INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Naturally occurring or induced (e.g. by radiation) self-compatible mutants of self-incompatible species in the Solanaceae have provided useful materials for studying the mechanism of the S-RNase-mediated self-incompatibility (SI) system. For example, molecular genetic studies of a self-compatible line of Lycopersicon peruvianum have suggested that the RNase activity of S-RNases is an integral part of the SI mechanism because mutations in the S-RNase gene rendering its protein product without the RNase activity cause the breakdown of SI (Kowyama et al., 1994a; Royo et al., 1994). Moreover, cytological and molecular genetic studies of x-ray irradiation-induced self-compatible mutants have uncovered a phenomenon, called competitive interaction, where duplication of the S-locus region renders pollen grains carrying two different pollen S-alleles unable to function in SI (de Nettancourt, 1977; Golz et al., 2000). Studies of self-compatible mutants have also led to the identification of loci unlinked to the S-locus that are required for the function of the S-RNase gene or the yet unidentified pollen S-gene. For example, mutations in the HT gene, which encodes a small Asn-rich protein (McClure et al., 1999), are partially responsible for the breakdown of SI in a cultivar of Lycopersicon esculentum (Kondo et al., 2002).

We have examined previously the SI status of all 19 natural taxa of the genus Petunia (Solanaceae; Tsukamoto et al., 1998), including three allopatric infraspecific taxa of the species Petunia axillaris (Lam.) Britton, Sterns & Poggenb., a historical parent of garden petunias (Sink, 1984). P. axillaris subsp. axillaris is self-incompatible, but subsp. parodii (Steere) Cabrera and subsp. subandina T. Ando (a recently described subspecies; Ando, 1996) are self-compatible. In Uruguay, subsp. axillaris and subsp. parodii show disjunctive distribution (Ando et al., 1994). From our survey of more than 100 natural populations of P. axillaris in Uruguay, we found that all populations of subsp. parodii were distributed in northern Uruguay and were entirely composed of self-compatible individuals. In contrast, all natural populations of subsp. axillaris were distributed in southern Uruguay and most of them were composed of only self-incompatible individuals. Some populations of subsp. axillaris contained both self-incompatible and self-compatible individuals (Ando et al., 1998), and they are referred to as "mixed populations."

The mixed populations we have identified offer a good opportunity for studying the causes for the breakdown of SI. We previously analyzed the SI behavior of plants derived from a mixed population, designated U1, of subsp. axillaris, and showed that the breakdown of SI in the self-compatible individuals was caused by defects in the pistil function of an S-haplotype. Specifically, S₁₃-RNase, the S₁₃-allele product of the S-RNase gene, was absent. We have termed this phenomenon "stylar-part suppression (sps)" of an S-haplotype (Tsukamoto et al., 1999). We further showed that the failure to produce S₁₃-RNase in the self-compatible individuals was not caused by deletion of the S₁₃-RNase gene or by mutations in the promoter of the S₁₃-RNase gene, but rather by a modifier locus that suppressed the expression of the S₁₃-RNase gene (Tsukamoto et al., 2003).

Here, we report the study of another mixed population, designated U83, of subsp. axillaris. We identified two S-haplotypes, designated S_C1 and S_C2, that were associated with self-compatibility, and determined that S_C2-haplotype was a mutant form of S₁₇-haplotype we had identified previously in the U1 population, with the mutation affecting only the pollen function.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

SI Status and S-Genotypes of 50 Plants of the U83 Population

We raised 50 plants (designated U83-1 to U83-50) from seeds collected from the U83 population and determined their SI status. Thirteen plants did not set any capsules upon self-pollination at the mature flower stage, whereas the other 37 plants set full-sized capsules, each containing approximately 900 seeds. For the former 13 plants, the growth of virtually all pollen tubes stopped in the upper third segment of their styles (results not shown), a typical response of solanaceous type SI. For the latter 37 plants, most of the pollen tubes reached the basal part of the style (results not shown), suggesting that their pistils failed to reject self-pollen. The ploidy level for all these 37 self-compatible plants was determined by flow cytometry as described by Tsukamoto et al. (1999), and all the plants were found to be diploid. This result ruled out polyploidy as a cause of the breakdown of SI in these self-compatible plants.

