It has been suggested that, in Arabidopsis, auxin controls the timing of anther dehiscence, possibly by preventing premature endothecium lignification. We show here that auxin content in anthers peaks before the beginning of dehiscence and decreases when endothecium lignification occurs. We show that, in the auxin-perception mutants afb1-3 and tir1 afb2 afb3, endothecium lignification and anther dehiscence occur earlier than wild-type, and the gene encoding the transcription factor MYB26, which is required for endothecium lignification, is over-expressed specifically at early stages; in agreement, MYB26 expression is reduced in naphthalene acetic acid-treated anthers, and afb1 myb26 double mutants show no endothecial lignification, suggesting that auxin acts through MYB26. As jasmonic acid (JA) controls anther dehiscence, we analysed how auxin and JA interact. In the JA-defective opr3 mutant, indehiscent anthers show normal timing of endothecium lignification, suggesting that JA does not control this event. We show that expression of the OPR3 and DAD1 JA biosynthetic genes is enhanced in afb1-3 and tir1 afb2 afb3 flower buds, but is reduced in naphthalene acetic acid-treated flower buds, suggesting that auxin negatively regulates JA biosynthesis. The double mutant afb1 opr3 shows premature endothecium lignification, as in afb1-3, and indehiscent anthers due to lack of JA, which is required for stomium opening. By treating afb1 opr3 and opr3 inflorescences with JA, we show that a high JA content and precocious endothecium lignification both contribute to induction of early anther dehiscence. We propose that auxin controls anther dehiscence timing by negatively regulating two key events: endothecium lignification via MYB26, and stomium opening via the control of JA biosynthesis.
In Arabidopsis, late stamen development consists of three developmental programs: anther dehiscence, pollen maturation and filament elongation. Coordination of these three processes ensures the release of mature pollen grains onto the stigma when filament elongation is completed.
Anther dehiscence and pollen maturation begin when the epidermis, endothecium, middle layer and tapetum inside the anther have differentiated, and meiosis has been completed. Anther dehiscence is characterized by three main events: development of lignified secondary walls in endothecial cells, followed by degradation of septum cells leading to a bi-locular anther, and finally breakage of the stomium, a group of specialized epidermal cells, which allows the release of pollen grains at anthesis (Goldberg et al., 1993). The first step in the anther dehiscence process is expansion of endothecial cells and the start of degeneration of the tapetum and middle layer. Development of lignified secondary walls occurs after endothecial cells have expanded when the degeneration of tapetum and middle layer is observed, and this correlates with the first pollen mitotic division. Endothecium lignification is essential for anther opening, as it is responsible for the tension that leads to stomium breakage (Keijzer, 1987; Bonner and Dickinson, 1989; Mitsuda et al., 2005; Yang et al., 2007). It has been demonstrated that the transcription factor MYB26 plays a regulatory role in endothecium lignification, as wall thickening is not observed in the endothecial cells of the ms35/myb26 mutant that is defective in MYB26, which displays indehiscent anthers. MYB26 acts upstream of the lignin biosynthesis pathway, and MYB26 expression in anthers is mainly observed from the start of pollen mitosis I to the end of the bicellular pollen stage (Dawson et al., 1999; Yang et al., 2007).
In a previous study (Cecchetti et al.,2008), we showed that auxin regulates the start of anther dehiscence by controlling the timing of endothecium lignification, as the triple and quadruple auxin receptor mutants tir1 afb2 afb3 and tir1 afb1 afb2 afb3 exhibited precocious endothecium lignification and early anther dehiscence; pollen maturation was also precocious in these mutants, and stamen filament growth was reduced, leading to shorter filaments compared to wild-type. Auxin is synthesized in anthers at pre-meiotic and meiotic stages by YUC2 and YUC6, and accumulation of auxin, as indicated by the activity of the auxin-inducible promoter DR5 (Ulmasov et al., 1997), was observed at the end of meiosis. DR5 activity decreases when endothecium lignification occurs, and is no longer detectable at the bi-locular stage, suggesting that auxin acts at the start of late stamen development (Cecchetti et al., 2008).
