The Molecular Genetics of Self-incompatibility in Petunia hybrida

Annals of Botany 85 (Supplement A): 105±112, 2000 doi:10.1006/anbo.1999.1062, available online at http://www.idealibrary.com on

The Molecular Genetics of Self-incompatibility in Petunia hybrida T. P. RO B B I N S * , R . M . H A R B O R D , T. SO N N E V E L D and K . C L A R K E Plant Science Division, School of Biological Sciences, University of Nottingham, Sutton Bonington Campus, Loughborough, LE12 5RD, UK Received: 21 July 1999

Accepted: 20 October 1999

The cultivated petunia (Petunia hybrida) has been a popular system in which to study genetic, physiological and biochemical aspects of gametophytic self-incompatibility. As with other members of the Solanaceae a number of S-RNase genes have been isolated for functional S-alleles. We have identi®ed S-RNase sequences for two additional functional S-alleles, Sv and S3 . These alleles are more similar to alleles from other families of the Solanaceae (Nicotiana and Solanum) than to any petunia alleles reported previously. The total number of S-alleles in P. hybrida is at least ten in spite of its cultivated origin. However, most cultivars of P. hybrida are in fact self-compatible and this appears to arise from the prominence of a single previously described allele So . The implications of this observation for the origin of self-compatibility in P. hybrida are discussed. The S-locus of P. hybrida has recently been mapped using an indirect method involving T-DNA insertions. Seven T-DNA insertions that were previously shown to be closely linked to the S-locus were physically mapped on the long arm of chromosome III using ¯uorescent in-situ hybridization. The most tightly linked T-DNA insertions are in a sub-centromeric position. This is consistent with the centric fragments of P. in¯ata obtained by irradiation mutagenesis that carry additional S-loci and confer a pollenpart mutant phenotype. An S-linked restriction fragment length polymorphism (RFLP) marker, CP100 was used to con®rm this chromosomal assignment and has provided evidence for S-locus synteny in the Solanaceae. # 2000 Annals of Botany Company Key words: Petunia hybrida, petunia, self-incompatibility, self-compatibility, S-RNase, PCR, T-DNA.

I N T RO D U C T I O N Over the past 30 years the cultivated petunia, Petunia hybrida, has contributed to many aspects of self-incompatibility research. The large ¯owers are amenable to controlled pollinations and provide sucient anther and pistil material for biochemical analyses. The contributions of P. hybrida to both genetic and biochemical studies have been reviewed previously (Ascher, 1984). In this review, recent molecular genetic studies will be discussed in relation to the earlier ®ndings. The contribution of P. hybrida will be placed in the context of the major advances in our understanding of S-ribonuclease (S-RNase) based incompatibility systems in other Solanaceae, and in particular the wild relative P. in¯ata. P. hybrida is believed to have been derived in the early part of the last century from hybridization of two or more wild petunia species (Sink, 1984). Molecular ®ngerprinting (Peltier et al., 1994; Cerny et al., 1996) has supported the view that major contributions were from a purple ¯owered species ( probably P. integrifolia) and a white ¯owered species (P. axillaris or P. parodii). The former species are invariably self-incompatible outbreeders whereas the latter species can be either self-incompatible or self-compatible (Tsukamoto et al., 1999). Although a number of selfincompatible cultivars of P. hybrida have been identi®ed * For correspondence. Fax ‡44 (0) 115-951-6334, e-mail tim. [email protected]

0305-7364/00/0A0105+08 $35.00/00

(Linskens and Straub, 1978; Ascher, 1984; this paper) the predominant condition is self-compatibility with abundant seed set on self-pollination. As with all other solanaceous species studied to date, the self-incompatibility in P. hybrida is gametophytic with a single multiallelic S-locus. Polymorphic pistil proteins were found to be associated with the S-alleles originally described by Linskens and Straub (Kamboj and Jackson, 1986). These were subsequently puri®ed and N-terminal sequences were obtained for three alleles S1 , S2 and S3 (Broothaerts et al., 1989). These sequences showed homology with the C1 conserved domain of pistil S-proteins previously identi®ed in Nicotiana alata and Lycopersicon peruvianum (Ioerger et al., 1991). These proteins were subsequently shown to have ribonuclease activity (Broothaerts et al., 1991) similar to that reported for the pistil S-proteins of N. alata (McClure et al., 1989). These pistil S-proteins therefore show all the properties anticipated of S-ribonucleases (S-RNases) that have now been extensively characterized in the Solanaceae (reviewed in Kao and McCubbin, 1996). The ®rst S-RNase cDNA sequences reported in P. hybrida were obtained by screening cDNA libraries with an oligonucleotide probe based on a conserved region (Clark et al., 1990). They reported S-RNase sequences for three alleles also named, S1 , S2 and S3 , but obtained from an independent source (Ascher, 1984). A comparison of the putative mature S-RNase sequences encoded by these alleles con®rms that they are distinct from the Linskens alleles previously characterized by Broothaerts (Fig. 1). In # 2000 Annals of Botany Company

