- © 1999 American Society of Plant Physiologists
Abstract
Gametophytic self-incompatibility in plants involves rejection of pollen when pistil and pollen share the same allele at the S locus. This locus is highly multiallelic, but the mechanism by which new functional S alleles are generated in nature has not been determined and remains one of the most intriguing conceptual barriers to a full understanding of selfincompatibility. The S11 and S13 RNases of Solanum chacoense differ by only 10 amino acids, but they are phenotypically distinct (i.e., they reject either S11 or S13 pollen, respectively). These RNases are thus ideally suited for a dissection of the elements involved in recognition specificity. We have previously found that the modification of four amino acid residues in the S11 RNase to match those in the S13 RNase was sufficient to completely replace the S11 phenotype with the S13 phenotype. We now show that an S11 RNase in which only three amino acid residues were modified to match those in the S13 RNase displays the unprecedented property of dual specificity (i.e., the simultaneous rejection of both S11 and S13 pollen). Thus, S12S14 plants expressing this hybrid S RNase rejected S11, S12, S13, and S14 pollen yet allowed S15 pollen to pass freely. Surprisingly, only a single base pair differs between the dual-specific S allele and a monospecific S13 allele. Dual-specific S RNases represent a previously unsuspected category of S alleles. We propose that dualspecific alleles play a critical role in establishing novel S alleles, because the plants harboring them could maintain their old recognition phenotype while acquiring a new one.
INTRODUCTION
Among the cell–cell recognition phenomena present in living organisms, self-incompatibility (SI) plays a major evolutionary role because it constitutes an important mechanism for preventing inbreeding. SI is present in hermaphroditic animals such as tunicates (Grosberg, 1988), in fungi (Kronstad and Leong, 1990), and in many Angiosperm families (de Nettancourt, 1977). In the most widespread type of SI, gametophytic SI, the genotype of the haploid pollen determines its own incompatibility phenotype. For the Solanaceae, the gametophytic SI phenotype is specified by a highly multiallelic S locus (de Nettancourt, 1977, 1997) whose only known product is a ribonuclease (S RNase; McClure et al., 1989) expressed in the transmitting tissue of the style (Anderson et al., 1986). Gain-of-function experiments have shown that expression of an S RNase transgene is sufficient to alter the SI phenotype of the pistil but not that of the pollen (Lee et al., 1994; Murfett et al., 1994; Matton et al., 1997), and thus the identity of the pollen S gene (unknown to date) is likely to be different from that of the S RNase (Kao and McCubbin, 1997). RNase activity, although essential for expression of the SI phenotype (Huang et al., 1994), seems not to be involved in the specificity of the cell–cell recognition phenomenon. In closely related S RNases, such specificity has been shown to depend on the amino acid sequence at the two hypervariable regions (HVa and HVb) (Matton et al., 1997), whereas in distantly related S RNases, a role for amino acids located elsewhere in the molecule (in addition to the hypervariable regions) has been suggested (Kao and McCubbin, 1997; Zurek et al., 1997).
Multiallelism at the S locus is impressive in both sporophytic and gametophytic SI systems (de Nettancourt, 1997). In the sporophytic system, both point mutations and intragenic recombination appear to have contributed to S allele polymorphism (Kusaba et al., 1997; Nasrallah, 1997). In the gametophytic system, several studies (Clark and Kao, 1991; Coleman and Kao, 1992; Saba-El-Leil et al., 1994; Matton et al., 1995, 1997; Ishimizu et al., 1998) have suggested that point mutations rather than intragenic recombination (Fisher, 1961; Pandey, 1970; Ebert et al., 1989) are the primary source of S allele polymorphism. Phenotypically distinct yet highly similar S RNase sequences have been described in Solanum chacoense (Saba-El-Leil et al., 1994) and Pyrus pyriflora (Ishimizu et al., 1998).
Studies of population dynamics predict that newly generated S alleles will spread in the population because of their reproductive advantage, until their frequency approaches that of other alleles. Several mechanisms to explain how new S alleles could be generated in nature have been proposed (de Nettancourt, 1977, 1997; Charlesworth, 1995). The major difficulty with these hypotheses is that the generation of new S allele specificities, either in the pollen or in the style alone, would inevitably result in the breakdown of the SI system. Only if appropriate changes in the pollen and stylar sequences occur concurrently can the SI phenotype be maintained.