The S-genotypes of these 50 plants were determined as described in "Materials and Methods," and the results are shown in Table I. A total of 14 different S-haplotypes were identified, and interestingly, two of them (S₄ and S₁₂) had previously been identified in the U1 population (Tsukamoto et al., 1999). The 12 "new" S-haplotypes were designated S₁₉ through S₂₈, S_C1, and S_C2. The haplotypes that occurred most frequently and second most frequently in these 50 plants were S₂₁ (found in 12 plants) and S₁₉ (found in 10 plants). S-homozygotes for all the 12 new haplotypes were generated by bud-selfing of appropriate U83 plants.

Stylar RNases Produced by Self-Compatible Plants and Their Progeny

The total stylar protein of each of the 37 self-compatible plants was analyzed by isoelectric focusing (IEF). Twenty-five plants produced an RNase (designated S_C1-RNase) with a pI of 10.0; the results of representative plants are shown in Figure 1, A and C. The other 12 self-compatible plants produced an RNase (designated S_C2-RNase) with a pI of 10.3; the results of representative plants are shown in Figure 1, B and D. Some S-RNases had a similar pI as S_C1- or S_C2-RNase. For example, U83-45 produced a single intense RNase band, even though it carried S₂₈- and S_C1-haplotypes. It is interesting to note that none of the 37 self-compatible plants derived from the U83 population produced both S_C1- and S_C2-RNases, even though we were able to generate such plants by crossing an S_C1S_C1 plant with an S_C2S_C2 plant. As expected, all the progeny of this cross were self-compatible and produced both S_C1- and S_C2-RNases.

View larger version (69K): [in this window] [in a new window]   Figure 1.   IEF of representative self-compatible plants derived from the U83 population. Gels were subjected to Coomassie Brilliant Blue staining (A and B) or RNase activity staining (C and D). The positions of SC1-RNase and SC2-RNase are indicated with arrows. The positions of pI markers are indicated to the right of the gels.

From the self-compatible plants that produced S_C1-RNase, we chose U83-1, U83-6, U83-12, and U83-28 for bud-selfing, and examined the segregation of S-genotypes in each bud-selfed progeny. The genotype of U83-1 was S₁₉S_C1 (Table I), and it produced both S₁₉-RNase (pI 10.3) and S_C1-RNase (Fig. 2, A and C). The S-genotypes of 62 bud-selfed progeny plants of U83-1 were determined as described in "Materials and Methods": eight were S₁₉S₁₉ (self-incompatible, rejecting S₁₉ pollen but accepting S_C1 pollen), 33 were S₁₉S_C1 (self-compatible, rejecting S₁₉ pollen but accepting S_C1 pollen), and 21 were S_C1S_C1 (self-compatible, accepting both S₁₉ and S_C1 pollen). For each of these three S-genotypes, total stylar protein was extracted from three plants and subjected to IEF. In all cases, the presence of S_C1-RNase or S₁₉-RNase in the style correlated with the plants carrying S_C1- or S₁₉-haplotype (Fig. 2, A and C), so these two RNases were the products of their respective S-haplotypes. SDS-PAGE analysis showed that the molecular masses of S_C1- and S₁₉-RNases were approximately 26.5 and 31.0 kD, respectively (results not shown), comparable with those of other S-RNases reported. Furthermore, all of the 40 progeny plants obtained from bud self-pollination of U83-1-3 (S_C1S_C1), a bud-selfed progeny plant of U83-1, were self-compatible. U83-6, U83-12, and U83-28 only produced the S_C1-RNase, and all of the 30 bud-selfed progeny plants obtained from each plant were self-compatible and produced only the S_C1-RNase (results not shown), consistent with the S-genotype of U83-6, U83-12, and U83-28 being S_C1S_C1.

View larger version (82K): [in this window] [in a new window]   Figure 2.   IEF of stylar RNases of U83-1, U83-2, and selected plants from their bud-selfed progeny. Gels were subjected to Coomassie Brilliant Blue staining (A and B) or RNase activity staining (C and D). The positions of SC1-RNase and SC2-RNase are indicated with arrows. The positions of pI markers are indicated to the right of the gel.