106

Robbins et al.ÐSelf-incompatibility in Petunia approaches (Lee, Huang and Kao, 1994) is consistent with the view that the pollen determinant of self-incompatibility ( pollen-S) is controlled by a separate, as yet unidenti®ed, gene at the S-locus. I D E N T I F I CAT I O N O F T WO N OV E L S - R NA S E S I N S E L F - I N CO M PAT I B L E P. H Y B R I D A

F I G . 1. A comparison of N-terminal sequences reported for S-RNase genes of P. hybrida. The N-terminal sequences of the Linskens alleles reported by Broothaerts et al. (1989) are aligned with those of the Ascher alleles derived from cDNA sequences (Clark et al., 1990). The amino acids conserved between the C1 domain of all the Solanaceae S-RNases (Ioerger et al., 1991) are indicated with asterisks. Also included are the Sx and So alleles reported by Ai et al. (1992), the Sb allele reported by Broothaerts et al. (1991) and the Sv sequence reported here. Allele names in parentheses are those proposed to avoid future confusion.

addition they are all distinct from the three S-RNase sequences reported for the wild species P. in¯ata (Ai et al., 1990). Two additional S-RNase sequences have been reported in P. hybrida, So and Sx , the former of which appears to be non-functional (Ai, Kron and Kao, 1991; Ai, Tsai and Kao, 1992). There may be as many as ten distinct S-alleles in P. hybrida cultivars suggesting a fairly complex breeding history involving more than two parent individuals. Expression analysis in P. hybrida at the protein level showed that the Linskens S2 glycoprotein accumulates in the top half of the pistil during the latter stages of ¯oral bud expansion (Broothaerts et al., 1989). The accumulation of corresponding mRNAs was studied using the Ascher S1 ±S3 alleles by Clark et al. (1990). They showed that all three alleles were highly expressed in styles but only weakly in ovaries. The accumulation of stylar mRNA is consistent with the onset of self-incompatibility as revealed previously in Nicotiana alata (Cornish et al., 1987). As in other Solanaceae, it is possible to overcome self-incompatibility by pollinating immature pistils, before the S-RNase has accumulated to sucient levels, and this allows the generation of homozygous S-allele stocks. In addition to the anticipated expression in the pistil, very low levels of expression were detected using reverse transcriptase-polymerase chain reaction (RT-PCR) in immature anthers for the S1 but not the S3 RNase (Clark and Sims, 1994). These ®ndings are similar to those in N. alata (Dodds et al., 1993), and recent studies in Lycopersicon peruvianum call into question the relevance of these low levels of expression for self-incompatibility (Dodds et al., 1999). The pollen-part mutations described in P. in¯ata (Brewbaker and Natarajan, 1960) and the inability to alter pollen phenotype using S-RNase transgenic

We have studied two self-incompatible petunia cultivars, one (T2U) homozygous for the S3 allele (Linskens and Straub, 1978), the other cultivar (V13) homozygous for a previously uncharacterized allele provisionally called Sv (Harbord, Napoli and Robbins, 2000). To further characterize these alleles a degenerate PCR cloning procedure was adopted to identify S-RNase sequences expressed in pistils. A degenerate primer based on the C2 conserved domain of S-RNases from the Solanaceae (Ioerger et al., 1991) was used to amplify partial S-RNase sequences by RT-PCR of pistil RNA (Xue et al., 1996). Several identical cloned PCR products with homology to other S-RNases were obtained for each allele. Con®rmation that these cDNA sequences corresponded to the Sv and S3 RNases was obtained by designing allele speci®c primers based on the hypervariable domain and the 30 untranslated region. These were used for genomic PCR analysis of a small family that segregated for both alleles (Fig. 2). Allele speci®c PCR products were found to cosegregate perfectly with the S-genotypes determined by test pollination. In a more extensive analysis of 90 progeny from the same cross, the two allele speci®c PCR products segregated as alleles of the same locus (Clarke and Robbins, data not shown). These ®ndings suggest that the cloned sequences represent functional alleles and not S-like RNases that are unlinked to the S-locus such as X2 in P. in¯ata (Lee, Singh and Kao, 1992). Additional allele speci®c primers were subsequently used to isolate full coding region sequences by 50 rapid ampli®cation of cDNA ends-polymerase chain reaction (50 RACE-PCR) as described by Xue et al. (1996). The full amino acid sequences predicted for the S3 and Sv S-RNase precursors are aligned in Fig. 3. Both S-RNase sequences contain the conserved domains C1±C5 typical for S-RNase sequences from the Solanaceae (Ioerger et al., 1991). The Sv sequence is most closely related to S7 from Nicotiana alata (67% identity) and the most similar allele in P. hybrida is Sx (51% identity; Ai et al., 1992). The S3 sequence is most closely related to S11 from Solanum chacoense (66% identity) and the most similar allele in P. hybrida is again Sx (58% identity). The higher level of interspeci®c rather than intraspeci®c similarity is an established feature of many S-RNase sequences in the Solanaceae (Ioerger, Clark and Kao, 1990). The predicted amino acid sequence of S3 can be compared with the N-terminal sequence for the puri®ed S3 protein published previously (Fig. 1). There is full agreement between the cDNA and protein sequence for all 19 amino acid positions that were determined unambiguously (Broothaerts et al., 1989). The predicted molecular weight (22.1 kD) is in approximate agreement with that reported

