- American Society of Plant Biologists
Abstract
The coordinate evolution of self-incompatibility (SI) and stigma-anther separation, two mechanisms that promote cross-pollination in plants, has been a long-standing puzzle in evolution and development. Using a transgenic self-incompatible Arabidopsis thaliana model, we performed screens for mutants exhibiting a modified SI response. A mutation in the RNA-dependent RNA polymerase RDR6, which functions in trans-acting short interfering RNA (ta-siRNA) production, was found that simultaneously enhances SI and causes stigma exsertion, without associated increases in SRK transcript levels. While rdr6 mutants had been previously shown to exhibit stochastic stigma exsertion, our results demonstrate that the S-locus receptor kinase (SRK) gene further enhances pistil elongation and stigma exsertion in this mutant background, a process that requires SRK catalytic activity and correlates with SRK transcript levels. These results suggest that positive regulators or effectors of SI and pistil development are regulated by ta-siRNA(s). By establishing complex connections between SI and stigma exsertion through the sharing of a ta-siRNA–mediated regulatory pathway and the dual role of SRK in SI and pistil development, our study provides a molecular explanation for the coordinate evolution of these processes.
INTRODUCTION
Plants with perfect flowers have evolved a variety of reproductive traits that promote cross-fertilization. Among the best-known physiological outbreeding devices are genetic self-incompatibility (SI) systems that prevent self-fertilization (Takayama and Isogai, 2005; Rea and Nasrallah, 2008). Although SI ensures that pollen cannot fertilize ovules on the same plant, it is often accompanied by additional adaptations in floral architecture that increase the chance of cross-pollen receipt, such as increased separation in stigma and anther heights that causes stigmas to protrude above the anthers, a trait known as stigma exsertion. By contrast, self-fertile species often have stigmas positioned at the same level as, or slightly below, the anthers, an arrangement that increases the efficiency of self-pollen deposition (Barrett, 2002). These associations of physiological and morphological floral traits in the outcrossing flower syndrome suggest that SI and stigma exsertion evolve coordinately, leading to the hypothesis that the co-occurrence of these two redundant mechanisms for promoting outcrossing is somehow advantageous (Darwin, 1877). However, the manner in which this coordinate evolution is achieved remains a major unresolved problem in evolution and development.
Having evolved from a self-incompatible ancestor, the model plant Arabidopsis thaliana, a highly self-fertile member of the Brassicaceae, has the potential to shed light on processes that accompany transitions in the mating system (Kusaba et al., 2001; Sherman-Broyles et al., 2007; Tang et al., 2007; Shimizu et al., 2008;). Most A. thaliana accessions have flowers in which the close proximity of stigmas and anthers favors self-pollination, and the species also lacks the SI system typical of self-incompatible members of the Brassicaceae (reviewed in Takayama and Isogai, 2005; Rea and Nasrallah, 2008). In particular, all A. thaliana accessions analyzed to date lack functional alleles of the two highly polymorphic genes that together form the SI specificity–determining S-locus haplotype or S haplotype (Sherman-Broyles et al., 2007; Tang et al., 2007; Shimizu et al., 2008): the S-locus receptor kinase (SRK) gene, which encodes a single-pass transmembrane kinase displayed on the surface of stigma epidermal cells, and the S-locus cysteine-rich protein (SCR) gene, which encodes a small pollen coat–localized protein that is the ligand for SRK. In the absence of functional SRK or SCR proteins whose haplotype-specific interaction normally triggers the inhibition of pollen tube development at the stigma surface in self-incompatible plants, the A. thaliana stigma cannot mount an SI response.
We recently established a transgenic self-incompatible A. thaliana model by transforming this species with functional SRK and SCR gene pairs isolated from its sister species, the self-incompatible Arabidopsis lyrata (Nasrallah et al., 2002, 2004). In a few accessions, the SRK and SCR transgenes confer a robust and stable SI phenotype, whereby SRK-expressing stigmas inhibit pollen expressing the cognate SCR throughout their development, resulting in lack of seed (Nasrallah et al., 2004; Boggs et al., 2009). This pollination phenotype is indistinguishable from that exhibited by self-incompatible A. lyrata plants, indicating that some accessions of A. thaliana have retained all genes required for SI except for the S-locus recognition genes (Nasrallah et al., 2004). By contrast, SRK-SCR transformants of the reference Columbia-0 (Col-0) accession express transient SI, whereby stigmas of mature floral buds and young flowers exhibit strong SI but subsequently exhibit breakdown of SI in older flowers (Nasrallah et al., 2002; Liu et al., 2007).
This age-dependent breakdown of SI offers two advantages for genetic analysis of SI: abundant seed can be obtained for mutant screens because Col-0∷SRK-SCR plants set seed upon autonomous self-pollination, and the developmental weakening of SI in these plants allows screening for mutations that enhance SI in older flowers as well as for mutations that cause breakdown of SI in young flowers. Taking advantage of these attributes, we conducted mutant screens and identified a mutation in the RNA-dependent RNA polymerase RDR6 that simultaneously enhances SI and causes stigma exsertion, the extent of which is magnified in the presence of a functional SRK gene. The study illustrates how experimental manipulation of the A. thaliana mating system, first by generating transgenic plants that express SI and subsequently by mutagenesis of these transgenic plants, can uncover processes responsible for the coordinate changes in stigma-pollen recognition and stigma-anther separation that accompany transitions between the outcrossing and selfing modes of mating, at least in the crucifer family, and thus illuminate the evolutionary forces that shape plant reproduction.