For the self-compatible plants producing S_C2-RNase, we chose U83-2, U83-25, and U83-26 for the segregation study. U83-2 produced two RNases, S₂₀-RNase (pI 10.2) and S_C2-RNase (Fig. 2, B and D). Among the 62 bud-selfed progeny plants of U83-2 examined, six were S₂₀S₂₀ (self-incompatible, rejecting S₂₀ pollen but accepting S_C2 pollen), 27 were S₂₀S_C2 (self-compatible, rejecting S₂₀ pollen but accepting S_C2 pollen), and 29 were S_C2S_C2 (self-compatible, accepting both S₂₀ and S_C2 pollen). IEF was carried out for the bud-selfed progeny of U83-2, and the presence of S_C2-RNase and S₂₀-RNase correlated with the presence of S_C2-haplotype and S₂₀-haplotype, respectively (Fig. 2, B and D). SDS-PAGE analysis showed that the molecular masses of S_C2- and S₂₀-RNases were approximately 26.0 and 29.0 kD, respectively (results not shown). As expected, all of the 40 progeny plants obtained from bud self-pollination of U83-2-4 (S_C2S_C2), a plant from the bud-selfed progeny of U83-2, were self-compatible. U83-25 and U83-26 plants produced only the S_C2-RNase band, and the 30 bud-selfed progeny plants obtained from each plant were all self-compatible and produced only S_C2-RNase (results not shown), consistent with the S-genotype of U83-25 and U83-26 being S_C2S_C2.

Reciprocal Crosses of S_C1- and S_C2-Homozygotes with 40 Tester S-Homozygotes

We chose U83-1-3 (S_C1S_C1) and U83-2-4 (S_C2S_C2) from the bud-selfed progeny of U83-1 and U83-2, respectively, to examine the genetic behavior of the S_C1- and S_C2-haplotypes. These two homozygotes were reciprocally crossed with 40 S-homozygotes, S₁S₁ through S₄₀S₄₀, of subsp. axillaris. U83-1-3 (S_C1S_C1) set full-sized capsules when pollinated with pollen from any of the 40 S-homozygotes, and reciprocal crosses also set full-sized capsules. For U83-2-4 (S_C2S_C2), all 40 S-homozygotes set full-sized capsules when pollinated with its pollen, and in reciprocal crosses, pollen from all but the S₁₇S₁₇ homozygote (U1-20-16) set full-sized capsules. That is, the S_C2S_C2 plant rejected S₁₇ pollen, but its pollen was not rejected by the S₁₇S₁₇ homozygote. Interestingly, S₁₇-haplotype was previously identified from another natural population, U1 (Tsukamoto et al., 1999), and it was not carried by any of the 50 plants of the U83 population examined (Table I).

Growth of pollen tubes was examined 48 h after pollination of S_C2S_C2 (U83-2-4). When S_C2S_C2 was self-pollinated, a large number of pollen tubes reached the basal part of the style (Fig. 3, A and B). In contrast, upon self-pollination of the self-incompatible S₁₇S₁₇ plant (U1-20-16), the growth of pollen tubes was arrested in the upper third segment of the style (results not shown). When the pistils of S_C2S_C2 were pollinated with pollen from the S₁₇S₁₇ plant, the growth of pollen tubes was arrested in the upper third segment of the style (Fig. 3, C and D). In the reciprocal cross using the S₁₇S₁₇ plant as the pistil parent and S_C2S_C2 as the pollen parent, a large number of pollen tubes were able to reach the basal part of the style (Fig. 3, E and F). To confirm that the S₁₇ pollen tubes grew normally in compatible crosses, we used pollen from the S₁₇S₁₇ plant to pollinate an S₁S₁ plant (U1-1-18). The results showed that all the S₁₇ pollen tubes were able to grow to the basal part of the pistil of the S₁S₁ plant (Fig. 3, G and H).

View larger version (73K): [in this window] [in a new window]   Figure 3.   Observation by fluorescent microscopy of pollen tube growth in the stigmatic zone (B, D, F, and H) and in the basal region (A, C, E, and G) of the style 48 h after pollination.

Properties of S_C2- and S₁₇-RNases

We next compared the stylar products of S_C2-haplotype and S₁₇-haplotype, i.e., S_C2-RNase and S₁₇-RNase. Total stylar protein of S_C2S_C2 (U83-2-4) and S₁₇S₁₇ (U1-20-16) plants were separated by Mono-S cation-exchange column chromatography. Both elution profiles contained one prominent peak at the same position (Fig. 4, A and B), typical of elution profiles of stylar proteins of self-incompatible Petunia spp. (Lee et al., 1994; Tsukamoto et al., 1999). To confirm that the common peak was S_C2-RNase/S₁₇-RNase, total stylar protein of S₂₀S₂₀ (U83-2-9) and S₂₀S_C2 (U83-2) was separately chromatographed (Fig. 4, C and D). Comparison of these profiles resulted in the positive assignment of the S_C2-RNase/S₁₇-RNase peak. We carried out RNase activity assays using the peak fractions of S_C2-RNase (Fig. 4A) and S₁₇-RNase (Fig. 4B) and found that the specific activity of S_C2-RNase (130 A₂₆₀ units min¹ mg¹) was almost identical to that of S₁₇-RNase (134 A₂₆₀ units min¹ mg¹).