Robbins et al.ÐSelf-incompatibility in Petunia

107

F I G . 2. S-RNase allele speci®c PCR products cosegregate with S-phenotypes in P. hybrida. A small family of plants segregating for the S3 and Sv alleles were generated by crossing an S3Sv stock with pollen from an S3Sb stock. The progeny were genotyped by test pollination with the S3Sb stock and found to segregate 1 : 1 for S3Sb and SvSb as anticipated. Progeny determined to carry the S3 or Sv allele are indicated together with the results obtained from two di€erent allele speci®c genomic PCR reactions. A 500 bp product from Sv allele speci®c PCR was found to cosegregate with the Sv phenotype (upper panel) and a 450 bp product from the S3 allele speci®c PCR with the S3 phenotype (lower panel). Methods: Genomic DNA was prepared from 11 genotyped plants as described elsewhere (Harbord et al., 2000) and approx. 30±50 ng was used for genomic PCR. Genomic PCR was carried out in a manufacturer's bu€er with 3 mM MgCl2 , 0.2 mM dNTPs and 0.5 units Taq polymerase (Bioline UK, Ltd) for 35 cycles of: 948C 30 secs, speci®ed annealing temperature 30 secs, 728C 60 secs. The Sv allele speci®c primers (V13-F1 50 GGACGAAGCTGATTGTAAGGG and V13-R1 50 CGATTTTCATATATTGGC) were used at 0.1 mM and 458C annealing temperature. The S3 allele speci®c primers (PhS3-F1 50 CACTACATTCGCGGGTAAGATGCTC and PhS3-R1 50 CGCATGTATCACTTTGACGACAGG) were used at 0.2 mM and 608C annealing temperature.

previously after deglycosylation (25.1 kD; Broothaerts et al., 1991). The Sv allele was known to be functionally distinct from S3 and Sb in test pollinations but it was not clear how it was related to other previously described P. hybrida S-alleles. Based on the predicted sequence of the C1 region it is possible that Sv is equivalent to the S2 allele from Linskens allowing for some ambiguous amino acid sequences. Unfortunately we have not been able to obtain any S2 stocks and we prefer to identify Sv as a novel S-RNase until shown otherwise. It is worth pointing out that the S2 and S3 alleles of Ascher also appear identical in this region yet they are known to be functionally distinct (Dana and Ascher, 1986) and have di€erent cDNA sequences (Clark et al., 1990). To avoid future confusion between the S1 ±S3 alleles described by Linskens and Straub (1978) and the S1 ±S3 alleles from Ascher that were sequenced by Clark et al. (1990) it seems desirable to introduce a sux, L, as shown in Fig. 1. Since this report contains the ®rst full length sequence for one of Linsken's alleles we propose that this allele should subsequently be called S3L . It is also important in any sequence comparisons to distinguish the P. hybrida alleles from the structurally distinct alleles of P. in¯ata that are also called S1 ±S3 (Ai et al., 1990). During the preparation of this manuscript two additional P. hybrida S-RNase sequences were submitted to the European Molecular Biology Laboratory (EMBL) database (accession numbers AB016522, AB016523). These amino acid sequences are identical to the Sv and S3L alleles described

in this report although they are presumably derived from distinct cultivars. MOLECULAR BASIS OF S E L F - CO M PAT I B I L I T Y S U G G E S T S A CO M M O N O R I G I N In spite of its contribution to self-incompatibility research, most cultivars of P. hybrida are in fact self-compatible. Selfcompatibility in P. hybrida has been studied by Ascher and colleagues and has been reviewed previously (Ascher, 1984). Ascher uses the term pseudo-self-compatibility (PSC) to describe both physiological and genetic causes for a breakdown in self-incompatibility. In some cases there is complex polygenic control and intermediate levels of seed set are observed. In at least one case it has been shown that PSC is not a result of reduced expression of an S-RNase (Clark et al., 1990). Ascher also describes lines with 100% PSC or rather full self-compatibility and in one case this arises from a loss of pollen function (Dana and Ascher, 1986). Interestingly, this line seems to lack a centric fragment which is characteristic of the pollen-part mutants arising from duplications described in Petunia in¯ata (Brewbaker and Natarajan, 1960). The mutation appears to be tightly, but not absolutely, linked to the S-locus. The numerous self-compatible cultivars of P. hybrida developed for commercial and research purposes o€er a diverse source of self-compatible stocks. In a previous study a commercial F1 hybrid was crossed to self-incompatible