RESULTS
A Recessive Mutation Causing Enhanced SI
Seeds from a transgenic Col-0∷SRKb-SCRb strain that was homozygous for the A. lyrata–derived SRKb and SCRb genes (Nasrallah et al., 2002) were mutagenized with ethyl methanesulfonate, and M2 plants were screened for mutants exhibiting a modified SI response. An M2 family, designated m014, was identified that segregated for plants in which the majority (>95%) of siliques lacked seed and only a small number of siliques contained few seed (less than five per silique). Self-pollinations showed that these mutant plants exhibited strong SI in older flowers, unlike Col-0∷SRKb-SCRb plants. At the +3 stage of flower development (see legend to Figure 1 for numerical designation of floral developmental stages), a stage at which stigmas from Col-0∷SRKb-SCRb were highly receptive to self-pollen (Figure 1A), mutant m014∷SRKb-SCRb stigmas exhibited complete inhibition of self-pollen (Figure 1B). This difference was not due to disruption of stigma or pollen function in mutant plants because reciprocal pollinations to wild-type Col-0 were successful (Figures 1C and 1D) and produced ample seed set.
Enhanced SI and Stigma Exsertion in m014∷SRKb-SCRb Caused by a Mutation in RDR6.
(A)to (D) Self- and cross-pollination assays. Self-pollination of +3-stage flowers from Col-0∷SRKb-SCRb (A) and m014∷SRKb-SCRb(B) and reciprocal crosses to wild-type Col-0 showing that stigma (C) and pollen (D) of m014∷Sb are functional. The numerical designation of developing flower buds adopted in studies of Brassica SI is used: a floral bud at flower opening (anthesis) is floral bud stage 0 (equivalent to stage 13 of Smyth et al., 1990), with progressively younger buds along the inflorescence designated by negative numbers (−1, −2, etc.), and progressively older buds designated by positive numbers (+1, +2, etc). Bars = 50 μm in (A), (C), and (D) and 20 μm in (B).
(E) and (F) floral characteristics of Col-0∷SRKb-SCRb (E) and m014∷SRKb-SCRb plants (F) showing stigma exsertion in mutant plants.
(G) The RDR6 gene and the positions of the rdr6-en missense mutation and other previously reported rdr6 mutant alleles (Peragine et al., 2004).
(H) Self-pollen inhibition on a +3-stage stigma of an F1 plant derived from a cross between an rdr6-en∷SRKb-SCRb and an rdr6-11 homozygote. Bar = 20 μm.
In addition to the enhanced SI phenotype, m014∷SRKb-SCRb mutant plants also exhibited increased stigma-anther separation. In contrast with the flowers of Col-0 and Col-0∷SRKb-SCRb, in which stigmas are positioned at the same level as, or slightly below, the anthers (Figure 1E), m014∷SRKb-SCRb stigmas were highly exserted above the anthers (Figure 1F). To understand the association between the enhanced SI and exserted stigma phenotypes in m014∷SRKb-SCRb plants and to determine the genetic basis of the m014 mutation, a mutant plant was backcrossed to the Col-0∷SRKb-SCRb strain used for mutagenesis. Analysis of F2 plants derived from this backcross demonstrated that the two floral phenotypes were tightly linked and cosegregated as a single recessive trait (122 wild type:35 mutant; P > 0.250). These results indicate that a single mutation causes both the enhanced SI and stigma exsertion observed in m014∷SRKb-SCRb plants and thus affects two different mechanisms that promote cross-pollination.
Positional cloning identified a C-to-T missense mutation that changes a Cys at position 768 to a Tyr in the first exon of At3g49500 (Figure 1G), a gene that encodes the RNA-dependent RNA polymerase RDR6. Crossing a homozygous m014∷SRKb-SCRb plant with a plant homozygous for the previously reported Col-0 rdr6-11 allele (Peragine et al., 2004; ABRC catalog number CS24285) produced F1 plants that exhibited an enhanced SI response and strong stigma exsertion similar to m014∷SRKb-SCRb plants (Figure 1H). This result confirmed that the mutant rdr6 allele in m014 is responsible for the enhanced SI and floral phenotypes observed in this mutant. We therefore designated this allele rdr6-en (for rdr6-enhancer).
For further analysis of the role of RDR6 in SI and pistil development, we used the well-characterized null rdr6-11 allele (Peragine et al., 2004). We crossed a Col-0∷SRKb-SCRb (hereafter Col-0∷Sb) plant carrying a single integration of the SRKb/SCRb transgene pair with an rdr6-11 homozygote (Figure 1G) and identified four rdr6-11∷SRKb-SCRb (rdr6-11∷Sb) homozygous lines and four rdr6-11 homozygous sib lines lacking SRKb/SCRb. Detailed analysis of these lines demonstrated that the enhanced SI phenotype and reduced seed set of rdr6-en∷Sb plants were recapitulated in rdr6-11∷Sb plants (Figures 2A and 2B ). This result indicates that the Cys-to-Tyr mutation in rdr6-en is as disruptive to RDR6 function as the premature stop codon in rdr6-11.
Seed Production in Col-0∷Sb, rdr6-11, and rdr6-11∷Sb Plants and Expression of SRKb in Developing Stigmas.
(A)and (B) Seed production in the three lines as a measure of the degree and uniformity of successful pollination. Images of inflorescences (A) and counts of autonomous seed set (B) show abundant and uniform seed production in Col-0∷Sb (all siliques are seed filled and expanded), stochastic reduction in seed set in rdr6-11 (mixture of expanded and unexpanded siliques containing few or no seed), and the dramatic and uniformly reduced seed production in rdr6-11∷Sb (absence of fully expanded siliques and only occasional partially expanded siliques containing few seeds). In (B), the number of seeds per silique derived from autonomous self-pollination (i.e., in the absence of pollinators) was obtained by analysis of 40 siliques (10 siliques from each of four plants) from each line. In Col-0∷Sb plants, among the siliques that contained >20 seeds, 32 siliques contained >30 seeds and seven siliques contained 21 to 30 seeds.