By using site-directed mutagenesis to dissect the elements required for determining allelic specificity, we have produced a mutant that bears on the question of new S allele generation. Our experimental system, the tuber-bearing wild potato species S. chacoense, involves two phenotypically distinct S alleles (S11 and S13) whose mature proteins differ by only 10 amino acids (Saba-El-Leil et al., 1994), four of which are located in the hypervariable regions (Figure 1B). We have shown that the substitution of all four of these amino acids in the hypervariable regions of an S11 RNase with those of an S13 RNase fully converts the S11 phenotype into an S13 phenotype (HVab S RNase; Matton et al., 1997). Here, we show that altering only three of these four amino acids (HVapb S RNase) can result in a new S allele with the unexpected property of dual-specificity incompatibility, or DSI (i.e., capable of rejecting simultaneously two phenotypically and genotypically distinct pollen types). DSI alleles, if produced in nature, could represent an essential step in the generation of new S alleles from preexisting ones in a mechanism that would prevent any breakdown of the SI system.
Although never before observed in the context of SI, dual specificity has been observed in other plant cell–cell recognition phenomena. For example, the disease resistance RPM1 gene of Arabidopsis confers resistance to Pseudomonas syringae, expressing either the avrRpm1 avirulence gene or the unrelated avrB pathogen signals (Grant et al., 1995). Furthermore, the polygalacturonase-inhibiting pgip-2 gene of bean encodes a protein that interacts with and specifically inhibits the endopolygalacturonases from both Aspergillus niger and Fusarium moniliforme (Leckie et al., 1999). Dual specificity may be more common in cell–cell recognition systems than currently thought, and if so, it may play an important role in the evolution of new recognition specificities in general.
RESULTS
The HVapb Transgene Is Derived from a Partial Domain Swap between the S11 and S13 Alleles
We have previously shown that replacing the four amino acids that differ between the HVa and HVb regions of S11 and S13 RNases resulted in complete conversion of the S11 rejection phenotype into that of S13 (the HVab allele; Figure 1B) (Matton et al., 1997). Using a style-specific tomato chitinase promoter (Harikrishna et al., 1996) to drive expression of a hybrid S11 allele in vivo, we produced an S RNase (the HVapb RNase; Figure 1B) in which three of these four amino acids in the HVa and the HVb regions of the S11 RNase were replaced with those of the S13 RNase. Thus, the HVapb RNase differs in only one amino acid from the previously reported HVab RNase, which displays an exclusively S13 SI phenotype (Figure 1B).
Genetic Crosses Show That the HVapb Transgene Rejects Both S11 and S13 Pollen
The breeding behavior of G4 plants (S12S14 genotype) transformed with the HVapb transgene was fully assessed by genetic crosses repeated over two consecutive years using pollen from suitable tester stocks. In these crosses, an incompatible reaction is expected between the endogenous S12 and S14 RNases present in the styles of the transformed plants and S12 and S14 pollen when produced by G4 plants (whether transformed or not). Any other pollen types (S11 or S13, for example) should experience no stylar barrier to fertilization and produce seeded fruits after pollination unless an additional incompatibility barrier is generated by expression of the transgene. The absence of fruits is a sensitive indicator of pollen rejection, because fruits can be set when even a few pollen tubes reach the ovary.
Structure of the HVapb Transgene Construct.
(A) The chimeric transgene contains the entire coding sequence of a mutated S11 gene downstream from a style-specific tomato chitinase promoter (Harikrishna et al., 1996). The hatched box corresponds to the chitinase promoter and includes the chitinase 5′; untranslated region (UTR). The black box corresponds to the S11 intron.
(B) The amino acid sequences in the hypervariable regions HVa and HVb of the S11 wild-type allele are shown with the residues targeted for site-directed mutagenesis in boldface. Dots represent unchanged amino acids. The residues differing in the hypervariable regions of the S13 wild type (wt) and the previously described HVab mutant (Matton et al., 1997) are shown for comparison along with their associated phenotypes.
We found that seven individuals from a total of 33 different plants transformed with the HVapb transgene (21% of the transgenic plants) fully rejected S11, S12, S13, and S14 pollen. This high number of plants with a new and complete SI phenotype is similar to that observed with other S transgenes in this species (27 to 36%; Matton et al., 1997) and is believed to be due to the strong chitinase promoter used in the gene constructs. The genetic crosses (Table 1) showed that these seven plants had acquired the ability to reject two phenotypically distinct pollens (S11 and S13), despite being transformed with only one type of S transgene. This was most clearly seen in the simultaneous rejection of pollen produced from an S11S13 genotype in addition to pollen from plants containing either S11 or S13 in different genetic backgrounds (Table 1). The transgenic plants behaved normally with respect to their endogenous SI phenotype (S12S14), as shown by complete lack of fruit set after self-pollinations or crosses with pollen from the untransformed S12S14 genotype.
The possibility that the transgenic plants were female sterile was excluded by the observations that fruits were set after pollinations with (1) compatible 2n pollen from G4 tetraploids (data not shown) and (2) haploid pollen from an S14S15 individual (Table 1). These later crosses also ruled out the possibility that the transgenic plants could reject pollen with any S allele constitution. We conclude from these genetic data that the HVapb RNase is dual specific because it recognizes and rejects the phenotypically distinct S11 and S13 pollen yet does not block unrelated pollen types. These seven plants thus acquired a full DSI phenotype. The remaining plants were scored as either partially or fully compatible because they let pass S11 and S13 pollen to varying degrees (Table 1). As occurs with fully DSI plants, all retained their endogenous S12S14 phenotype.