108

Robbins et al.ÐSelf-incompatibility in Petunia

F I G . 3. Full amino acid sequences of two previously unreported P. hybrida S-RNase sequences. The full precursor protein sequences of the Sv and S3L S-RNases are shown based on cDNA sequences. The amino terminus of the mature S3 protein is underlined (see Fig. 1). The ®ve conserved domains for S-RNases from the Solanaceae (C1±C5), the two major hypervariable domains (HVa and HVb) and several conserved cysteine (C) residues are indicated (Ioerger et al., 1991). Vertical lines show positions that are identical between the two P. hybrida S-RNase sequences. Methods: Total pistil RNA was prepared from S-allele homozygotes maintained by bud sel®ng and used for 30 RACE-PCR using the method and C2 degenerate primer described by Xue et al. (1996). PCR products were cloned into the TA cloning vector (Invitrogen) and clones with homology to S-RNases identi®ed. To obtain full length sequences the 50 RACE-PCR procedure was used with a commerical kit (Bethesda Research Laboratories) and allele speci®c primers (sequences available on request). All DNA sequences are based on at least three independent clones and analysed using the DNAstar software and the Clustal alignment method (Madison, USA). The DNA sequences have been submitted to the EMBL database under the accession numbers AJ271062(SV) and AJ271065(S3L).

P. in¯ata to test the genetic basis of the self-compatibility (Ai et al., 1991). Two distinct S-alleles derived from the P. hybrida parent, Sx and So , were identi®ed in progeny by SDS-PAGE analysis. All progeny carrying the So allele were self-compatible whereas progeny carrying the Sx allele were either self-compatible or self-incompatible. This suggested that the So allele is non-functional and the Sx allele is conditionally functional perhaps due to the segregation of modi®er genes. Subsequently cDNA clones were obtained for both S-RNases (Fig. 1) and both alleles were shown to have ribonuclease activity (Ai et al., 1992). We have obtained similar results in crosses between the self-incompatible cultivar V13 (SvSv) and a self-compatible inbred cultivar V26 (Harbord, Napoli and Robbins, 2000). The self-compatible stock was assumed to be homozygous for a non-functional S-allele, termed Sc ˆ self-compatible. An F1 hybrid was generated and found to be selfcompatible but incompatible with pollen from the V13 parent. This indicated that the Sv allele is functional in the pistil of these F1 plants. Using the early bud pollination method this F1 hybrid was backcrossed to the selfincompatible parent V13 and 65 backcross individuals were tested for self-compatibility. It was anticipated that half of these backcross progeny would be homozygous SvSv and half heterozygous SvSc . Approximately half (52.3%)

were self-compatible and the remainder were either fully self-incompatible (27.7%) or partially incompatible (20.0%) setting intermediate levels of seed (Harbord et al., 2000). The latter class may be equivalent to the pseudo-selfcompatible types described previously by Ascher (1984). Intermediate levels of seed set were also observed in the progeny of P. hybrida  P. in¯ata crosses that carry the Sx allele (Ai et al., 1990). To analyse the segregation of S-alleles in this backcross two approaches were utilized. The ®rst approach was indirect, using an S-linked RFLP marker CP100 which suggested that all self-incompatible progeny were homozygous for the Sv-linked allele of CP100 (Harbord et al., 2000). This suggested that the Sc allele may be nonfunctional and this was tested further by the cloning of an S-RNase from the V26 line (Robbins, unpubl. res.). Using the same approach described above for the functional alleles Sv and S3 , 30 RACE-PCR was used to amplify sequences that contain the conserved C2 domain from pistil RNA of the self-compatible V26 cultivar. Three independent clones with S-RNase homology were identical in sequence to the previously published So allele (Ai et al., 1992). Allele speci®c primers were designed based on the full length published So sequence and found to amplify identical size bands in So and Sc (data not shown). These

Robbins et al.ÐSelf-incompatibility in Petunia

109

F I G . 4. An So allele-speci®c PCR product cosegregates with self-compatibility in the backcross ScSv  SvSv . A family of 65 plants was generated from a (V13  V26)  V13 backcross and tested for self-incompatibility (see Harbord et al., 2000). This family should segregate for the Sc allele of V26 and a selection of 12 progeny that were unambiguously self-incompatible (SI) or self-compatible (SC) were tested for the presence of the Sc(So) allele by genomic PCR. The expected 800bp product was detected in a V26 control and all SC progeny. Methods: Genomic PCR was carried out as decribed in Fig. 3 except that the primers were used at 0.2 mM and an annealing temperature of 578C. The primers used (SoF2: 50 GGTTATTGCAAAATTAAGGG and V26-R5: 50 ATGTTCTCTCTTCGAAGTTCGCG) were based on the published 50 UTR sequence of So (Ai et al., 1992) and the 30 UTR of So(Sc) from V26 (Robbins, unpubl. res.).