View larger version (12K): [in this window] [in a new window]   Figure 4.   Cation-exchange chromatography profiles of style extracts from self-compatible SC2SC2 (A) and S20SC2 (D) plants and self-incompatible S17S17 (B) and S20S20 (C) plants.

Isolation of Full-Length cDNA Clones Encoding S_C1-, S_C2-, and S₁₇-RNases

We first used RT-PCR to amplify partial-length cDNAs for S_C1-, S_C2-, and S₁₇-RNases from poly(A⁺) RNA isolated from styles of U83-1-3 (S_C1S_C1), U83-2-4 (S_C2S_C2), and U1-20-16 (S₁₇S₁₇), respectively. The primers used corresponded to two of the conserved regions of solanaceous S-RNases, C2 (NFTIHGLWP), and C5 (ELNEIGICF; Tsai et al., 1992; Tsukamoto et al., 2003). An approximately 430-bp fragment of the expected size was obtained from all three plants. This fragment was purified and cloned for each S-RNase. Three separate cDNA libraries were constructed for these three plants using poly(A⁺) RNA isolated from stylar tissues collected 1 d before anthesis. Screening of the cDNA libraries using the cloned RT-PCR product obtained from each corresponding plant as a probe resulted in the isolation of three, six, and five positive clones from the S_C1S_C1, S_C2S_C2, and S₁₇S₁₇ cDNA libraries, respectively. The sequences of the longest clones for S_C1-RNase (GenBank accession no. AY180048), S_C2-RNase (GenBank accession no. AY180049), and S₁₇-RNase (GenBank accession no. AY180050) were determined. Except for 2 additional bp in the 5' non-coding region, the nucleotide sequence of the S_C2-RNase cDNA was completely identical to that of the S₁₇-RNase cDNA.

The deduced amino acid sequences of S_C1-RNase and S₁₇-RNase/S_C2-RNase contained 221 and 220 residues, respectively. S_C1-RNase was most similar to S₁-RNase of P. axillaris subsp. axillaris, sharing 53.2% sequence identity (Tsukamoto et al., 2003), and S₁₇-RNase was most similar with S_X-RNase of Petunia hybrida Vilm., sharing 68.0% sequence identity (Ai et al., 1991).

Genetic Behavior of S_C1- and S_C2-Haplotypes

The SI behavior of S_C1- and S_C2-haplotypes was further examined in plants carrying the S₁₃-haplotype. This haplotype was chosen because it was identified from a different population (U1), allowing us to determine whether S_C1 and S_C2 behaved similarly in a different genetic background. Moreover, S₁₃-RNase had a lower pI than S_C1- and S_C2-RNases, so it could be easily distinguishable from the latter two by IEF. We reciprocally crossed U83-1-3 (S_C1S_C1), a plant from the bud-selfed progeny of U83-1, and U83-2-4 (S_C2S_C2), a plant from the bud-selfed progeny of U83-2, with U1-11-14 (S₁₃S₁₃). We analyzed 10 progeny (F₁ generation) of each cross, and found all the F₁ plants to be self-compatible. All the F₁ plants of the reciprocal crosses between S_C1S_C1 and S₁₃S₁₃ accepted S₁₇ pollen but rejected S₁₃ pollen. All the F₁ plants of the reciprocal crosses between S_C2S_C2 and S₁₃S₁₃ rejected both S₁₃ and S₁₇ pollen.

Next, we bud self-pollinated selected plants from each of the four F₁ generations mentioned above and examined segregation of the S-haplotypes in their bud-selfed progenies (F₂ generations). For S_C1-haplotype, two F₁ plants from S_C1S_C1 × S₁₃S₁₃ and two from the reciprocal cross were chosen, and a total of 231 individuals of the F₂ generations of these four F₁ plants were analyzed (Table II). The segregation ratio for each F₂ generation was expected to be 1 (S₁₃S₁₃):2 (S₁₃S_C1):1 (S_C1S_C1). However, the S-genotypes of all except four plants were either S₁₃S_C1 (self-compatible, rejecting S₁₃ pollen) or S_C1S_C1 (self-compatible, accepting S₁₃ pollen), with more S_C1S_C1 individuals than S₁₃S_C1 individuals in all the F₂ generations. The four S₁₃S₁₃ plants (self-incompatible, rejecting S₁₃ pollen) were obtained only in two of the four F₂ generations. One possibility for this large deviation from the expected segregation ratio is that not all S₁₃ pollen tubes were able to escape rejection by the pistil during bud self-pollination. The immature buds (of 20 mm in length) of the F₁ plants from S_C1S_C1 × S₁₃S₁₃ and S₁₃S₁₃ × S_C1S_C1 used for bud self-pollination were found by IEF to produce a small amount of S₁₃-RNase in the style (results not shown). It should be noted that self-incompatible S₁₃S₁₃ progeny was never obtained when the S₁₃S_C1 plants were self-pollinated at the mature flower stage.

