plants with the Ogura-related cyto-plasm (dot). For the third class, corresponding to thesmallest inserts, the 5� ends of the RNAs were scat-tered in the second half of the orf138 coding sequence,with no obvious major position.

Western blots of protein extracts from plants carry-ing the wild Ogura-related cytoplasm with an antibodyraised against the ORF138 protein (Grelon et al. 1994)conWrmed that these plants produced no detectableORF138 protein (data not shown).

Discussion

An Ogura-related cytoplasm of ancient origin is present in European populations of R. raphanistrum

According to theoretical models, the presence of a ste-rility-inducing cytoplasm leads to the selection ofrestorers of fertility. The discovery of restorers for theOgura CMS in natural French populations of Raph-anus raphanistrum (Eber et al. 1994), naturally led tosearches for the Ogura CMS in these populations.However, we found no female plants in natural popula-tions of R. raphanistrum, including the population inwhich these restorers were Wrst detected. Fortunately,the mitochondrial gene responsible for the Ogura CMSis known and can be readily detected by PCR. Wesearched for the orf138 gene in wild plants or theirmaternal descent and identiWed a few plants carryingthis gene. Four of the 17 populations studied includedindividuals with the orf138 gene. As the populations

sampled were not evenly distributed geographically, itis diYcult to determine the distribution of the Ogura-related cytoplasm. However, it is clear that, when pres-ent, this cytoplasm is not frequent. In contrast, Muray-ama et al. (2004) reported a high frequency of theorf138 gene in populations of wild radish in Japan,most of which were gynodioecious. The situation inEuropean populations of R. raphanistrum thereforeclearly diVers from that of Japanese populations ofwild radish, at least as far as the Ogura CMS is con-cerned. The low frequency of orf138 in our populationssuggests that this cytoplasm has either only recentlybeen introduced or is an ancient cytoplasm destined tobecome extinct. Brassica hybrids with the Ogu-INRAcytoplasm, derived from the Ogura cytoplasm, havebeen cultivated in France for a decade. This might sug-gest acquisition of the Brassica cytoplasm of hybrids bywild Raphanus raphanistrum through spontaneousinterspeciWc crosses. However, this possibility can beruled out because the Ogu-INRA cytoplasm has char-acteristics absent from the wild Ogura-related cyto-plasm. Firstly, plants with the Ogura cytoplasm have asecond copy of the tRNAfMet gene; plants with theOgu-INRA cytoplasm do not (Bonhomme et al. 1999).Secondly, the Ogu-INRA cytoplasm includes a plastidgenome from Brassica species (Pelletier et al. 1983; Bon-homme et al. 1999), whereas the plastid sequences foundin the Ogura-related cytoplasm were identical to thoseof the original radish Ogura cytoplasm. Thirdly, theintergenic sequence between orf138 and orfB in the rad-ish Ogura and Ogura-related cytoplasms diVers slightlyfrom that of the Ogu-INRA cytoplasm. In addition, theorf138 coding sequence of wild R. raphanistrum plantsdiVers at position 99 (silent polymorphism) from theOgura sequence present in Brassica hybrids. Interest-ingly, the sequence of the orf138 gene present in thenovel Ogura-related cytoplasm described here is identicalto the type B sequence in wild and cultivated radishes(Yamagishi and Terachi 2001). This sequence was mostfrequent between the wild radish and R. raphanistrumplants studied by these authors, suggesting that type B isthe ancestral sequence for orf138. These data, and thepresence of fertility restorers in European populationsof R. raphanistrum, are consistent with the Ogura-related cytoplasm in these populations being ancientrather than recently introduced.

The European Ogura-related cytoplasm is very similar to the original Ogura cytoplasm, but does not cause sterility

Terachi et al. (2001) described three types of cytoplasmin wild and cultivated radishes that diVered in terms of

123

Theor Appl Genet (2007) 114:1333–1343 1341

their orfB 5� Xanking sequences. One of these sequenceswas found to be strictly linked to orf138 and found inOgura and Ogura-related cytoplasms. These authorssuggested that the orf138 and orfB genes became associ-ated only once in the history of radish cytoplasms. How-ever, this assertion was based on observation of theorf138-orfB locus only, and the sequencing resultsobtained were not completely consistent with those ofother studies (see the discussion in Terachi et al. 2001).Our analysis of the orf138-orfB locus in EuropeanOgura-related cytoplasm yielded a sequence identical tothat reported by Terachi et al. (2001). We investigatedplastid genome polymorphism in the prospected Euro-pean populations, and clearly identiWed a sequence spe-ciWc for the Ogura-related cytoplasm for both plastidregions studied (ESM). This sequence was identical tothat of the original Ogura cytoplasm, taken as a refer-ence, indicating that the orf138-orfB locus probablydoes have a monophyletic origin. In addition, our DNAhybridization analysis of this locus showed no diVerencebetween the original Ogura cytoplasm and the Euro-pean Ogura-related cytoplasm (Fig. 2).