primers were used to monitor the segregation of the Sc allele in the backcross progeny described above. All the fully selfincompatible progeny were Sv/Sv whereas the fully selfcompatible progeny were Sv/So (Fig. 4). The partially self-incompatible progeny were Sv/Sv (data not shown) which was consistent with the genotyping using the CP100 marker (Harbord et al., 2000). This con®rms that the So allele is non-functional as reported by Ai et al. (1990), and suggests that unlinked modi®ers are also segregating that weaken the self-incompatibility in the Sv/Sv homozygotes. The ®nding that the same So allele is present in two apparently unrelated commercial stocks prompted us to conduct a survey of commercial cultivars and research stocks. Using the allele speci®c primers designed for So we obtained genomic PCR products of the expected size for 12 out of 15 commercial F1 hybrids tested and 13 out of 16 research stocks obtained from the Free University of Amsterdam (data not shown). This suggests that the So allele is present in approx. 80% of existing research stocks and commercial cultivars and is probably a major contributor to the self-compatibility observed in many P. hybrida varieties propagated today. However, it cannot be excluded that modi®er loci are also present and preliminary data suggest that at least two inbred lines other than V26 (W80 and W115) show backcross segregations consistent with modi®ers and polygenic control (Robbins and Harbord, unpubl. res.). The self-compatibility of V26 and other cultivars may therefore represent a compound e€ect of the So allele and other unlinked compatibility factors. It is also possible that self-compatible cultivars carry functional S-alleles that are rendered inactive due to such modi®ers as proposed for the original source of the Sx allele (Ai et al., 1991). It is interesting to speculate on why the So allele should predominate in self-compatible varieties (Fig. 5). One possibility is that the self-compatible allele was present in the original crosses between wild species that gave rise to P. hybrida. Indeed, P. axillaris, which is believed to be one

of the parental species, is frequently self-compatible in the wild (Tsukamoto et al., 1999). This is in contrast to P. integrifolia which has been proposed as the other major species contributing to P. hybrida (Peltier et al., 1994; Cerny et al., 1996), which is an obligate outbreeder. It will be interesting to see whether the So allele is present in any of these natural populations of P. axillaris. An alternative possibility is that the self-compatibility arose as a result of hybridization. Some evidence to support this view is that the So allele is expressed at high levels, has all the structural features of an S-RNase, and most signi®cantly has been shown to have ribonuclease activity (Ai et al., 1992). These ®ndings are consistent with the view that the S-RNase of the So allele is fully functional, and whilst RNase function is not the sole prerequisite for stylar expressed S-gene function in the Solanaceae, the self-compatibility is most likely to result from a mutation a€ecting pollen-S function. A similar situation may apply in Nicotiana sylvestris which is exceptional amongst self-compatible Nicotiana species in having an apparently functional S-RNase (Golz

F I G . 5. Origin of self-compatibility in Petunia hybrida. The possible origin of self-compatibility (SC) and self-incompatibility (SI) in modern cultivated P. hybrida is shown schematically.

110

Robbins et al.ÐSelf-incompatibility in Petunia

et al., 1998). Phylogenetic analysis places the N. sylvestris S-RNase with other functional S-RNases and the authors proposed the term `relic S-RNase'. The self-compatibility of N. sylvestris may also result from a mutation in pollen-S. In contrast to this, a rare self-compatible accession in Lycopersicon peruvianum revealed an active site mutation that abolishes ribonuclease activity (Royo et al., 1994). This is consistent with the observation that S-RNases with a modi®ed active site are also inactive in transgenic Petunia in¯ata (Huang et al., 1994). It will be interesting to examine whether pollen-part mutations are a frequent cause of self-compatibility in other cultivated species such as L. esculentum. C Y TO G E N E T I C A N D G E N E T I C M A P P I N G O F T H E S - LO C U S I N P. H Y B R I D A Until relatively recently the cytological position of the Slocus in P. hybrida was uncertain. There was a single report of linkage between the Grandi¯ora/Undulata locus on chromosome V and the S-locus based on segregation distortion (de Nettancourt, 1977). The seven chromosomes of P. hybrida can be divided into three classes based on centromere position (Smith, Oud and de Jong, 1973). The chromosomes are amenable to direct gene localisation by ¯uorescent in-situ hybridization (FISH) as demonstrated by ten Hoopen et al. (1996). However, the ability to map S-RNase genes directly by FISH has been hampered by the small target size (ten Hoopen and Robbins, unpubl. res.). A much larger target is provided by a T-DNA integrated into the genome and several transgenes have been localized previously on chromosome II (ten Hoopen et al., 1996). Recently, several T-DNA insertions have been mapped close to the S-locus of P. hybrida using a segregation distortion assay (Harbord et al., 2000). These T-DNA insertions o€ered an opportunity to map the S-locus by an indirect method. In collaboration with colleagues at the University of Amsterdam we have localized seven of these T-DNA insertions by ¯uorescent in-situ hybridization (ten Hoopen et al., 1998). All seven S-linked T-DNA insertions were localized to the long arm of chromosome III. The two T-DNA insertions most tightly linked to the S-locus (within 2 cM) were shown to be within 20% of total chromosome length of the centromere whereas more loosely linked T-DNAs had a more distal location. This suggests that the S-locus is located in the vicinity of the centromere, a ®nding consistent with a class of self-compatible mutants generated by irradiation in P. in¯ata (Brewbaker and Natarajan, 1960). These mutants carry `centric fragments', small centromere-bearing chromosome fragments that confer an additional S-speci®city to otherwise normal diploid plants (see Golz et al., 2000). An independent con®rmation that the S-locus of P. hybrida maps to chromosome III was obtained using an anonymous potato RFLP marker, CP100. This RFLP marker had previously been shown to map genetically to the same position as a potato S-RNase gene (Gebhardt et al., 1991). Similarly, in P. hybrida this RFLP marker was shown to cosegregate with S-phenotypes in a testcross