To determine how the DSI phenotype was related to expression of the transgene, we first ranked transformed plants on the basis of HVapb RNA levels by tissue printing (Cappadocia et al., 1993). A selection of these tissue prints is shown in Figure 2. Typically, five different styles from each plant were printed onto nylon membranes and hybridized first to the S13 cDNA probe (Figures 2A to 2C, left) and then to an 18S rRNA probe as a control for RNA levels (right). The tissue prints are arranged so that those with the highest and lowest levels of HVapb RNA are at the top of Figure 2A and at the bottom of Figure 2C, respectively, whereas controls for the specificity of the S13 cDNA probe are shown at left in Figure 2D. The pollen rejection phenotype (i.e., the number of fruits set per number of pollinations performed) is shown for both S11 and S13 pollen beneath the tissue prints for each transgenic plant. As shown, the full DSI phenotype (full rejection of both S11 and S13 pollen) was associated with those plants expressing the highest levels of transgene RNA, whereas low levels of transgene expression are associated with a partially or fully compatible pollen rejection phenotype. Interestingly, some individuals with intermediate levels of transgene expression fully rejected S13 pollen but only partially rejected S11 pollen (e.g., plants T-11, T-21, T-6, and T-1). This is not surprising, given that the HVapb S RNase sequence differs from the S11 sequence by three amino acids and from the HVab transgene (an S13 rejection phenotype) by only one amino acid (Figure 1B). The additional 12 plants lacking a detectable phenotype (Table 1) also had no detectable RNA expression in these tissue prints (data not shown).
A Dual-Specificity Pollen Rejection Phenotype Is Revealed by Genetic Crosses
Molecular Characterization of the DSI Plants
To confirm that the acquisition of DSI was due to a single S RNase whose characteristics were those expected based on the HVapb gene sequence, we subjected five of the plants with a complete DSI phenotype (T-16, T-18, T-20, T-30, and T-31) to a detailed molecular analysis. One transgenic plant that had not acquired a new rejection phenotype (T-19) was included as a negative control. DNA gel blots using an S13 cDNA probe confirmed that all transformants contained at least one copy of the transgene and that all transgenics were derived from independent transformation events (Figure 3). Control hybridizations with plants of known genotype confirmed that the probe hybridized readily to both S11 and S13 alleles. The transgene expression as assessed by RNA gel blot analyses (Figure 4A) confirmed that plants that had acquired the DSI phenotype expressed an mRNA whose length (1 kb) and amounts were similar to those observed in wild-type S11S12 or S13S14 plants. Hybridization with a constitutively expressed β-amylase probe as control (Figure 4B) showed that all samples contained similar amounts of RNA. Only a single size class of mRNA hybridized to the S13 cDNA probe in the transgenic plants.
HVapb Transgene Expression and Breeding Behavior for Selected Plants.
(A) For each transgenic plant T-9, T-20, T-16, T-18, and T-31, four to five styles were squashed on a charged nylon membrane, hybridized with an S13 cDNA probe, and exposed to film for 7 days (blots at left for each plant). All hybridizations and exposures were performed at the same time for all plants. The membranes were then stripped, hybridized with an 18S rRNA probe, and exposed to film overnight (blots at right for each plant). The pollen rejection phenotypes (number of fruits set/number of pollinations) with either S11 (left) or S13 (right) pollen are shown immediately below the tissue prints for each plant to allow direct comparisons with the levels of transgene expression. No fruits set after pollination represents a full incompatible phenotype.
(B) Transgenic plants T-11, T-21, T-6, T-1, and T-24 were treated as described in (A).
(C) Transgenic plants T-25, T-14, T-10, T-33, and T-19 were treated as described in (A).
(D) Nine styles from either S11S12 (top) or S12S14 (bottom) genotypes were squashed on a charged nylon membrane, hybridized to an S13 cDNA probe (left), and exposed to film at the same time as for the transgenic plants. The membranes were then stripped and hybridized to an 18S rRNA probe (right).