We showed, by crosses with tester genotypes, thatfertility restorers were not Wxed in the prospected natu-ral populations, as some individuals were not homozy-gous for restorers. However, the small numbers of seedsobtained from these crosses precluded a genetic analy-sis of restoration. In addition, we did not include in thisstudy crosses involving plants with cytoplasms diVerentfrom the Ogura-related cytoplasm. Further geneticstudies are required to determine the number of resto-ration loci present in French natural populations of R.raphanistrum, the frequency of restorer alleles, andtheir possible linkage to the identiWed Rfo locus (Brownet al. 2003; Desloire et al. 2003; Koizuka et al. 2003).Nevertheless, two backcrosses with the maintainer rad-ish tester line failed to reveal male sterility in the mater-nal progenies of plants heterozygous for fertilityrestorers. This unexpected result shows that the newlyidentiWed Ogura-related cytoplasm does not induce ste-rility. As the type B orf138 coding sequence is associ-ated with male sterility-inducing cytoplasms in Japanesewild radishes (Yamagishi and Terachi 1997), we con-cluded that this silent polymorphism is not the cause ofthe lack of sterility phenotype. We therefore analyzedorf138 expression in male fertile plants with the Euro-pean Ogura-related cytoplasm.

The orf138 transcript is interrupted by a processing event favored in the wild Ogura-related cytoplasm

RNA hybridization analyses clearly showed that theexpression proWle of the orf138-orfB locus diVered

between the two cytoplasms (Fig. 3). Whereas theorf138-orfB cotranscript (Bonhomme et al. 1992; Bel-laoui et al. 1997) accumulates in Ogura plants, asmaller RNA was the major orfB RNA detected inplants with the Ogura-related cytoplasm. This RNAwas not detected with a probe spanning the 5� part ofthe orf138 sequence, indicating that this RNA does notcontain the entire orf138 coding sequence. Mapping ofthe 3� and 5� ends of this RNA indicated that its 3� endwas the same as that of the orf138-orfB mRNA inOgura plants, but that its 5� end lay within the orf138coding sequence. We have shown that the orf138-orfBlocus is transcribed from a promoter upstream fromthe fMtRNA gene, and that the 5� end of the orf138-orfB mRNA undergoes processing (Bellaoui et al.1997). It therefore appears likely that a diVerent, oradditional, processing event produces the 5� end of themajor orfB mRNA in the Ogura-related cytoplasm. Asmall amount of orf138-orfB Ogura mRNA wasdetected in plants with the Ogura-related cytoplasm,and, reciprocally, a small amount of the truncated orfBmRNA was detected in Ogura plants (Fig. 3). Thisobservation suggests that the observed diVerences inthe expression proWle of this locus are based on a quan-titative, rather than qualitative mechanism. The inter-ruption of a sterility gene transcript by a maturationevent is very similar to the action of a fertility restorer(Hanson and Bentolila 2004). However, in this case,the processing event depends on cytoplasm type ratherthan on a nuclear locus. Firstly, the RNA mapped inthis study was the major orfB mRNA only in plantscarrying the European Ogura-related cytoplasm, andnever in plants carrying the Ogura cytoplasm, althoughthe introduction of a restorer from a wild plant ren-dered these plants fertile (data not shown). Secondly,this orfB mRNA also predominated in plants from thesecond backcross with the L7F maintainer line (M2,M4, and M5 in Fig. 3). As these three plants derivedfrom second backcrosses with the L7F maintainer lineand presented the same orfB transcription proWle, thisproWle is very unlikely to result from the residual pres-ence of a restorer allele in all three plants. A similar sit-uation has been described in Arabidopsis thaliana, inwhich cox3 transcript maturation diVers in the C24 andCol0 cytoplasms (Forner et al. 2005). However, in thiscase, diVerences in distant upstream sequences pro-vided a possible explanation for the diVerences in mat-uration sites. In the case presented here, we haveidentiWed no sequence diVerence likely to account forour observations. It is unlikely that the single nucleo-tide substitution inside the orf138 gene aVects the pro-cessing of the orf138 RNA because this substitution isalso present in male-sterility inducing cytoplasms of

123

1342 Theor Appl Genet (2007) 114:1333–1343

Asian genotypes (Yamagishi and Terachi 1997). Nev-ertheless, no sequencing was carried out outside theexpressed region and, although DNA hybridizationanalysis detected no diVerences between the Oguraand Ogura-related cytoplasms, these two cytoplasmsmay diVer in other regions of the mitochondrialgenome.