(C) Gel blot analysis of poly(A)+ RNA from stigmas at the −1, +1, and +3 developmental stages. The blot was hybridized first with a probe derived from the SRKb first exon, which hybridizes with a 3.0-kb full-length transcript (asterisk) and an alternative 1.6-kb transcript (circle) corresponding to exon 1 and subsequently with an actin probe. The normalized values of the 3.0-kb SRKb transcript (with Col-0∷Sb set at 1.0) are indicated.
(D) Seed production after manual self- and cross-pollination. The histogram shows the average number of seeds per silique (±1.96 se) produced after manual application of pollen on +1-stage stigmas. The average number of seeds in each pollination experiment was obtained by analysis of 30 siliques (10 siliques from each of plants from each line).
Enhanced SI in the rdr6 Genetic Background
One function of RDR6 is in posttranscriptional silencing of transgenes via a small RNA-mediated mRNA degradation pathway (Mallory and Vaucheret, 2006), and a loss-of-function mutation in this gene could potentially ameliorate expression of the SRKb transgene and thus cause an enhancement of SI in the stigma. However, analysis of steady state levels of SRKb transcripts showed that SRKb transcript levels were not increased but rather were somewhat reduced in rdr6-11∷Sb stigmas relative to Col-0∷Sb stigmas at all stages of development (Figure 2C), ruling out the possibility that enhanced SI was due to increased accumulation of SRKb transcripts.
For a detailed analysis of the SI response in rdr6-11∷Sb plants, we performed pollinations by manual application of pollen grains on stigmas at the +1 stage, which is the stage at which self-pollen deposition on the stigma normally occurs in wild-type A. thaliana and also the stage at which SI is already weakened in Col-0∷Sb plants (Nasrallah et al., 2002). To this end, we assayed rdr6-11∷Sb, rdr6-11 lacking the transgenes, and Col-0∷Sb plants by self-pollination and reciprocal cross-pollinations between mutant and wild-type plants. For each of these pollinations, the success rate of pollen tube growth into stigma epidermal cells was evaluated by counting the number of seeds in each of 30 resulting siliques and the average number of seeds per silique from each line was calculated. Self-pollination of Col-0∷Sb and rdr6-11 lacking SRKb/SCRb, as well as reciprocal pollinations between rdr6-11∷Sb and wild-type Col-0, confirmed that rdr6-11 mutant plants were fertile and produced full siliques (Figure 2D). By contrast, self-pollination of rdr6-11∷Sb stigmas resulted in very low numbers of seeds per silique (Figure 2D), indicating that mutant stigmas maintained SI into late stages of flower development even upon prolonged pollen-stigma contact. This conclusion was reinforced by pollination assays in which pollen tube growth was examined 24 h after manual application of self-pollen to +1-stage stigmas. Numerous pollen tubes were observed to have already penetrated the stigma surface and to have reached ovules throughout the ovary in Col-0∷Sb plants. By contrast, strong pollen inhibition was observed at the surface of rdr6-11∷Sb stigmas.
Enhancement of Stigma Exsertion by S-Locus Transgenes in the rdr6 Mutant Background
Mutations in RDR6 are known to affect various plant processes, including virus resistance (Qu et al., 2005) and vegetative and flower development (Peragine et al., 2004). In particular, the rdr6-11 mutant was previously reported to exhibit alterations in flower architecture and stochastic loss of coordination between stamen and pistil elongation, resulting in variable seed production (Peragine et al., 2004). However, these alterations had not been subjected to detailed analysis. We therefore compared floral morphology and seed production in the wild type, an rdr6-11 homozygous line, and an rdr6-11∷Sb homozygous line. The latter two mutant lines were found to differ in the extent of seed set and stigma exsertion, not only from the wild type, but unexpectedly, also from each other. The rdr6-11∷Sb line exhibited much stronger stigma exsertion and significantly less seed production than the rdr6-11 mutant line. In rdr6-11 plants, stigma exsertion was pronounced only in the first few flowers produced after the transition from the vegetative to reproductive phase but became less apparent and more variable among different flowers as the plants grew older. By contrast, stigmas remained strongly exserted throughout the life span of rdr6-11∷Sb plants.
A detailed analysis of autonomous seed set (Figure 2B) and floral architecture (Figures 3A to 3C ) in wild-type and rdr6-11 plants containing or lacking the SRKb/SCRb transgenes explained the difference in stigma exsertion between rdr6-11 and rdr6-11∷Sb plants. In wild-type plants, the majority of siliques were fully expanded, and the uniformly abundant seed set by these plants is consistent with the placement of anthers at or slightly above the level of the stigma, which allows autonomous deposition of numerous pollen grains on the stigma. In rdr6-11 plants lacking SRKb/SCRb, the number of seeds per silique was highly variable, ranging from total lack of seed in some siliques to full seed set in other siliques. This phenotype is consistent with the observation that anthers were often, but not always, recessed below the stigma in rdr6-11 plants, which would cause variable numbers of pollen grains to be deposited on individual stigmas. By contrast, the reduction in seed set was both more uniform and more dramatic in rdr6-11∷Sb plants, as previously noted for rdr6-en∷Sb plants: 92% (37/40) of the siliques lacked seed, 8% (3/40) of the siliques contained only one to four seeds, and no siliques contained more than four seeds. This near lack of seed correlated with the more uniform and more extensive stigma exsertion (Figures 3A to 3C) and, therefore, uniform lack of pollen deposition on stigmas, observed in rdr6-11∷Sb flowers relative to rdr6-11 flowers. These observations indicate that the enhancement of stigma exsertion in the rdr6 mutant background is caused by the SRKb/SCRb transgenes themselves. This unexpected effect of the SRKb/SCRb transgenes on stigma exsertion may be related to expression of SRKb in nonstigmatic pistil tissues, which is evident in both Col-0∷Sb transgenic plants and A. lyrata Sb plants (Figure 3D).