family of 48 individuals (Harbord et al., 2000). In a di€erent mapping population the CP100 marker was found to cosegregate with a peroxidase isozyme locus (PrxA) previously mapped to chromosome III (ten Hoopen et al., 1998). Consistent with this localization was the observation that S-linked T-DNAs were loosely linked to a leaf visual marker, yg3 previously assigned to chromosome III. These observations are not in agreement with the reported linkage between the S-locus and the Undulata mutation on chromosome V. However, this was based on segregation distortions that were subsequently called into question by Reimann-Phillip (de Nettancourt, 1977). In other members of the Solanaceae, the S-locus has been assigned to a chromosome genetically. Thus the S-locus of Lycopersicon peruvianum was assigned to chromosome I by segregation distortion (Tanksley and Loaiza-Figueroa, 1985) and the S-RNase of Solanum tuberosum was placed on the homeologous chromosome by RFLP analysis (Gebhardt et al., 1991). There is evidence of synteny around the S-locus in the Solanaceae, the CP100 marker is S-linked in petunia (Harbord et al., 2000) and potato (Gebhardt et al., 1991), and an anodal peroxidase is linked to the S-locus in both petunia (ten Hoopen et al., 1998) and tomato (Bernatzky and Tanksley, 1986). Evidence is thus gathering in support of the view that no major rearrangement of the S-locus region has occurred during speciation of the Solanaceae although local rearrangements sucient to inhibit recombination between alleles cannot be excluded. During the preparation of this manuscript it was reported that a P. hybrida S-RNase gene has been localized directly by FISH at, or very near to, the centromere of chromosome III (Entani et al., 1999). Genomic clones of the S-RNase revealed that the gene was embedded in repetitive sequences present at the centromere of all petunia chromosomes. A centromeric localization for the S-locus opens up the possibility that recombination around the S-locus may be suppressed by centromeric heterochromatin. Recombination is likely to be suppressed at the S-locus to maintain linkage between a coevolved gene complex including the pollen recognition gene and the S-RNase gene. Comparison of S-RNase sequences in the Solanaceae lends support to this view (Clark and Kao, 1991). If it is a selective advantage for the S-locus to be located within a centromere then this should be observed in other members of the Solanaceae. There is some support for this in tomato based on S-linked RFLP markers (Bernatzky and Miller, 1994) showing linkage to centromeric repeats (Broun and Tanksley, 1996). However, a centromeric location for the S-locus in other members of the Solanaceae may only be a re¯ection of synteny (ten Hoopen et al., 1998). For this reason it will be particularly interesting to establish the cytological position of S-RNase loci in other more distantly related families such as the Scrophulariaceae and Rosaceae (Richman, Broothaerts and Kohn, 1997). FUTURE PERSPECTIVES The total number of S-alleles in P. hybrida currently stands at ten, and since the same alleles are now being recovered

Robbins et al.ÐSelf-incompatibility in Petunia independently in unrelated cultivars it may not rise signi®cantly in future surveys. This remains a surprisingly high level of diversity considering that plant breeders probably selected for self-compatibility at an early stage. This has almost certainly contributed to the prevalence of the nonfunctional So allele in current stocks of P. hybrida. The molecular basis of self-compatibility of the So allele is uncertain, but the available data are consistent with a functional S-RNase. One possibility is that it is a pollenpart mutation of a previously functional allele, perhaps as a result of a duplication. It will be interesting to compare the situation in P. hybrida with other cultivated Solanaceae that were derived from self-incompatible species. Understanding the mechanisms that have led to self-compatibility during cultivation may facilitate the genetic engineering of selfcompatibility into self-incompatible species of both the Solanaceae and the Rosaceae which share the same SRNase mechanisms (Richman et al., 1997). Self-compatibility can also arise from the presence of modi®er genes as demonstrated in P. hybrida. The identity of such genes is dicult to predict in view of the simple models of self-incompatibility currently proposed (reviewed by Kao and McCubbin, 1996). However, P. hybrida o€ers several examples of such modi®er genes (Ascher, 1984; Ai et al., 1991; Harbord et al., 2000) that could be genetically characterised. Such modi®er genes may not show allele speci®city and therefore o€er a more general method of inactivating self-incompatibility systems. Finally, as in all of the self-incompatibility systems discussed in this volume, a major goal remains the identi®cation of the pollen determinant of S-locus speci®city. The pollen-S gene must reside at the S-locus and high resolution genetic maps, including tightly linked ¯anking markers such as CP100, should allow the limits of the S-locus to be de®ned. The centromeric location and the suppression of recombination anticipated would predict that this may correspond to a physically large region. The application of S-linked transgenes to preselect for rare recombination events around the S-locus (Harbord et al., 2000) should facilitate this approach. In addition, these S-linked TDNAs carry maize transposable elements which opens up the possibility of genetic screens to identify pollen-S. AC K N OW L E D GE M E N T S The authors would like to thank Dr Wim Broothaerts for supplying seed carrying the S3L allele originally characterized by Prof. H. F. Linskens. Also Dr Ronald Koes (Free University of Amsterdam) for providing V13, V26 and other inbred lines of P. hybrida and Dr Christiane Gebhardt (Max Planck Institute, Cologne) for the potato RFLP marker, CP100. This work was supported by the UK Biotechnology and Biological Sciences Research Council. L I T E R AT U R E C I T E D Ai Y, Kron E, Kao T-h. 1991. S-alleles are retained and expressed in a self-compatible cultivar of Petunia hybrida. Molecular and General Genetics 230: 353±358.