Small amounts of orf138-orfB mRNA were detect-able in plants with the Ogura-related cytoplasm, but noORF138 protein could be detected in protein extractsfrom these plants (data not shown). This small amountof mRNA therefore appears to be insuYcient for pro-tein production, accounting for the lack of sterilityinduction by the wild cytoplasm.

We conclude that the Ogura-related cytoplasm ofwild plants in European natural populations carries anorf138 locus of the “ancestral” type that has lost itsability to induce male sterility due to processing of itstranscript, disrupting the orf138 coding sequence. Thepresence of fertility restorers in these populationsstrongly suggests that this cytoplasm was once able toinduce male sterility. To our knowledge, such a situa-tion is unprecedented in the CMS systems generallystudied in natural populations, because the sterilitygenes have not been identiWed and cannot be followedindependently of sterility phenotype. Theoretical pre-dictions concerning CMS maintenance in gynodioe-cious populations suggest that a loss of sterilityinduction would be expected when restorer allelesbecome Wxed in the population (Bailey et al. 2003). Inthe populations studied here, it is clear that the restor-ers are not Wxed. However, we cannot rule out thepossibility that they were once Wxed rendering thesterility-associated cytoplasm neutral and allowingunknown cytoplasmic changes to impair orf138 expres-sion, resulting in the observed male fertility “reversion”.The low frequency of the wild Ogura-related cyto-plasm in European populations suggests that there mayhave been a subsequent drift eVect on this cytoplasm.Murayama et al. (2004) found that restorers were notWxed in Japanese wild radish populations, and sug-gested that this may be due to a cost of restoration—adecrease in the Wtness of plants carrying the fertilityrestorer but not the sterility-inducing cytoplasm. Wedo not yet know whether the restorers present in Euro-pean populations are the same as those present in Jap-anese populations, and cannot evaluate the possiblecosts associated with these restorers. However, assum-ing that European restorers do have a cost, they wouldbe expected to accumulate mutations subject to possi-ble selection after the “reversion” of the Ogura-relatedcytoplasm. It should be possible to test this hypothesisby comparing restorer and non-restorer alleles from

European populations and analyzing their phylogenicrelationship. The original situation described here pro-vides an opportunity for experimental studies of thefate of restorer genes in the absence of the correspond-ing male sterility cytoplasm. The striking diVerencebetween the situations of the Ogura CMS system inJapanese and European natural populations is diYcultto understand at this stage. It would be interesting tocompare the demographic situations of the popula-tions, to establish the evolutionary scenario underlyingthis “chronicle of a death foretold”.

Acknowledgments We thank P. Touzet and M. Dufaÿ for usefulcomments during the preparation of the manuscript. We thank H.Mireau for critical reading of the manuscript. We thank Dr. Bon-net for generously providing the radish maintainer line for OguraCMS. We also thank P. Saumitou-Laprade for kindly helping uswith the choice of sequenced plastid genome regions, and thecommunication of primer sequences. We would like to thankA.M. Chèvre, F. Eber, P. Touzet, G. Guéritaine, J. Noun L. Gea-gea, and the Gauguin family for their precious help in the pros-pecting of natural populations. The Lebanese populations wereprospected in the framework of a French-Lebanese program(04EF22/L12) coordinated by M. Durand-Tardif, funded by the“Coopération pour l’Evaluation et le Développement de laRecherche”. SG was supported by a Saint-Exupéry fellowshipfrom the French Embassy in Buenos Aires. YR was supported bya fellowship from the P. R. China.

References

Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG,Smith JA, Struhl K (1990) Current protocols in molecularbiology. Greene, Wiley Interscience, New York

Bailey MF, Delph LF, Lively CM (2003) Modeling gynodioecy:novel scenarios for maintaining polymorphism. Am Nat161(5):762–776

Bellaoui M, Pelletier G, Budar F (1997) The steady-state level ofmRNA from the Ogura cytoplasmic male sterility locus inBrassica cybrids is determined post-transcriptionally by its 3�region. EMBO J 16(16):5057–5068

Bentolila S, Alfonso AA, Hanson MR (2002) A pentatricopep-tide repeat-containing gene restores fertility to cytoplasmicmale-sterile plants. Proc Natl Acad Sci USA 99(16):10887–10892