Protein extracts from the styles of plants expressing the DSI phenotype were analyzed by gel electrophoresis. First, two-dimensional gel electrophoresis (Figure 5) showed that the transgenic plants T-16, T-20, T-30, and T-31 contained a protein found neither in the untransformed host plant (S12S14) nor in the control plant T-19. This new protein had an apparent molecular weight identical to that previously observed for transgenic plants expressing an authentic S11 transgene (Matton et al., 1997) but a slightly more acidic pI, consistent with the arginine-to-leucine substitution in the HVb region (Figure 1). To confirm that the new protein was similar in sequence to the S11 RNase, we probed protein gel blots of stylar extracts with an antibody raised against a 15amino acid peptide corresponding to the S11 HVa region. This antibody specifically recognizes the S11 RNase, because the S13 RNase did not cross-react (Figure 6A, left). The antibody did cross-react, however, with a protein in the stylar extracts of transformed plants that had acquired the DSI phenotype (Figure 6A). No reaction was observed with control plant T-19, despite a protein load similar to that in other lanes (Figure 6B). The amount of HVapb RNase is comparable in plants that acquired the DSI phenotype; plant T-30, which appeared to contain less immunoreactive protein, also contained less total protein, as shown by Ponceau red staining (Figure 6B). However, these protein blots cannot be used to compare levels of the HVapb RNase with the S11 RNase because these two proteins have a different sequence in the region of antibody binding (the HVa region; Figure 1B).
The immunoreactive protein generally began to accumulate in styles ∼2 days before flower opening (Figure 6C), similar to the pattern of S2 and S3 RNase expression reported in this species (Xu et al., 1990). The accumulation of the S RNase can also be observed by general protein stains (Figure 6D). Because unopened flower buds can be self-pollinated and set fruit, the ability of the unopened buds of transgenic plants to reject S11 pollen was also tested at 3 and 2 days before bud opening. When styles of plant T-20 were observed by UV microscopy 30 hr after pollination, buds pollinated 3 days before opening showed numerous pollen tubes (>50) in the ovarian region, whereas buds pollinated 2 days before opening showed few (<5) pollen tubes (data not shown). The time at which these latter flower buds were examined thus correlated with the time detectable levels of the HVapb RNase were observed (1 day before bud opening). The presence of immunologically detectable levels of the HVapb RNase is thus a prerequisite for pollen rejection. By the time of flower opening (day 0), plant T-20 was fully incompatible with both S11 and S13 pollen (no pollen tubes reached the ovary), and the pollen rejection phenotype was microscopically identical to that of plant T-16 described below.
Transgenic Plants Contain from One to Four Copies of the Transgene.
Ten-microgram samples of genomic DNA were digested with EcoRI, electrophoresed on agarose gels, and transferred to nylon membranes. After hybridization with an S13 cDNA probe, films were exposed for 3 days. The S13 cDNA probe hybridizes with both S11 and S13 alleles in untransformed plants (leftmost three lanes) but does not recognize the S12 or S14 alleles in the host plant. Numbers at left show the position of molecular weight standards (in kilobases).
DSI Transgenic Plants Express Wild-Type Levels of mRNA.
(A) Ten micrograms of total stylar RNA was electrophoresed on agarose gels containing formaldehyde and transferred to nylon membranes. The membranes were hybridized with the S13 cDNA probe and exposed to film for 2 hr. The hybridization signal from five of the six transformed plants carrying the HVapb transgene was similar to S RNase transcript levels in the untransformed plants (Matton et al., 1997) (left lanes). The S12S14 genotype (G4) is the transformation host. The size of the hybridizing mRNA is 1 kb.
(B) The membrane used in (A) was rehybridized with a β-amylase gene probe and exposed to film for 3 hr to show that similar amounts of RNA were present in all samples.
Lastly, when stylar protein extracts were tested for RNase activity by in-gel RNase assays, a new RNase activity whose electrophoretic mobility was identical to that of the authentic S11 RNase (expressed by transgenic plant T-64 S11) (Matton et al., 1997) was observed (Figure 7, arrows). This activity was absent in the untransformed (S12S14) host plant or in plant T-19. This demonstrates that the transgene product has RNase activity. Taken together, we conclude from these data that the acquisition of the DSI pollen rejection phenotype is due to the expression of a single active S RNase whose molecular characteristics are those expected from the HVapb sequence.
Simultaneous S11 and S13 Pollen Tube Growth Arrest in Styles of DSI Plants
As a final confirmation of the DSI phenotype, the styles of HVapb-expressing individuals were examined by UV microscopy 48 hr after pollination with pollen from an S11S13 individual. Both S11 and S13 pollen tubes, observed in the midstyle of plant T-16 (Figure 8C, upper right panel), had thickened callose walls, were few in number, and did not penetrate to the ovarian region (Figure 8C, bottom right). These characteristics are identical to those of a fully incompatible cross, such as when S12 and S14 pollen are observed in the styles of untransformed G4 plants (Figure 8A, right panels). This SI response to incompatible pollen was not observed when pollen from an S11S13 individual grew in the styles of untransformed (G4) plants (Figure 8A, left panels) or in the styles of plants not expressing the transgene (plant T-19; Figure 8C, left panels). In these fully compatible crosses, pollen tubes passed freely through the styles and entered the ovary.