Stigma Exsertion and Expression of SRKb in Pistil Tissues.
(A) Extent of stigma exsertion in +1-stage flowers of Col-0∷Sb, rdr6-11, and rdr6-11∷Sb plants. The four floral organs are labeled in the Col-0∷Sb panel and the anther and stigma, which cap the stamen and pistil, respectively, are labeled in the rdr6-11∷Sb panel.
(B) Relative heights (in millimeters) of stigma and anther in each of 40 +1-stage flowers from the three lines. The anther height shown is the average height of the four long anthers in each flower.
(C) Average stigma-anther separation in the three lines. For each of the 40 flowers analyzed per line, the distance between stigma and anther was assigned a positive value if the stigma was exserted above the anthers and a negative value if the stigma was recessed below the anthers. The stigma-anther separation values, presented as means ±1.96 se, were significantly different among the three lines (P value < 0.01).
(D)Gel blot analysis of poly(A)+ RNA purified from stigmas and the remainder of the pistil (style + ovary) from –1-stage buds of the three A. thaliana lines and A. lyrata (A.l.) Sb plants carrying the native SRKb allele. Hybridization and analysis of hybridization signals were as in Figure 2C. The blot was hybridized first with a probe derived from the SRKb first exon, which hybridizes with a 3.0-kb full-length transcript (asterisk) and an alternative 1.6-kb transcript (circle) corresponding to exon 1, and subsequently with an actin probe.
To determine if the observed increase in stigma exsertion was due to an increase in pistil length or to a decrease in stamen length, we compared stigma and anther heights in the wild type and rdr6 mutants containing or lacking the SRKb/SCRb transgenes. We found that the Sb transgenes had no significant effect on pistil and stamen length in wild-type Col-0 (Figure 4A ) and that in the absence of SRKb/SCRb, the rdr6 mutation promoted pistil length and reduced stamen length (i.e., it had opposite effects on pistil and stamen elongation; Figure 4B). By contrast, the presence of SRKb/SCRb in rdr6 plants promoted pistil elongation, both in the style and the ovary (Figures 4C and 4D) but had no obvious effects on stamen length (Figure 4B). This specific enhancement of pistil elongation in the otherwise morphologically normal rdr6-11∷Sb pistils is not caused by an increase in the size of cells in pistil tissues (Figure 5 ; Table 1 ).
Differential Elongation in Col-0, Col-0∷Sb, rdr6-11, and rdr6-11∷Sb Pistils.
(A) Relative pistil and stamen lengths in Col-0 and Col-0∷Sb plants. Neither pistil nor stamen length show significant differences between the two strains (P = 0.5262 and 0.3047).
(B) Relative pistil and stamen lengths in Col-0∷Sb, rdr6-11, and rdr6-11∷Sb. Pistil length differs significantly among the three lines (P < 0.01). Stamen length in rdr6-11 was significantly shorter than that in Col-0∷Sb (P < 0.01), whereas stamen length in rdr6-11∷Sb was not significantly different from rdr6-11 (P = 0.254).
(C)and (D) Relative lengths of the stigma+style region (C) and ovary (D) in Col-0∷Sb, rdr6-11, and rdr6-11∷Sb plants. For each line, the average length (±1.96 se) of each structure was calculated from measurements performed as described in Figure 3 on 40 flowers at the +1 stage (10 flowers from each of four plants).
Scanning Electron Micrographs of +1-Stage Pistils from Col-0∷Sb and rdr6-11∷Sb Plants.
The higher magnification micrographs show close-ups of the style (top) and a region in the middle of the ovary (bottom) from the same pistil of each genotype.
Comparison of Cell Length in Col-0∷Sb and rdr6-11∷Sb Pistils Using a Paired t Test
A Functional SRK Gene Is Responsible for Enhanced Pistil Elongation in rdr6 Mutants
To test the hypothesis that the observed effect of the SRKb/SCRb transgenes on stigma exsertion is related to expression of SRKb in nonstigmatic pistil tissues, we introduced SRKb alone or SCRb alone into the rdr6-11 mutant background by crossing an rdr6-11 homozygous plant with either a Col-0∷SRKb or a Col-0∷SCRb transgenic plant. Three independent lines each of rdr6-11∷SRKb, rdr6-11∷SCRb, and rdr6-11 lacking both SRKb and SCRb were selected for comparative studies of stigma exsertion. Consistent with the hypothesis that SRKb alone was responsible for enhanced stigma exsertion in rdr6-11∷Sb plants, only rdr6-11∷SRKb plants, and not rdr6-11∷SCRb plants, exhibited more extensive stigma exsertion and reduction in seed set than rdr6-11 plants lacking SRKb/SCRb (Figure 6A ). Furthermore, the extent of stigma exsertion, like the strength of SI, was correlated with the steady state levels of SRKb transcripts in rdr6-11∷SRKb pistils, as demonstrated by comparing independent rdr6-11∷SRKb transformants (Figure 6B): transformants that expressed the highest SRKb mRNA levels exhibited the most extensive stigma-anther separation and the most severe inhibition of SCRb-expressing pollen.
Correlation of SI, Stigma Exsertion, and Seed Set with SRKb Transcript Levels in rdr6-11∷Sb Plants.
(A) Enhancement of stigma exsertion by SRKb. Only rdr6-11∷SRKb flowers, but not rdr6-11∷SCRb flowers, exhibited enhanced stigma exsertion relative to untransformed rdr6-11 flowers.