Bonhomme S, Budar F, Férault M, Pelletier G (1991) A 2.5 kbNcoI fragment of Ogura radish mitochondrial DNA is corre-lated with cytoplasmic male-sterility in Brassica cybrids.Curr Genet 19:121–127

Bonhomme S, Budar F, Lancelin D, Small I, Defrance M-C, Pelle-tier G (1992) Sequence and transcript analysis of the Nco2.5Ogura-speciWc fragment correlated with cytoplasmic male ste-rility in Brassica cybrids. Mol Gen Genet 235:340–348

Bonhomme S, Budar F, Lancelin D, Pelletier G (1999) DNA se-quence conferring cytoplasmic male sterility, mitochondrialgenome, nuclear genome, mitochondria and plant containingsaid sequence and process for the preparation of hybrids.European Patent EP0909815

Bonnet A (1977) Breeding in France of a radish F1 hybrid ob-tained by use of cytoplasmic male sterility. Eucarpia Crucife-rae Newsl 2:5

123

Theor Appl Genet (2007) 114:1333–1343 1343

Brown GG, Formanova N, Jin H, Wargachuk R, Dendy C, PatilP, Laforest M, Zhang J, Cheung WY, Landry BS (2003) Theradish Rfo restorer gene of Ogura cytoplasmic male sterilityencodes a protein with multiple pentatricopeptide repeats.Plant J 35(2):262–272

Budar F, Delourme R, Pelletier G (2004) Male sterility. In: PuaEC, Douglas (eds) Biotechnology in agriculture and forestry54. Springer, Berlin, pp 43–64

Budar F, Touzet P, Pelletier G (2006) Cytoplasmic male sterility.In: Ainsworth C (ed) Flowering and its manipulation. Black-well, Dordrecht 20:147–180

Charlesworth D (2002) What maintains male-sterility factors inplant populations? Heredity 89(6):408–409

Cui X, Wise RP, Schnable PS (1996) The rf2 nuclear restorer geneof male-sterile T-cytoplasm maize. Science 272(5266):1334–1336

Darwin C (1877) The diVerent forms of Xowers on plants of thesame species. John Murray, London

Dellaporta J, Wood J, Hicks J (1983) A plant DNA miniprepara-tion: version II. Plant Mol Biol Rep 1:19–21

Desloire S, Gherbi H, Laloui W, Marhadour S, Clouet V, Catto-lico L, Falentin C, Giancola S, Renard M, Budar F, Small I,Caboche M, Delourme R, Bendahmane A (2003) IdentiWca-tion of the fertility restoration locus, Rfo, in radish, as amember of the pentatricopeptide-repeat protein family.EMBO Rep 4(6):588–594

Duroc Y, Gaillard C, Hiard S, Defrance MC, Pelletier G, Budar F(2005) Biochemical and functional characterization of ORF138,a mitochondrial protein responsible for Ogura cytoplasmic malesterility in Brassiceae. Biochimie 87(12):1089–1100

Eber F, Chèvre AM, Baranger A, Vallée P, Tanguy X, Renard M(1994) Spontaneous hybridization between a male-sterileoilseed rape and two weeds. Theor Appl Genet 88:362–368

Forner J, Weber B, Wietholter C, Meyer RC, Binder S (2005)Distant sequences determine 5� end formation of cox3 tran-scripts in Arabidopsis thaliana ecotype C24. Nucleic AcidsRes 33(15):4673–4682

Grelon M, Budar F, Bonhomme S, Pelletier G (1994) Ogura cyto-plasmic male-sterility (CMS)-associated orf138 is translatedinto a mitochondrial membrane polypeptide in male-sterileBrassica cybrids. Mol Gen Genet 243:540–547

Grivet D, Heinze B, Vendramin GG, Petit R (2001) Genome walk-ing with consensus primers: application to the large single copyregion of chloroplast DNA. Mol Ecol Notes 1:345–349

Hanson MR, Bentolila S (2004) Interactions of mitochondrialand nuclear genes that aVect male gametophyte develop-ment. Plant Cell 16 (Suppl):S154–S169

Havey MJ (2004) The use of cytoplasmic male sterility for hybridseed production. In: Daniell H, Chase C (eds) Molecularbiology and biotechnology of plant organelles. Springer,Heidelberg, pp 623–634

Koizuka N, Imai R, Fujimoto H, Hayakawa T, Kimura Y, Kohno-Murase J, Sakai T, Kawasaki S, Imamura J (2003) Geneticcharacterization of a pentatricopeptide repeat protein gene,orf687, that restores fertility in the cytoplasmic male-sterileKosena radish. Plant J 34(4):407–415