Two-Dimensional Gels Show That a New Protein Is Present in DSI Transgenic Plants.
Fifty micrograms of stylar protein was electrophoresed on twodimensional gels and visualized by Coomassie Brilliant Blue R 250 staining. The spot labeled HVapb is not found in the untransformed host (S12S14) or in plants without a detectable phenotype (T-19). This new protein has the same molecular mass as the S11 RNase but a slightly more acidic pI, as expected based on the sequence modifications.
DISCUSSION
Plants with the Full DSI Phenotype Express the HVapb Transgene
The DSI phenotype is the ability to completely reject two phenotypically distinct pollens (S11 and S13) and was observed in seven of the 33 transgenic plants produced. This high number of plants acquiring a new SI phenotype is in good agreement with our previous results and is believed to result from the strong chitinase promoter used in the constructs (Matton et al., 1997). The conclusion that the new phenotype results from expression of the transgene is based on (1) expression of HVapb RNA in plants that have acquired the DSI phenotype (Figures 2 and 4), (2) expression of S RNase in plants that have acquired the DSI phenotype (Figures 5, 6A, and 7), (3) generally increased RNA levels in plants with higher levels of pollen rejection (Figure 2), and (4) an increase in the pollen rejection phenotype with increasing HVapb RNase accumulation (Figure 6C).
DSI Transgenic Plant Style Extracts Cross-React with an Anti-S11 RNase Antibody.
(A) Twenty micrograms of stylar extract prepared from buds at the time of flower opening (day 0) was electrophoresed on an SDSpolyacrylamide gel, transferred to nitrocellulose, and challenged with the anti-S11 antibody. Flowers were obtained from untransformed plants with known S allele constitutions, a transgenic plant lacking an observable phenotype (T-19), or DSI transgenic plants (T-16, T-18, T-20, T-30, and T-31). Each day 0 sample was taken from a time-series gel similar to that shown in (C). To ensure equal development times, all gels contained samples from S11S12 and S12S14 plants as positive and negative controls, respectively. The DSI transgenic plants contain a protein similar in size to the authentic S11 RNase, which reacts with the antibody, although HVapb RNase levels cannot be directly compared with those of the S11 RNase due to the differences in their amino acid sequences in the region of antibody binding. The size of the HVapb RNase (26 kD) is shown at left.
(B) Ponceau-S staining of the nitrocellulose in (A) before immunoblotting.
(C) Twenty micrograms of stylar extract was prepared from buds harvested from plant T-20 at the indicated times (in days) of flower development (day 0 represents flower opening), electrophoresed, and subjected to immunoblotting as given in (A). Films were exposed approximately twice as long as those shown in (A) to visualize the low levels of protein in unopened flowers.
(D) Ponceau-S staining of the nitrocellulose in (C) before immunoblotting.
DSI Transgenic Plants Contain a New RNase Activity.
In-gel RNase activity assays using 50 μg of stylar protein (Yen and Green, 1991) show a band of 26 kD (arrow) present in extracts from transgenic plants expressing either an authentic S11 (T-64-S11) or the HVapb transgene (T-16, T-18, T-20, T-30, and T-31). This band is absent in the untransformed G4 genotype and in plants without detectable transgene transcript (T-19).
The full DSI phenotype is typically associated with expression of the transgene at a level similar to endogenous S RNases for both RNA and protein. However, we find there are exceptions to this general rule. For example, plant T-20 has average levels of HVapb RNase yet lower than average levels of RNA (cf. Figures 4 and 6). Furthermore, plant T-18 has low levels of protein despite high levels of RNA (cf. Figures 4 and 6). These discrepancies may be related to differences in the stability of transgene products at both mRNA and protein levels in the independent transgenic lines, because our measurements reflect accumulated quantities (i.e., neither the synthesis nor the degradation rates of the S proteins or the S mRNAs are known). Furthermore, because transgene expression is developmentally regulated (Figure 6C), differences in sampling time or even in the timing of transgene expression could produce differences in the accumulated levels of mRNA or protein. Also, other factors, such as the effects of pollination (perhaps not negligible, given that the wound-inducible chitinase promoter drives our transgene) or temperature, may contribute to variations in the levels of molecular correlates to the phenotype.