(B) Effect of SRKb expression levels on SI, stigma-anther separation, and seed set. Four independent rdr6-11∷SRKb T1 plants (designated as numbers 1 through 4), which exhibited different degrees of seed set, were used for quantitative real-time RT-PCR analysis. The values (±sd) were derived from analysis of three biological replicates. Ten flowers from each T1 plant were used to assess the strength of SI [average number of pollen tubes (±1.96 se) after manual application of SCRb-expressing pollen on −1-stage stigmas] and the enhancement of stigma exsertion [average stigma-anther separation (±1.96 se)]. The average number of seeds (±1.96 se) reflects the extent of stigma-anther separation.
To determine if SRKb kinase activity is required for the enhancement of pistil elongation in the rdr6-11 mutant background, a kinase-negative mutant form of SRKb was generated. Previous experiments using Brassica SRKs had shown that a substitution of the conserved Lys residue in the putative ATP binding site abolishes the catalytic activity of the SRK kinase (Goring and Rothstein, 1992; Stein and Nasrallah, 1993). We therefore replaced the corresponding Lys (Lys-529) in SRKb with an Arg and introduced the SRKb gene carrying the Lys529Arg mutation [SRKb(Lys529Arg)] into rdr6-11 plants by Agrobacterium tumefaciens–mediated transformation. Approximately 100 independent transgenic T1 plants were generated, and 60 of these plants were tested for SI by pollinating stigmas with pollen expressing SCRb. In all 60 transformants, rdr6-11∷SRKb(Lys529Arg) stigmas were fully receptive to SCRb-expressing pollen (with numerous pollen tubes observed per stigma), confirming that these stigmas could not mount an SI response and that the SRKb(Lys529Arg) mutant was inactive. Furthermore, none of the 100 T1 rdr6-11∷SRKb(Lys529Arg) transformants showed the enhancement of stigma exsertion and drastic reduction in seed production observed in rdr6-11∷SRKb transformants (Figure 7A ). This ineffectiveness of the SRKb(Lys529Arg) mutant in enhancing stigma exsertion was not due to suboptimal SRKb(Lys529Arg) transcript levels because the stigmas of some of rdr6-11∷SRKb(Lys529Arg) transformants accumulated much higher transcript levels than those of rdr6-11∷SRKb plants having strong stigma exsertion (Figure 7B). These results demonstrate that enhanced stigma exsertion in rdr6∷SRKb plants, like SI, requires a catalytically active SRK kinase.
Requirement of a Catalytically Active SRKb for Stigma Exsertion in rdr6 Plants.
(A) Floral morphology of rdr6-11∷SRKbmu (kinase-negative mutant) and untransformed rdr6-11 plants.
(B) Comparison of relative SRKb mRNA levels measured by real-time quantitative RT-PCR in different rdr6-11∷SRKbmu transgenic lines and the transgenic line shown in Figure 3A, which contains a functional SRKb and exhibits enhanced stigma exsertion.
Enhancement of SI and Stigma Exsertion Are Dependent on trans-Acting Short Interfering RNA Pathways
The fact that RDR6 functions in the production of short interfering RNAs (siRNAs) suggests that small RNAs function in the regulation of SI and pistil elongation. To determine if trans-acting siRNAs (ta-siRNAs), rather than other siRNAs, are responsible for the floral and pollination phenotypes of rdr6∷Sb plants, we tested the effect of mutations in two genes known to be involved in two different siRNA pathways: RDR2, an RNA-dependent RNA polymerase that functions in transcriptional silencing through an siRNA-directed chromatin silencing pathway (Zhang et al., 2007), and ARGONAUTE7/ZIPPY (AGO7/ZIP), which functions in posttranscriptional silencing via ta-siRNA–mediated regulation and has been implicated in pistil development (Adenot et al., 2006; Fahlgren et al., 2006; Garcia et al., 2006; Hunter et al., 2006). Plants carrying inactivating T-DNA insertions in RDR2 or AGO7/ZIP (see Methods) were crossed to Col-0∷Sb plants, and plants homozygous for the rdr2 or ago7/zip mutations and carrying the Sb transgenes were analyzed. Only ago7/zip∷Sb plants, but not rdr2∷Sb plants, were found to exhibit enhanced SI and stigma exsertion, as shown in Figure 8 .
Enhanced SI and Stigma Exsertion in ago7/zip∷Sb Plants.
(A) and (B) Self-pollination assays of Col-0∷Sb and ago7/zip∷Sb flowers showing large numbers of elongated pollen tubes, indicative of complete breakdown of SI, in +1-stage Col-0∷Sb flowers (A) and lack of elongated pollen tubes, indicative of enhanced SI, in ago7/zip∷Sb flowers (B) after prolonged stigma-pollen contact (9 h).
(C) and (D) Floral characteristics of ago7/zip (C) and ago7/zip∷Sb (D) plants showing enhanced stigma exsertion in ago7/zip∷Sb.
(E) Average stigma-anther separation (±1.96 se) in ago7/zip and ago7/zip∷Sb obtained from analysis of 40 +1-stage flowers from each line. The stigma-anther separation values are significantly different between the two lines (P < 0.01).
DISCUSSION
This article demonstrates that the model plant A. thaliana, and in particular its reference Col-0 accession, although highly self-fertile and lacking both SI and stigma exsertion, can provide significant insight into major processes that underlie the outcrossing mode of breeding. In addition to identifying a major regulator of SI, our study has uncovered a dual role for SRK in SI and pistil elongation and revealed tight genetic integration between SI signaling and pistil developmental pathways.
RDR6 and Regulation of SI
Our identification of an rdr6 mutation that causes enhanced SI despite somewhat reduced SRKb transcript levels demonstrates a role for RDR6 as a negative regulator of SI. Notably, this role was revealed only because our mutant screens were performed in a genetic background that confers partial loss of SI. Because enhanced SI was observed in rdr6∷Sb and ago7/zip∷Sb plants and thus appears to be dependent specifically on ta-siRNA pathways, it is likely due to the lack of ta-siRNA–mediated regulation of one or more positive regulator or effector of the SI response. In wild-type Sb plants, regulation by ta-siRNA would cause reduced expression of the proposed positive factor(s) to marginal levels, thus limiting SI activity, while ta-siRNA loss in rdr6∷Sb or ago7/zip∷Sb plants would result in increased expression of the positive factor(s) and enhanced SI.