111

Ai Y, Tsai D-S, Kao T-h. 1992. Cloning and sequencing of cDNAs encoding two S proteins of a self-compatible cultivar of Petunia hybrida. Plant Molecular Biology 19: 532±528. Ai Y, Singh A, Coleman CE, Ioerger TR, Kheyr-Pour A, Kao T-h. 1990. Self-incompatibility in Petunia in¯ata: isolation and characterisation of cDNAs encoding three S-allele associated proteins. Sexual Plant Reproduction 3: 130±138. Ascher PD. 1984. Self-incompatibility. In: Sink KC, ed. Petunia: Monographs on theoretical and applied genetics Vol. 9. Berlin: Springer-Verlag, 92±109. Bernatzky R, Miller D. 1994. Self-incompatibility is codominant in intraspeci®c hybrids of self-compatible and self-incompatible Lycopersicon peruvianum and L. hirsutum based on protein and DNA marker analysis. Sexual Plant Reproduction 7: 297±302. Bernatzky R, Tanksley SD. 1986. Toward a saturated linkage map in tomato based on isozymes and random cDNA sequences. Genetics 112: 887±898. Brewbaker JL, Natarajan AT. 1960. Centric fragments and pollen-part mutation of incompatibility alleles in Petunia. Genetics 45: 699±704. Broothaerts WJ, van Laere A, Witters R, Preaux G, Decock B, van Damme J, Vendrig JC. 1989. Puri®cation and N-terminal sequencing of style glycoproteins associated with self-incompatibility in Petunia hybrida. Plant Molecular Biology 14: 93±102. Broothaerts WJ, Vanvinckenroye P, Decock B, van Damme J, Vendrig JC. 1991. Pentunia-hybrida S-proteins: Ribonuclease-activity and the role of their glycan side-chains in self-incompatibility. Sexual Plant Reproduction 4: 258±266. Broun P, Tanksley SD. 1996. Characterization and genetic mapping of simple repeat sequences in the tomato genome. Molecular and General Genetics 250: 39±49. Cerny TA, Caetano-Anolles G, Trigiano RN, Starman TW. 1996. Molecular phylogeny and DNA ampli®cation ®ngerprinting of Petunia taxa. Theoretical and Applied Genetics 92: 1009±1016. Clark KR, Okuley JJ, Collins PD, Sims TL. 1990. Sequence variability and developmental expression of S-alleles in self-incompatible and pseudo-self-compatible petunia. The Plant Cell 2: 815±826. Clark AG, Kao T-h. 1991. Excess nonsynonymous substitution at shared polymorphic sites among self-incompatibility alleles of the Solanaceae. Proceedings of the National Academy of Sciences (USA) 88: 9823±9827. Clark KR, Sims TL. 1994. The S-ribonuclease gene of Petunia hybrida is expressed in nonstylar tissue, including immature anthers. Plant Physiology 106: 25±36. Cornish EC, Pettitt JM, Bonig I, Clarke AE. 1987. Developmentally controlled expression of a gene associated with self-incompatibility in Nicotiana alata. Nature 326: 99±102. Dana MN, Ascher PD. 1986. Sexually localized expression of pseudoself compatibility (PSC) in Petunia hybrida Hort. I. Pollen inactivation. Theoretical and Applied Genetics 71: 573±577. Dodds PN, Bonig I, Du H, Rodin J, Anderson MA, Newbigin E, Clarke AE. 1993. S-RNase gene of Nicotiana alata is expressed in developing pollen. The Plant Cell 5: 1771±1782. Dodds PN, Ferguson C, Clarke AE, Newbigin E. 1999. Pollen-expressed S-RNases are not involved in self-incompatibility in Lycopersicon peruvianum. Sexual Plant Reproduction 12: 76±87. Entani T, Iwano M, Shiba H, Takayama S, Fukui K, Isogai A. 1999. Centromeric localisation of an S-RNase in Petunia hybrida Vilm. Theoretical and Applied Genetics 99: 391±397. Gebhardt C and 11 others. 1991. RFLP maps of potato and their alignment with the homoeologous tomato genome. Theoretical and Applied Genetics 83: 49±57. Golz JF, Clarke AE, Newbigin E. 2000. Mutational approaches to the study of self-incompatibility: Revisiting pollen-part mutants. Annals of Botany 85: 95±103. Golz JF, Clarke AE, Newbigin E, Anderson MA. 1998. A relic S-RNase is expressed in the styles of self-compatible Nicotiana sylvestris. The Plant Journal 16: 591±600. Harbord RM, Napoli CA, Robbins TP. 2000. Segregation distortion of T-DNA markers linked to the self-incompatibility (S) locus in Petunia hybrida. Genetics (in press). ten Hoopen R, Harbord RM, Maes T, Nanninga N, Robbins TP. 1998. The self-incompatibility (S) locus in Petunia hybrida is located on