Single Amino Acid Changes Can Alter the SI Phenotype
We have previously shown that full conversion of the S11 into the S13 pollen rejection phenotype was obtained by substituting the four residues in the hypervariable regions that differed between the S11 and S13 RNases (Matton et al., 1997). Here, we demonstrate that partial replacement of these four key amino acids can lead to the production of a functional S RNase with a DSI phenotype. A careful examination of the amino acids altered in the HVapb RNase (Figure 1), which rejects both S11 and S13 pollen, leads to two important conclusions concerning the relationship between S RNase sequence and the resulting incompatibility phenotype. First, a total of seven amino acid differences (six of which lie outside the hypervariable regions) separate the sequence of the HVapb RNase from the wild-type S13 RNase. Because both of these S RNases reject S13 pollen (Figure 1), this represents the largest number of amino acid changes described to date that can be accommodated by an S RNase without elimination of its pollen rejection capability. Second, when the sequence of the HVapb RNase is compared with that of the previously produced HVab RNase (Figure 1), only a single amino acid is found to differ between the two. This amino acid substitution (K→N), due to a single base pair transversion (A→C), is the only difference between the previously described S13-specific HVab mutant (which does not reject S11 pollen; Matton et al., 1997) and the HVapb mutant (which does) (Figure 1). This thus represents the smallest difference between two S RNase sequences with a different pollen rejection phenotype and provides convincing experimental support for the argument that single point mutations can generate new S allele specificities (Clark and Kao, 1991; Saba-El-Leil et al., 1994).
DSI Transgenic Plants Display a Typical Pollen Rejection Phenotype with Two Distinct Pollen Types.
(A) and (C) Squashes of midstyle (top) and stylar bases (bottom) of either untransformed G4 or transformed plants T-19 or T-16 were stained with aniline blue 48 hr after pollination (Martin, 1959) with pollen from either an S11S13 or an S12S14 tester stock. All pollen tube growth was arrested in plants expressing the HVapb transgene (T-16) but not in plants lacking detectable transgene expression (T-19). Controls show a fully compatible pollination of untransformed G4 plants with S11 and S13 pollen (left) and an incompatible pollination of G4 plants with self-S12 and self-S14 pollen.
(B) A schematic view of the pistil illustrates which regions of the styles were examined for the presence of pollen tubes.
The recognition characteristics of our antibody, directed against the HVa region of the S11 RNase (an anti-S11 antibody), also support the hypothesis that single amino acid changes can alter S RNase recognition. The anti-S11 antibody recognizes both S11 and HVapb RNases but does not recognize the S13 RNase. It is thus reasonable that the pollen component of the S11 allele, capable of binding an S11 RNase with an affinity similar to that of the anti-S11 antibody, should also be able to discriminate between the HVapb and the S13 RNases (which differ only in a single amino acid in the HVa region). We also note that the recognition of both S11 and HVapb RNases by our antibody is in agreement with the intermediate nature of the HVapb RNase. Because the anti-S11 antibody has a surface complementary to the S11 RNase, its behavior thus mimics in vitro the expected in vivo interaction between the pollen S component and its corresponding S RNase. Clearly, these specific reactions imply that the hypervariable regions of S RNases lie exposed at the surface of the protein, as proposed by recent threedimensional models of RNase structure (Parry et al., 1998). As a caveat to any sweeping generalization that might be drawn from these observations, however, we note that the substitution of asparagine by lysine (which eliminates antibody binding as well as S11 pollen recognition) replaces an uncharged residue with a larger, positively charged residue. It remains to be determined whether less drastic single amino acid substitutions would also be sufficient to alter the SI phenotype and antibody binding.
A single amino acid change has recently been shown to alter the recognition specificity of the polygalacturonase inhibiting protein (PGIP) of bean. PGIP is encoded by a gene family, and the proteins produced by different family members recognize and inhibit polygalacturonases from different fungi with different specificities. For example, PGIP-1 inhibits polygalacturonases from A. niger, whereas PGIP-2 inhibits polygalacturonases from both A. niger and F. moniliforme (Leckie et al., 1999). The sequences of the monospecific PGIP-1 and the dual-specific PGIP-2, which differ by only eight amino acids, are analogous to our wild-type S13 and HVapb sequences. Interestingly, when the glutamine at position 253 of the PGIP-2 sequence is mutated to a lysine, the dual specificity of PGIP-2 is lost. Conversely, the monospecific PGIP-1 can be converted to dual specificity by replacement of the lysine at position 253 by glutamine. This behavior is clearly similar to our results comparing HVab with HVapb RNases, in which replacement of a charged lysine residue with a nonpolar amine produces a dual-specific recognition phenotype.
Dual-Specific S Alleles May Play a Key Role in Evolution
We believe that dual-specific alleles may play a pivotal evolutionary role in the generation of new S alleles, one of the last conceptual barriers to a full understanding of SI (the other being the identity of the pollen component of the SI system). In essence, the problem of S allele generation lies in the contrasting observations that whereas even small populations can contain many different S alleles (de Nettancourt, 1997), all attempts to generate S alleles with a new specificity by classic mutagenesis have failed and have resulted only in the production of nonfunctional alleles (de Nettancourt, 1977). Clearly, mutations in an S allele sequence are more likely to result in breakdown of SI than in a new SI phenotype.