Because rdr6:Sb stigmas are receptive to pollen lacking SCRb, the SI response is not constitutively active in these stigmas but is activated specifically by SCRb-expressing pollen. It is therefore unlikely that the proposed positive factor acts directly on SI signaling intermediates and thus bypasses the requirement for SRK in inhibition of pollen. Previous studies had identified positive effectors or regulators of the SI response, including genes for the cytosolic Ser/Thr M-locus Protein Kinase (Murase et al., 2004), the U-box E3 ubiquitin ligase Armadillo Repeat-Containing protein 1 (Gu et al., 1998), and the U-box protein Plant U-box protein 8 (Liu et al., 2007). However, these genes are unlikely to be responsible for enhanced SI in rdr6∷Sb plants because their expression levels did not increase in this mutant background.
Overlaps between the SI and Pistil Developmental Pathways and SRK-Mediated Promotion of Pistil Elongation
The pleiotropic effect of rdr6 mutations on SI and stigma exsertion unmasks overlaps between pathways for SI signaling and pistil elongation. How these pathways intersect through the siRNA pathway remains to be determined. However, among known RDR6 targets are the auxin response factors ARF2, ARF3/ETTIN, and ARF4 (Peragine et al., 2004; Williams et al., 2005). Furthermore, it has been shown that AGO7/ZIP is specifically responsible for the generation of ta-siRNAs from the trans-acting siRNA locus TAS3, which regulates ARF3 and ARF4 (Adenot et al., 2006; Fahlgren et al., 2006; Garcia et al., 2006; Hunter et al., 2006). Therefore, a tantalizing possibility is that SI and stigma exsertion are both mediated by auxin, a plant growth regulator that affects cell division and cell elongation and is a versatile integrator of various plant developmental and physiological processes (Fleming, 2006).
It should be noted that the recovery of mutations having pleiotropic effects was not unexpected in a mutagenesis screen for A. thaliana plants exhibiting a modified SI response. Indeed, the maintenance of SI signaling components in A. thaliana despite the lack of a functioning SI system in this species is consistent with the notion that the SI signaling pathway might overlap with signaling cascades that underlie other biological processes. More unexpected was our finding that SRKb enhances pistil elongation in rdr6 mutants because SRK had been thought to function exclusively in SI. Nevertheless, SRKb, like other SRKs (Kusaba et al., 2001), is expressed in nonstigmatic tissues of the pistil, and our results reveal that this expression has biological consequences, at least in the rdr6 mutant background. The requirement of a catalytically active SRK kinase for enhanced pistil elongation in rdr6∷SRKb plants further indicates that the SRK protein either activates or reinforces a signaling cascade involved in pistil development. Similar to the enhancement in SI, enhancement of pistil elongation in rdr6∷Sb and rdr6∷SRKb plants might be caused by increased expression of a factor that functions in concert with SRK to promote pistil elongation.
The fact that SRK affects pistil elongation in the absence of its SCR ligand implies that the receptor is constitutively active in the nonstigmatic pistil cells of the rdr6 genetic background. Why SRK kinase activity is tightly regulated in stigma epidermal cells but not in other cells of the rdr6∷SRKb-SCRb pistil is not understood. Two thioredoxin h-like proteins have been proposed to function in the stigma as negative regulators of SI by maintaining SRK in an inactive state in the absence of SCR (Cabrillac et al., 2001). Suppression of these proteins was found to cause a partial constitutive inhibition of both self-pollen and non-self pollen (Haffani et al., 2004). Both genes are expressed in the transmitting tract of the pistil (Tung et al., 2005), and it is possible that these or other negative regulators of SRK kinase activity are reduced or ineffective in SRK-expressing nonstigmatic rdr6 pistil cells.
Coevolution of SI and Stigma Exsertion
The concept of adaptive advantage was proposed to explain the maintenance in the same species of both physiological and morphological mechanisms that redundantly promote outcrossing (reviewed in Barrett, 2002). Three hypotheses have been proposed to explain the adaptive advantage of a morphological barrier to self-fertilization in self-incompatible species: promoting pollen transfer, reducing pollen wastage, and reducing self-pollen interference (Barrett, 2002). However, although several independent observations have supported the role of stigma exsertion in promoting cross-pollination in self-compatible plants (Kohn and Barrett, 1992; Motten and Stone, 2000, and references within), the adaptive value of this trait in self-incompatible species has not been demonstrated in a large number of studies that have addressed this issue (Bjorkman, 1995; Koelling and Karoly, 2007, and references within). Furthermore, a molecular explanation of how these physiological and morphological floral traits are integrated during the evolution of plant sexual diversity has not been formulated.
Our results provide a molecular explanation for the coordination of SI and pistil development during evolution and for the observation that dramatic changes in floral architecture can occur rapidly upon loss of SI during evolutionary switches from out-crossing to self-fertility (Foxe et al., 2009). Our finding that RDR6 is a common regulator of SI and stigma exsertion and that SRK functions in both SI and pistil elongation shows how, rather than acting on both traits, selection can act on one trait and simultaneously affect the other trait. Interestingly, pleiotropy is increasingly viewed as being responsible for associations between other floral characters that are considered to be important for the evolution of mating systems (Juenger et al., 2000; Conner, 2002; Juenger et al., 2005), although the molecular mechanisms that underlie these associations are not understood.