For the S_C2-haplotype, three F₁ plants from the progeny of S_C2S_C2 × S₁₃S₁₃ and three from its reciprocal cross were chosen for analysis of their bud-selfed progenies (F₂ generations). A total of 339 individuals in the F₂ generations were analyzed (Table III). The S-genotypes of all but 10 plants were either S₁₃S_C2 (self-compatible, rejecting S₁₃ and S₁₇ pollen) or S_C2S_C2 (self-compatible, rejecting S₁₇ pollen, accepting S₁₃ pollen). The reason described above might explain the much lower than expected numbers of S₁₃S₁₃ plants (self-incompatible, rejecting S₁₃, accepting S₁₇ pollen) obtained, as well as the excess of S_C2S_C2 plants over S₁₃S_C2 plants.

All the genetic results obtained from the plants carrying S_C1 or S_C2 described in this section and earlier showed that the breakdown of SI in their self-compatible progeny cosegregated with S_C1 or S_C2, and that the pistil function of S_C2 was normal and behaved as did S₁₇. Thus, the defect in S_C2 lay in the pollen function.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

In this work, we studied the SI behavior of 50 plants derived from a mixed population, U83, of P. axillaris subsp. axillaris, with the goal of determining the cause of the breakdown of SI in the self-compatible individuals. We identified a total of 14 different S-haplotypes carried by these plants. The presence of this number of S-haplotypes in a small number of plants is typical of the highly polymorphic S-locus of all the SI systems studied so far. For example, 16 different S-haplotypes were identified from 41 plants of a natural population of Ipomoea trifida (Kowyama et al., 1994b), and 30 S-haplotypes were identified from 36 plants of a natural population of Papaver rhoeas (Lawrence and O'Donnell, 1981). In a previous study, we identified 18 different S-haplotypes from 33 plants derived from another mixed population, U1, of subsp. axillaris (Tsukamoto et al., 1999).

Interestingly, two of the S-haplotypes identified in the U83 population, S₄ and S₁₂, have been identified previously in the U1 population located approximately 100 km from the U83 population (Tsukamoto et al., 1999). However, it is not unusual to find genetically identical S-haplotypes in different natural populations. For example, Kowyama et al. (1994b) identified one S-haplotype of I. trifida from all six natural populations they examined (four in Mexico, one in Guatemala, and one in Colombia).

One major reason for studying mixed populations was to determine the causes of the breakdown of SI in naturally occurring self-compatible "mutants." Some mutant S-haplotypes are likely to co-exist with their progenitor wild-type haplotypes in a natural population, making it possible to identify the cause of the breakdown of SI by genetic and molecular studies of self-compatible mutants and their wild-type progenitors. Analysis of self-incompatible individuals of L. peruvianum identified from the same locale as a self-compatible line that produced an S-RNase lacking the RNase activity have led to the identification of the progenitor haplotype of the self-compatible haplotype (Bernatzky et al., 1995). We have identified 13 mixed populations from a survey of 73 natural populations of subsp. axillaris (Ando et al., 1998). Our previous study of one mixed population, U1, has led to the identification of a mutant haplotype, designated S₁₃^sps, and its progenitor wild-type haplotype, S₁₃. The cause of the mutant phenotype was attributed to a modifier locus that suppressed the expression of the S₁₃-RNase gene in the style (Tsukamoto et al., 1999, 2003).

Here, we found that 37 of the 50 plants derived from the U83 population were self-compatible and that all the self-compatible plants carried one of two S-haplotypes, S_C1 and S_C2. The results of the genetic crosses of S_C1 and S_C2 homozygotes with all the functional S-genotypes we have identified in U1, U67, and U83 populations showed that none of the 12 functional S-haplotypes of the U83 population was the wild-type form of either S_C1-haplotype or S_C2-haplotype. The pollination results and the observation of pollen tube growth (Fig. 3) suggest that the S_C2-haplotype is genetically identical to S₁₇-haplotype except that it functions normally in the pistil but fails to function in the pollen. Interestingly, the S₁₇-haplotype was identified from the U1 population (Tsukamoto et al., 1999).