How then do new specificities arise? New S phenotypes have been reported after extensive inbreeding (de Nettancourt, 1977), but unfortunately no molecular information is available about their nature. It has been suggested that the generation of new allelic specificities might proceed by a series of sequential mutagenic steps difficult to monitor because intermediates might require sophisticated analyses for their detection (Lewis, 1951; Fisher, 1961). Indeed, a dual-specificity allele could represent just such an intermediate, and in this study, availability of S11 and S13 pollen was necessary for its detection. The lack of appropriate genetic systems to detect DSI could thus explain why dual specificities have not been previously observed in SI systems. A further disadvantage is that unlike other systems in which both partners to cell–cell recognition are known, the nature of the pollen component to SI remains elusive.
In our hypothetical model for generating new S phenotypes, we propose that the transition from one allelic specificity to another occurs when stylar and pollen parts evolve step by step (Figure 9). A dual-specific allele is crucial to this scheme because it allows a new recognition specificity to evolve in the plant without it losing its original incompatibility phenotype. Based on both previous work (Matton et al., 1997) and this work, we envisage the following order of three key mutational steps. First, a stylar RNase (S11) with specificity for P11 pollen mutates and evolves the capacity (like the dual-specific HVapb mutant described here) to recognize and reject a type of pollen (P13) not yet present in the population. The SI system does not break down because the dual-specific allele retains the ability to reject P11. Second, the pollen component P11 also evolves by point mutations. The new specificity that will eventually arise (P 13) would not be recognizable by S11 but does not result in SI breakdown because P13, although phenotypically distinct from P11, is still recognized by the dual-specific allele. Finally, the stylar dual-specific component mutates further to a stage (equivalent to the HVab mutant) at which the recognition of P11 is finally lost but the recognition of P13 is retained. At this stage, a new S allele, consisting of a new male and female pair, has been generated. The intermediate in this process has never lost its SI phenotype, although the specificity of the recognition reaction has been altered.
We do not know the nature of the positive selective pressure that operates to maintain stylar and pollen components as functional pairs, but its effects appear similar to the gene-for-gene coevolution proposed for host-parasite relationships (Thompson and Burdon, 1992). There is clearly a need to invoke such a positive selective pressure because nonfunctional alleles do not seem to accumulate in natural populations (Campbell and Lawrence, 1981; Richman and Kohn, 1996; see also references in de Nettancourt, 1977). Our results reconcile such a positive selective pressure with the creation of a new and distinct stylar-pollen pair (whose components should not, by definition, recognize the original components from which they have evolved) by providing a means of maintaining a functional SI system while new S alleles are generated. Our interpretation suggests that the selective force will be different from inbreeding depression, which would act after mutation in one of the components if this mutation produced a nonfunctional stylar-pollen pair. In our view, it seems advantageous for any positive selective pressure to act before generation of a nonfunctional allele, because nonfunctional alleles would experience no barrier to propagation through the population.
Hypothetical Model of New S Allele Generation via Point Mutations.
Diagram depicts the evolution from a wild-type S11 RNase to a wildtype S13 RNase. The solid arrows represent the amino acid replacements (numbers shown in parentheses) that have either been introduced into the hypervariable regions by experimental site-directed mutagenesis or occurred outside the recognition site. The dotted lines represent recognition of specific tester pollen as experimentally determined by genetic analysis. The open arrow represents hypothetical mutations that must take place within the recognition site of the pollen component. The transitional dual-specific allele (HVapb) contains the maximum number of amino acid changes still allowing recognition of P11. wt, wild type.
The model as presented here makes no assumptions about the nature of the pollen component to SI. It also makes no predictions concerning the identity or mode of action of the pollen component (because the consequences of a “membrane receptor” or a “cytosolic inhibitor” are irrelevant to the model). The strength of the model lies in the elimination of simultaneity during the production of matching pairs of stylar and pollen components. This is vital given that the probability of simultaneous mutations arising in two different genes is far lower than the probability of a mutation arising in a single gene. Our model, derived from our experimental data on S RNases, places the stylar component as the DSI allele that first reaches out toward a new specificity. However, it is equally likely that a similar process could happen in the pollen (i.e., that the pollen component could evolve into a dual-specific allele) because the stylar and pollen parts are genetically linked (i.e., a dual-specific allele in either part must carry its partner through any genetic cross). It will be of great interest to compare the sequences and recognition characteristics of pollen components with the S11 and S13 alleles when the pollen components of gametophytic SI are at last identified.
METHODS
Plant Material
The diploid (2n = 2x = 24) Solanum chacoense self-incompatible genotypes used in this study include line PI458314 (which carries the S11 and S12 alleles), line PI230582 (which carries the S13 and S14 alleles), their F1 hybrid G4 (which carries the S12 and S14 alleles and which was selected for its high in vitro regenerability [Van Sint Jan et al., 1996]), another F1 hybrid (which carries the S11 and S13 alleles), and line PI458316 (which carries the S14 and S15 alleles). The parental lines were obtained from the Potato Introduction Station (Sturgeon Bay, WI).