Evolutionary Switches in Mating System and Small RNAs
The emergence of a selfing species from an outcrossing ancestor has been generally accepted as the universal path for plant mating system evolution (Stebbins, 1957; Barrett, 2002). Based on the complexity of the SI system, it had been thought that evolutionary switches from SI to self-compatibility are irreversible in natural populations. Indeed, phylogenetic and maximum-likelihood analyses of transition rates between SI and self-compatibility in the Solanaceae support this view (Igic et al., 2006). However, a phylogenetic analysis of species in the Asteraceae that exhibit varying degrees of SI has detected substantial rates of reversal to SI from both self-compatible and pseudo-self-compatible species (Ferrer and Good-Avila, 2007). The shift from the pseudo-self-compatibility exhibited by Col-0∷Sb plants to the robust SI phenotype observed in rdr6∷Sb plants is consistent with the notion that the switch from outcrossing to selfing can be reversed in crucifers, at least during transitional phases of mating-system evolution, via modulation of a small RNA that regulates a target gene involved in the SI response.
The data further suggest that evolutionary switches to self-fertility can result from the emergence of a small RNA that suppresses the SI response. Certainly, the evolutionary histories of small RNAs and their targets are consistent with such a process. Comparative studies of small RNAs and their targets have revealed that a target gene exists prior to the emergence of its small RNA (Floyd and Bowman, 2004). Importantly, the evolution of small RNAs is thought to be a highly dynamic process because a large number of these RNAs are not conserved among different plant species (reviewed in Axtell and Bowman, 2008). This feature of small RNA evolution might explain how a positive effector of SI that is regulated by ta-siRNA(s) can exhibit different regulation in self-fertile and self-incompatible plant lineages and also how such ta-siRNA(s) might be lost from self-fertile lineages to generate self-incompatible plants.
RDR6 and the Fine-Tuning of Reproductive Processes
Our study provides a new perspective on the function of RDR6 and ta-siRNAs in plants by demonstrating a role for these molecules in the fine-tuning of processes critical for reproduction in plants. Loss-of function mutations in genes involved in ta-siRNA production generally cause only subtle vegetative developmental phenotypes that have no obvious impact on plant performance, suggesting that plants use ta-siRNAs sparingly to regulate endogenous genes. This view is supported by the observation that plants accumulate only very low levels of ta-siRNAs compared with microRNAs (Williams et al., 2005). By contrast, our results show that floral traits that are fine-tuned for a specific pollination mode are much more sensitive to defects in ta-siRNA production than vegetative traits, with significant consequences for the success rate of self- or cross-pollination.
The stochastic loss of coordination between stamen and pistil elongation resulting in reduced seed production in A. thaliana rdr6 plants suggests that RDR6 is integrated into the regulatory network of flower development and is essential for the maintenance of morphological features critical for self-fertilization. We propose that, in self-fertile A. thaliana, gene regulation by the RDR6 pathway normally buffers pistil development against external, and possibly internal, perturbations and thus maintains low variability for the close positioning of stigma and anther, a feature critical for the effective self-pollination and uniform seed set observed in most, if not all, accessions of the species. In this view, RDR6 would function as a canalizing gene (i.e., a gene that confers stability or robustness on inherently noisy developmental networks; Flatt, 2005). Currently, Hsp90 is the only canalizing gene known (Queitsch et al., 2002). However, canalizing functions have been proposed for microRNAs (Hornstein and Shomron, 2006) and Wingless signaling pathway genes (Arias and Hayward, 2006). Our results suggest that ta-siRNAs and the SRK signaling pathway have similar canalyzing functions in the establishment of floral characters that enforce the selfing and outcrossing modes of mating.
METHODS
Mutagenesis and Mutant Screens
The transgenic self-incompatible Arabidopsis thaliana line used in this study was generated by transforming the Col-0 accession with Arabidopsis lyrata SRKb and SCRb, as described previously (Nasrallah et al., 2002). Seeds from this transgenic line were used for ethyl methanesulfonate mutagenesis. M2 seeds from each resulting M1 plant were collected and screened as individual families for plants exhibiting reduced seed set relative to unmutagenized Col-0∷SRKb-SCRb plants. Approximately 16 M2 plants from each of ∼100 M1 families were analyzed. A family, designated m014∷SRKb-SCRb, segregated for reduced seed set. SI phenotype was evaluated by comparing self-pollen inhibition in +3 flowers of mutant and wild-type plants in the m014∷SRKb-SCRb family, and mutant plants were reciprocally pollinated with wild-type Col-0 to test for stigma and pollen functions.
Map-Based Cloning and Gene Validation
The m014 line was subjected to three rounds of backcrossing to Col-0∷ SRKb-SCRb plants, and an F2 mapping population was established by crossing a mutant SRKb-SCRb plant with a wild-type plant of the Landsberg erecta accession. Using 212 F2 mutant plants and simple sequence length polymorphism markers designed from the Landsberg erecta random sequence database (www.tigr.org), the mutation was mapped to chromosome 3 within a 100-kb interval flanked by At3g49420 and At3g49690. To identify the gene disrupted by the mutation, the coding regions of the 28 annotated genes in this interval were amplified from the m014 line using gene-specific primers designed from the Col-0 genome sequence (www.tigr.org). PCR products were subjected to automated sequencing on an Applied Biosystems 3730xl DNA analyzer, using big dye terminator chemistry and AmpliTaq-FS DNA polymerase, at the BioResource Center at Cornell University (Ithaca, NY). Sequences were compared with the wild-type Col-0 genome sequence (www.tigr.org) to identify the mutation.
A Col-0 strain homozygous for the rdr6-11 allele (Peragine et al., 2004; seed stock number CS24285) was obtained from the ABRC (Columbus, Ohio). To confirm the presence of a nonsense mutation in the rdr6-11 allele of this strain, a region of the RDR6 gene was amplified from genomic DNA and sequenced using the following gene-specific primers: 5′-CTTAGATCTTCCAACGGAGTAG-3′ and 5′-GCTCCCCAAAACAAGGCTCATTA-3′. Salk lines 059661 and 080533 carrying T-DNA insertions in RDR2 (At4g11130) and AGO7/ZIP (At1g69440), respectively, were obtained from the ABRC.