That S_C2-haplotype is defective in the pollen function of S₁₇-haplotype was further confirmed by the characterization of S_C2-RNase and S₁₇-RNase and subsequent cloning and sequencing of their cDNAs. The sequences of these two S-RNases are completely identical, and both have normal RNase activity, which is required for rejection of self-pollen during the SI response (Huang et al., 1994).

The molecular nature of the mutation that affects the pollen function of S₁₇-haplotype remains to be determined. Many pollen function mutants, either naturally occurring or induced by radiation, have been shown to be caused by duplication of the entire S-locus or the part of the S-locus that contains the pollen S-allele (de Nettancourt, 1977; Golz et al., 1999, 2001). The duplicated region often exists as a centric fragment but has been found translocated to a different chromosomal region or linked to the S-locus of another S-haplotype through unequal crossover (Golz et al., 2001). It is thought that pollen grains that contain two different S-alleles fail to function in SI interactions due to competitive interaction. For example, SI breaks down in an S₁S₂ plant with duplicated pollen S₂-allele. If the duplicated region exists as a centric fragment, this plant will produce four different genotypes of pollen, S₁, S₂, S₁dS₂, and S₂dS₂ (with "d" standing for "duplicated," as defined by Golz et al., 1999), but only S₁dS₂ pollen fails to function in SI interactions. Thus, self-compatible S₁S₁dS₂ and S₁S₂dS₂ progeny will be produced upon self-pollination at the mature flower stage. Bud-selfing, however, will also produce self-incompatible S₁S₁, S₁S₂, S₂S₂, and S₂S₂dS₂ progeny. Similar results from ordinary selfing and bud-selfing will be obtained if the duplicated region is translocated to a chromosomal region unlinked to the S-locus. If the duplicated region is linked to the S₁-locus, this plant will produce S₁-dS₂ (with the hyphen indicating linkage between the S₁-locus and the duplicated S₂-locus region) and S₂ pollen, and ordinary selfing will produce self-compatible S₁-dS₂S₂ and S₁-dS₂S₁-dS₂ progeny. Bud-selfing will also produce the self-incompatible S₂S₂ genotype.

All the results of bud-selfing we carried out in this work are consistent with the S_C2-haplotype containing a duplicated S-locus region of unknown S-haplotype linked to the S₁₇-locus. For example, bud-selfing of U83-2 (S₂₀S_C2) yielded self-incompatible S₂₀S₂₀, self-compatible S_C2S_C2, and self-compatible S₂₀S_C2, but did not result in self-incompatible S₁₇S₁₇ (i.e. S_C2S_C2 without the duplicated S-locus region) or self-incompatible S₂₀S₁₇, as would be expected if the duplicated S-locus region exists as a centric fragment or is translocated to a different chromosomal region. Thus, if the breakdown of SI in S_C2-haplotype is caused by competitive interaction, duplication of the S-locus might have occurred in a plant carrying the S₁₇-haplotype and some other S-haplotype, with the duplicated region of this other S-haplotype linked to the S₁₇-locus. The genotype of U83-2 could be denoted as S₂₀S₁₇-dS_n, and the genotype of its self-compatible S_C2S_C2 progeny would be S₁₇-dS_nS₁₇-dS_n.

The loss of pollen function could also be caused by mutation in the pollen S-gene, resulting in either complete absence of its protein product or production of a defective protein product. In the case of S_C2, the mutation would specifically affect the pollen function of S₁₇-haplotype. For example, the genotype of an S₂₀S_C2 plant could be denoted S₂₀S₁₇^ppm, with S₁₇^ppm indicating pollen-part mutation of the pollen S₁₇-gene. Bud-selfing will produce self-incompatible S₂₀S₂₀ progeny and self-compatible S₂₀S₁₇^ppm and S₁₇^ppmS₁₇^ppm progeny. The results of all the genetic crosses involving the S_C2-haplotype can be explained by the existence of S₁₇^ppm. However, it has been suggested that mutations in the pollen S-gene would be lethal to the pollen (Golz et al., 2001). A prevailing model explaining how S-haplotype-specific inhibition of pollen tubes can be achieved predicts that the pollen S-allele product contains an S-allele-specific domain and an RNase inhibitor domain and that the latter would interact with the RNase activity domain of all non-self S-RNases but not that of the self S-RNase (McCubbin and Kao, 2000). This would allow the self S-RNases, but not non-self S-RNases, to degrade pollen tube RNAs, leading to growth inhibition of self-pollen tubes. If this inhibitor model is correct, a pollen tube carrying a defective pollen S-allele would be unable to inhibit the RNase activity of any S-RNase and would suffer growth arrest in any style.