Krishnasamy S, MakaroV CA (1993) Characterization of the rad-ish mitochondrial orfB locus: possible relationship with malesterility in Ogura radish. Curr Genet 24(1–2):156–163

Kuhn J, Binder S (2002) RT-PCR analysis of 5� to 3�-end-ligatedmRNAs identiWes the extremities of cox2 transcripts in peamitochondria. Nucleic Acids Res 30(2):439–446

Loudet O, Chaillou S, Camilleri C, Bouchez D, Daniel-Vedele F(2002) Bay-0 x Shahdara recombinant inbred line

population: a powerful tool for the genetic dissection ofcomplex traits in Arabidopsis. Theor Appl Genet 104(6–7):1173–1184

Lurin C, Andres C, Aubourg S, Bellaoui M, Bitton F, Bruyere C,Caboche M, Debast C, Gualberto J, HoVmann B, LecharnyA, Le Ret M, Martin-Magniette ML, Mireau H, Peeters N,Renou JP, Szurek B, Taconnat L, Small I (2004) Genome-wide analysis of Arabidopsis pentatricopeptide repeatproteins reveals their essential role in organelle biogenesis.Plant Cell 16(8):2089–103

Murayama K, Yahara T, Terachi T (2004) Variation of femalefrequency and cytoplasmic male-sterility gene frequencyamong natural gynodioecious populations of wild radish(Raphanus sativus L.). Mol Ecol 13(8):2459–2464

Ogura H (1968) Studies on the new male sterility in Japanese rad-ish, with special references to utilization of this sterility to-wards the practical raising of hybrid seeds. Mem Fac AgricKagoshima Univ 6:39–78

Pelletier G, Primard C, Vedel F, Chétrit P, Rémy R, Rousselle P,Renard M (1983) Intergeneric cytoplasmic hybridization inCruciferae by protoplast fusion. Mol Gen Genet 191:244–250

Perrin R, Meyer EH, Zaepfel M, Kim YJ, Mache R, Grienenber-ger JM, Gualberto JM, Gagliardi D (2004) Two exoribonuc-leases act sequentially to process mature 3�-ends of atp9mRNAs in Arabidopsis mitochondria. J Biol Chem279(24):25440–25446

Saur Jacobs M, Wade MJ (2003) A synthetic review of the theoryof gynodioecy. Am Nat 161(6):837–851

Terachi T, Yamaguchi K, Yamagishi H (2001) Sequence analysison the mitochondrial orfB locus in normal and Ogura male-sterile cytoplasms from wild and cultivated radishes. CurrGenet 40(4):276–281

Touzet P, Budar F (2004) Unveiling the molecular arms race be-tween two conXicting genomes in cytoplasmic male sterility?Trends Plant Sci 9(12):568–570

Wang Z, Zou Y, Li X, Zhang Q, Chen L, Wu H, Su D, Chen Y,Guo J, Luo D, Long Y, Zhong Y, Liu YG (2006) Cytoplas-mic male sterility of rice with Boro II cytoplasm is caused bya cytotoxic peptide and is restored by two related PPR motifgenes via distinct modes of mRNA silencing. Plant Cell18(3):676–687

Yamagishi H, Terachi T (1994a) Molecular and biological studieson male sterile cytoplasm in Cruciferae. II. The origin ofOgura male sterile cytoplasm inferred from segregation pat-tern of male sterility in the F1 progeny of wild and cultivatedradishes (Raphanus sativus L.). Euphytica 80:201–206

Yamagishi H, Terachi T (1994b) Molecular and biological studieson male-sterile cytoplasm in the Cruciferae. I. The origin anddistribution of Ogura male-sterile cytoplasm in Japanesewild radishes (Raphanus sativus L.) revealed by PCR-aidedassay of their mitochondrial DNAs. Theor Appl Genet87:996–1000

Yamagishi H, Terachi T (1996) Molecular and biological studieson male-sterile cytoplasm in the Cruciferae. III. Distributionof Ogura-type cytoplasm among Japanese wild radishes andAsian radish cultivars. Theor Appl Genet 93:325–332

Yamagishi H, Terachi T (1997) Molecular and biological studieson male-sterile cytoplasm in the Cruciferae. IV. Ogura-typecytoplasm found in the wild radish, Raphanus raphanistrum.Plant Breed 116:323–329

Yamagishi H, Terachi T (2001) Intra-and inter-speciWc variationsin the mitochondrial gene orf138 of Ogura-type male sterilecytoplasm from Raphanus sativus and Raphanus raphani-strum. Theor Appl Genet 103:725–732

123