RNA Analysis
For RNA gel blot analysis, pistils from the –1, +1, and +3 stages of development were dissected to generate one sample consisting of stigma tissue and another consisting of style and ovary. Approximately 100 stigmas or 50 style/ovary segments were pooled for poly(A)+ RNA purification using the FastTrack RNA isolation kit (Invitrogen) and subjected to RNA gel blot analysis as previously described (Kusaba et al., 2001). The blots were hybridized as previously described with 32P-labeled fragments derived from the first exon of SRKb and an A. thaliana actin gene. Hybridized blots were exposed to phosphor screens and developed using a STORM 860 PhosphorImager (Molecular Dynamics). Signal intensities were quantitated using the ImageQuant software package (Molecular Dynamics) and normalized using actin hybridization signals.
For quantitative real-time RT-PCR of SRKb transcripts, total RNA was isolated from 25 pistils dissected from −1-stage buds using the Trizol reagent (Invitrogen). One microgram of total RNA was treated with DNase I (Invitrogen), reverse-transcribed with oligo(dT) primers and Superscript II (Invitrogen), and the resulting single-stranded cDNA was subjected to quantitative real-time RT-PCR using an ABI Prism 7900HT sequence detection system and SYBR green fluorescence (iQ SYBR Green Supermix; Bio-Rad). The following primers were used: SRKb primers flanking the sixth intron of the gene (5′-AATAACCTGCTCGGCTACGC-3′ and 5′-GCTGAATCTACGATGAATGGATCT-3′), and primers specific for the Ubiquitin Conjugating (UBC) gene At5g25760 (5′-CTGCGACTCAGGGAATCTTCTAA-3′ and 5′-TTGTGCCATTGAATTGAACCC-3′). The relative amount of transcripts was calculated using the comparative CT (threshold cycle) method and normalized to the endogenous UBC reference. The mean CT and sd values were calculated from three replicates of each sample.
Pollination Assays, Imaging, and Measurement of Floral Structures, and Scanning Electron Microscopy
For pollination assays, floral buds were emasculated before anthesis, and stigmas were allowed to develop to specific developmental stages before manual application of pollen grains. For assays of +3-stage stigmas, pollinated flowers were left on a 0.4% (w/v) agar plate for 3 h before fixation and processing for visualization of pollen tube growth by epifluorescence microscopy (Nasrallah et al., 2002).
Images of manually dissected +1-stage flowers were captured using the Scion image program (www.scioncorp.com), and measurements were done using the ImageJ image processing program developed at the National Institutes of Health. To compare flower morphology among Col-0∷Sb, rdr6-11, and rdr6-11∷Sb plants, 40 flowers (10 flowers from each of four plants) from each genotype were used. For analysis of anther height, only the four long anthers, which are directly involved in self-pollination, were measured. Comparisons of stigma-anther separation (n = 40) and pistil length (n = 40) between pairs of genotypes were performed using the paired t test program from Simple Interactive Statistical Analysis (http://home.clara.net/sisa/). Evaluations of significance were done using double-sided P value.
For scanning electron microscopy, whole +1-stage flowers were processed essentially according to (Thorsness et al., 1991), and the samples were observed using a Hitachi S4500 scanning electron microscope (Cornell Integrated Microscopy Center, Cornell University). The size of pistil cells in the Col-0∷Sb and rdr6-11∷Sb lines was measured from scanning electron micrographs as follows. The image of the pistil was divided into five 200-μ sections, with one section corresponding to the style (Section 1 in Table 1) and four sections spanning the length of the ovary from its top (Section 2 in Table 1) to its base (Section 5 in Table 1). The length of 20 cells in each section was measured using the ImageJ program and compared using a paired t test as described above.
Construction of the SRKb Kinase-Negative Mutant
The SRKb genomic clone used for the generation of the transgenic Col-0∷Sb line (Nasrallah et al., 2002) was modified by replacing the conserved Lys residue in the putative ATP binding site of SRKb with an Arg using the overlap extension PCR method of site-directed mutagenesis described previously (Stein and Nasrallah, 1993). Two overlapping SRKb kinase genomic fragments were amplified using primer pairs A and B (5′-TAGTTCGGCGGCAATTGTAGATAG-3′ and 5′-GGATAGCCTTCTCACCGCAATCTCTTGTCC-3′) and primer pairs C and D (5′-CGGACAAGAGATTGCGGTGAGAAGGCTATCC-3′ and 5′-CCGGCGGTTTAGGCTGAG-3′) (residues in bold show the positions of the substitution on the primer). The SRKb kinase containing the A-to-G missense mutation was reconstituted from the two PCR fragments by recombinant PCR using primers A and D and subsequently used to replace the corresponding region in the wild-type SRKb gene. The resulting mutant SRKb gene was introduced into rdr6-11 mutant plants by the floral dip transformation method (Clough and Bent, 1998).
Acknowledgments
We thank Mike Scanlon and Steve Tanksley for use of imaging systems, Suzy Strickler for advice with real-time RT-PCR and analysis of rdr2 plants, Tiffany Crispell for technical assistance, and Carole Daugherty for scanning electron microscopy. Seed for mutant strains was obtained from the ABRC. This work was supported by National Institutes of Health Grant GM057527 and USDA Grant 2006-35301-16894.
Footnotes
-
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: June B. Nasrallah (jbn2{at}cornell.edu).
- Received April 10, 2009.
- Revised August 14, 2009.
- Accepted August 22, 2009.
- Published September 18, 2009.
References