Luu et al. (2001) have proposed a modified inhibitor model that predicts that the pollen S-allele product only contains the S-allele specificity domain and that the RNase-inhibitor domain resides on a general RNase inhibitor. The RNase inhibitor would bind to all S-RNases except when an S-RNase is bound by its cognate pollen S-allele product. In the absence of a functional pollen S-allele product, the RNase activity of all S-RNases would be inhibited. Thus, contrary to the prediction of the simple inhibitor model, pollen carrying a defective pollen S-allele would not be lethal because it would be accepted by the pistil of any S-genotype.

Mutation in a modifier locus required for pollen function could also cause the breakdown of SI in S_C2. If this is the case, the modifier locus would have to be tightly linked to the S-locus of the S₁₇-haplotype. Otherwise, self-incompatible plants possessing the S₁₇-haplotype would have been obtained in the bud-selfed (F₂) progeny shown in Table III, had this modifier locus segregated away from the S-locus of the S₁₇-haplotype. The existence of such a modifier locus for the pollen function is possible; however, as mentioned above, any mutation that affects the function of a pollen S-allele would be lethal to pollen if the simple inhibitor model of SI interactions is valid.

If the S-locus-linked markers previously identified in P. inflata (McCubbin et al., 2000) are also linked to the S-locus of P. axillaris, they will be useful for determining whether the S_C2-haplotype contains duplication of the S-locus region of some other S-haplotype. Ultimately, identification of the pollen S-gene will reveal whether the defect in the pollen function of S_C2 lies in the mutation in the pollen S₁₇-gene.

Although all the properties of the S_C1-haplotype determined in the present study resemble those of the S_C2-haplotype, we could not conclude whether this haplotype is defective in the pistil function and/or pollen function. This is because we did not find the wild-type form of S_C1 among the 40 S-haplotypes we have identified from the U1, U67, and U83 populations. It may be worthwhile searching for the wild-type form of the S_C1-haplotype in the vast grassy plains of South America, particularly in Uruguay and Argentina, where P. axillaris widely occurs (Ando, 1996).

For the 13 mixed populations of P. axillaris we have identified, 10 contained mostly self-incompatible plants (Ando et al., 1998). For example, only 9% to 10% of the plants of the U1 population examined were self-compatible (Ando et al., 1998; Tsukamoto et al., 1999). The U83 population studied here was an exception: 61% (Ando et al., 1998) to 74% (Table I) of the plants studied were self-compatible. Breakdown of SI in all the self-compatible plants of the U1 population we studied was found to be caused by suppression of the stylar function of the S₁₃-haplotype (Tsukamoto et al., 2003). However, the breakdown of SI in the self-compatible plants of the U83 population was caused by a defect in the pollen function of the S₁₇-haplotype and another unknown mechanism associated with the S_C1-haplotype. Multiple causes for the breakdown of SI may be a reason why the U83 population contained more self-compatible than self-incompatible individuals.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Material

Plants of Petunia axillaris subsp. axillaris derived from three different populations in Uruguay were used in the present study. They were U1 (located at 0.3 km east of La Paz along a side road leading to Camino Uruguay Road, Montevideo), U67 (located near Route 15, 10.1 km southeast of Rocha to La Paloma, Rocha Department), and U83 (located near Route 60, between Minas and Pan de Azúcar, Lavalleja Department). A total of 42 S-homozygotes were used: S₁S₁ to S₁₈S₁₈ were identified from the U1 population (Tsukamoto et al., 1999); S₁₉S₁₉ to S₂₈S₂₈, S_C1S_C1, and S_C2S_C2 were identified from the U83 population in this work (see below); and S₂₉S₂₉ to S₄₀S₄₀ were identified from the U67 population (T. Tsukamoto, T. Ando, T. Omori, H. Watanabe, H. Kokubun, E. Marchesi, and T.-h. Kao, unpublished data).