Transgenic Plants and Mutagenesis
A HindIII-NcoI fragment (hatched box in Figure 1A) containing the promoter and the 5′; untranslated region of the chitinase gene was ligated to a genomic S11 RNase gene engineered to have an NcoI site at the ATG translation initiation codon of the S11 preprotein. The S11 gene fragment also contains a small (87 bp) intron (black box in Figure 1A) between the hypervariable regions HVa and HVb and 0.8 kb of the 3′; untranslated region. Mutants were obtained through sitedirected mutagenesis (Altered Sites II in vitro mutagenesis system; Promega, Madison, WI) using the following mutagenic oligonucleotides (mutated nucleotides are underlined): HVa, 5′;-GCCAAAACTTAATTATAAATTTTTCAGTGT-3′;; HVb, 5′;-GCTTCTGCTCTAAAGGACCAACC-3′;; and HVap, 5′;-GCCAAAACTTAATTATAACTTTTTCAGTGT-3′;. A mutagenic oligonucleotide containing the NcoI site (5′;TGCACCATGGTTAAATCACTGCTTAC-3′;) was used in conjunction with an oligonucleotide complementary to the 3′; end of the gene (5′;GAATTCAAGGACATACATTTG-3′;) to polymerase chain reaction amplify the S11 1.7-kb fragment (Pwo DNA polymerase; Roche Diagnostics, Quebec, Canada). The 3.1-kb HindIII-EcoRI chimeras were cloned in the transformation vector pBIN19 (Clontech, Palo Alto, CA) and introduced in Agrobacterium tumefaciens LBA4404 by electroporation. Plants were transformed by the leaf disc method using a highly regenerable S. chacoense genotype (called G4) carrying the S12 and S14 alleles (Van Sint Jan et al., 1996). Each construction was sequenced before and after transformation in Agrobacterium to ensure the integrity of the NcoI junction and of the mutated region.
Molecular Analyses
Total RNA was prepared as described (Jones et al., 1985). RNA gel blot analyses (Sambrook et al., 1989) were performed with 10 μg of total RNA, as determined spectrophotometrically, and an S13 cDNA probe (Matton et al., 1997). DNA was extracted by following an established procedure (Reiter et al., 1992). DNA gel blot analyses were performed as described (Sambrook et al., 1989) with 10 μg of genomic DNA per sample digested with EcoRI and probed with the S13 cDNA. Because there is only one EcoRI site in the transgene constructs, the number of bands on the DNA gel blot corresponds to the copy number of the transgene. Two-dimensional gel electrophoresis, in-gel RNase activity assays, and tissue prints were performed as described previously (Matton et al., 1997).
Antibodies and Immunochemistry
For protein gel blots, a polyclonal antibody specifically directed against the S11 RNase was produced by immunization of a rabbit (Cocalico Biological Inc., Reamstown, PA) with the synthetic peptide KPKLTYNYFSDKMLN (corresponding to the S11 RNase HVa region) on a branched multiple antigen peptide (Research Genetics, Huntsville, PA). Protein samples separated by SDS-PAGE and transferred to nitrocellulose were stained with 2% (w/v) Ponceau S to check the uniformity of sample loading. The membrane was then blocked by incubation in Tris-buffered saline containing 1.5% (w/v) BSA (fraction V; Sigma), 2.5% (w/v) instant skim milk powder, and 0.05% (v/v) Tween 80 for 1 hr at room temperature, stained with a 1:1000 dilution of the polyclonal anti-S11 antibody for 1 hr at room temperature, rinsed, and visualized with a 1:5000 dilution of a goat anti-rabbit horseradish peroxidase conjugate (Promega) and the Renaissance Western Blot Chemiluminescence Reagent kit (New England Nuclear Life Science Products, Boston, MA), according to the manufacturer's instructions.
Genetic Crosses
The genetic analyses were performed by using pollen from tester stocks with known S allele constitution. Crosses were scored as fully compatible when almost every pollination resulted in fruit set, whereas they were considered fully incompatible when pollinations never resulted in fruit formation. Pollen tube growth in the styles was monitored using the aniline blue staining method (Martin, 1959), as described previously (Matton et al., 1997).
Acknowledgments
We thank Sylvain Lebeurrier for plant acclimation and care and Drs. Bruce McClure and Charles Gasser for graciously providing the chitinase promoter. This work was supported by a fellowship from the Jean-Walter Zellidja foundation to D.T.L. and by research grants from the National Science and Engineering Research Council Canada and the Fonds pour la Formation des Chercheurs et l'Aide ågrave; la Recherche Québec.
- Received June 7, 1999.
- Accepted August 30, 1999.
- Published November 1, 1999.
REFERENCES