112

Robbins et al.ÐSelf-incompatibility in Petunia

chromosome III in a region, syntenic for the Solanaceae. The Plant Journal 16: 729±734. ten Hoopen R, Robbins TP, Fransz PF, Montijn BM, Oud O, Gerats AGM, Nanninga N. 1996. Localisation of T-DNA insertions in Petunia hybrida by ¯uorescent in-situ hybridisation: physical evidence for a suppression of recombination. The Plant Cell 8: 823±830. Huang S, Lee H-S, Karunanandaa B, Kao T-h. 1994. Ribonuclease activity of Petunia in¯ata S proteins is essential for rejection of self-pollen. The Plant Cell 6: 1021±1028. Ioerger TR, Clark AG, Kao T-h. 1990. Polymorphism at the selfincompatibility locus in Solanaceae predates speciation. Proceedings of the National Academy of Sciences (USA) 87: 9732±9735. Ioerger TR, Gohlke JR, Xu B, Kao T-h. 1991. Primary structural features of the self-incompatibility protein in the Solanaceae. Sexual Plant Reproduction 4: 81±87. Kamboj RK, Jackson JF. 1986. Self-incompatibility alleles control a low molecular weight, basic protein in pistils of Petunia hybrida. Theoretical and Applied Genetics 71: 815±819. Kao T-h, McCubbin AG. 1996. How ¯owering plants discriminate between self and non-self pollen to prevent inbreeding. Proceedings of the National Academy of Sciences (USA) 93: 12059±12065. Lee H-S, Huang S, Kao T-h. 1994. S-proteins control rejection of selfincompatible pollen in Petunia in¯ata. Nature 397: 560±563. Lee H-S, Singh A, Kao T-h. 1992. RNase X2, a pistil-speci®c ribonuclease from Petunia in¯ata, shares sequence similarity with solanaceous S proteins. Plant Molecular Biology 20: 1131±1141. Linskens HF, Straub J. 1978. A mutant collection of Petunia hybrida. Incompatibility Newsletter 10: 123±131. McClure BA, Haring V, Ebert PR, Anderson MA, Simpson RJ, Sakiyama F, Clarke AE. 1989. Style self-incompatibility gene products of Nicotiana alata are ribonucleases. Nature 342: 955±957.

de Nettancourt D. 1977. Incompatibility in angiosperms. In: Frankel R, Gall GAE, Linskens HF, eds. Monographs on theoretical and applied genetics. Berlin: Springer-Verlag, 48. Peltier D, Farcy E, Dulieu H, Berville A. 1994. Origin, distribution and mapping of RAPD markers from wild type Petunia species in Petunia hybrida Hort. lines. Theoretical and Applied Genetics 88: 637±645. Richman AD, Broothaerts W, Kohn JR. 1997. Self-incompatibility RNases from three plant families: homology or convergence?. American Journal of Botany 84: 912±917. Royo J, Kunz C, Kowyama Y, Anderson M, Clarke AE, Newbigin E. 1994. Loss of a histidine residue at the active site of S-ribonuclease is associated with self-compatibility in Lycopersicon peruvianum. Proceedings of the National Academy of Sciences (USA) 91: 6511±6514. Sink KC. 1984. Taxonomy. In: Sink KC, ed. Petunia. Monographs in theoretical and applied genetics. New York: Springer Verlag, 3±9. Smith FJ, Oud JL, de Jong JH. 1973. A standard karyogram of Petunia hybrida Hort. Genetica 44: 474±484. Tanksley SD, Loaiza-Figueroa F. 1985. Gametophytic self-incompatibility is controlled by a single major locus on chromosome 1 in Lycopersicon peruvianum. Proceedings of the National Academy of Sciences (USA) 82: 5093±5096. Tsukamoto T, Ando T, Kokubun H, Watanabe H, Masada M, Zhu X, Marchesi E, Kao T-h. 1999. Breakdown of self-incompatibility in a natural population of Petunia axillaris (Solanaceae) in Uruguay containing both self-incompatible and self-compatible plants. Sexual Plant Reproduction 12: 6±13. Xue Y, Carpenter R, Dickinson HG, Coen ES. 1996. Origin of allelic diversity in Antirrhinum S locus RNases. The Plant Cell 8: 805±814.