Examination of SI Behavior

The SI status of each plant was determined by the presence or absence of capsules after self-pollination at the mature flower stage. For bud self-pollination, several flower buds of 20 mm in length were emasculated and pollinated with fresh pollen collected from the same individual. For crosses, all flowers of the plants serving as females were carefully emasculated 1 d before anthesis and then pollinated with fresh pollen grains carrying selected S-haplotypes. The extent of pollen tube growth in the style was monitored 48 h after self-pollination using fluorescence microscopy after the style had been stained with decolorized aniline blue (Martin, 1959).

Analysis of Total Stylar Protein

IEF analysis of total protein extracted from the style (including stigma) was performed according to the method of Sassa et al. (1992). The methods for extracting total stylar protein, cation-exchange chromatography, and RNase activity assay were described by Tsukamoto et al. (1999). Stylar protein extracts were also analyzed by 12% (w/v) SDS-polyacrylamide gels (Laemmli, 1970).

Determination of S-Genotypes

Fifty individuals, designated U83-1 to U83-50, were raised from seeds collected from the U83 population. The procedure used in determining the S-genotypes is described below, using U83-1 as an example. U83-1 was bud self-pollinated to circumvent SI, and the SI status was examined for at least 30 plants of its bud-selfed progeny. One of the self-incompatible plants of its bud-selfed progeny was tentatively assigned S_AS_A genotype, because its styles produced only one S-RNase as revealed by IEF. All the other individuals of the U83 population were then pollinated with pollen from the S_AS_A plant to determine whether they carried the S_A-haplotype. U83-1 turned out to be self-compatible because self-pollination of mature flowers also yielded large-sized capsules. Among the self-compatible plants of the bud-selfed progeny of U83-1, those whose styles produced a different S-RNase than S_A-RNase and whose styles accepted pollen from the S_AS_A plant were assigned S_C1S_C1; and those whose styles produced both S_A- and S_C1-RNases and rejected pollen from the S_AS_A plant were assigned S_AS_C1. IEF was used to determine whether the remaining self-compatible plants from the U83 population contained S_C1-RNase.

When a self-compatible plant (e.g. U83-2) was found not to contain S_C1-RNase, the method described above for U83-1 was used to determine the S-genotypes of its bud-selfed progeny and identify self-compatible homozygotes. Using U83-2 as an example, the S-genotype of self-compatible plants of its bud-selfed progeny that produced only one S-RNase band on IEF was assigned S_C2S_C2. The other S-haplotype carried by U83-2, which segregated to self-incompatible plants of its bud-selfed progeny, was tentatively assigned S_B because plants homozygous for this haplotype accepted pollen from the S_AS_A homozygote. Pollen from the S_BS_B homozygote was used to pollinate U83-3 through U83-50 to determine whether they carried the S_B-haplotype. The above-mentioned process was repeated until the tentative S-genotypes of all U83 plants were determined.

To assign S-genotypes to the 50 U83 plants, each of the self-incompatible S-homozygotes (e.g. the S_AS_A homozygote) obtained as described above was separately pollinated with pollen of the S₁S₁ through S₁₈S₁₈ homozygotes from the U1 population (Tsukamoto et al., 1999). If pollen of all the tester S-homozygotes was accepted by an S-homozygote derived from the U83 population, the S-haplotype carried by this S-homozygote was assigned S_n, with n being the number 19 or higher. In all, S₁₉ to S₂₈ were assigned to these S-homozygotes. In a separate study, 14 different S-homozygotes were identified from another mixed population, U67, and test crosses using pollen of S₁S₁ through S₂₈S₂₈ homozygotes showed that two of these S-haplotypes (S₁₃ and S₂₅) were present in the U67 population. Thus, the other 12 S-homozygotes were assigned S₂₉ to S₄₀.

Isolation of Total RNA from Styles

Total RNA was extracted from styles of flower buds 1 d before anthesis using the method described by Tsukamoto et al. (2003).

RT-PCR

The conditions for RT-PCR (including the primers used) and cloning of PCR products were as described by Tsukamoto et al. (2003). cDNAs were sequenced using the Thermo Sequenase Cycle Sequencing Kit (Amersham, Piscataway, NJ) and analyzed on an A.L.F. sequencer (Pharmacia, Piscataway, NJ).

Construction of cDNA Libraries and Isolation and Sequence Analysis of cDNA Clones

cDNA libraries were constructed from 10 µg of poly(A⁺) RNA isolated from styles of S_C1S_C1, S_C2S_C2, and S₁₇S₁₇ plants as described by Tsukamoto et al. (2003). Screening of the libraries using PCR-derived DNA fragments as probes, and sequencing of cDNA clones were carried out as described by Tsukamoto et al. (2003).