It has been shown by various authors that jasmonic acid (JA) is also involved in the control of anther dehiscence, as mutants defective in JA biosynthesis (Stintzi and Browse, 2000; Ishiguro et al., 2001) or perception (Xie et al., 1998; Devoto et al., 2002) show indehiscent anthers and are consequently male-sterile. Five biosynthetic genes are involved in JA biosynthesis: DAD1, LOX, AOS, AOC and OPR3. It has been shown that DAD1 and OPR3 are expressed in stamen filaments and anthers, respectively, during late stamen development. JA is also involved in other processes occurring during late stamen development, as both dad1 and opr3 mutants show reduced pollen viability, and the opr3 mutant has shortened filaments (Sanders et al., 2000; Stintzi and Browse, 2000).
The effect of JA on anther dehiscence was analysed in dad1 mutants by Ishiguro et al. (2001), who proposed that JA causes anther dehydration, which is responsible for stomium breakage and opening at the final stages of anther development. Accordingly, Sanders et al. (2000) have shown that stomium breakage is delayed in the dde1 and opr3 mutant.
Indirect evidence suggesting that auxin may activate the biosynthesis of JA in late stamen development comes from the analysis of double mutants defective in the auxin response factors (ARFs) ARF6 and ARF8, which have indehiscent anthers and decreased jasmonate production in flower buds during late development (Nagpal et al., 2005). Furthermore, it has been shown that, in arf6 arf8 flower buds, the expression of DAD1 is severely decreased (Tabata et al., 2010). Very recently, an auxin-dependent regulatory role against JA has been proposed for ARF5/MONOPTEROS (Garrett et al., 2012), as the gain-of-function mutant mpabn shows indehiscent anthers that open upon JA treatment.
The present study was designed to extend our knowledge on how auxin controls the timing of endothecium lignification, and to clarify the interaction between auxin and JA in determining anther dehiscence. Using auxin and JA concentration measurements, auxin perception single and multiple mutants, naphthalene acetic acid (NAA) and JA treatments, JA biosynthetic mutants and analysis of the levels of relevant transcripts, we show that changes in auxin concentration control the timing of endothecium lignification and modulate the JA biosynthesis that is responsible for stomium opening.
Changes in IAA concentration control late anther developmental stages
In a previous paper (Cecchetti et al., 2008), we showed that the activity of the DR5 promoter in anthers is undetectable at stages 8 and 9 [pre-meiotic and meiotic stages of flower development respectively (Bowman, 1994)], and peaks at stage 10 before tapetum degeneration and endothecium lignification; DR5 activity decreases at stage 11 when the tapetum degenerates and endothecium lignification occurs, and is no longer detectable at stage 12 after septum lysis when tapetum degeneration is completed.
To verify whether these changes in DR5 activity corresponded to actual changes in auxin concentration controlling and coordinating the events leading to anther dehiscence, we measured the free indole-3-acetic acid (IAA) content in anthers at various developmental stages (see Methods S1). As shown in Table 1, free IAA content was quite low during stages 8 and 9 (pre-meiotic and meiotic stages), peaked at stage 10 (microspore stage), i.e. before endothecium lignification, decreased at stage 11 (pollen mitosis II stage) when lignification has occurred, and further decreased at stage 12 (tricellular pollen stage) when the anther is bi-locular. These data show that changes in auxin concentration do indeed take place during anther development, and are consistent with the reported changes in DR5 activity and the proposed role of auxin in stamen development (Cecchetti et al., 2008).
Table 1. Free IAA content in wild-type anthers at various developmental stages
The afb1 auxin perception mutant is defective only in anther dehiscence
Here, we aim to better define the role of auxin and shed light on its interaction with JA in controlling specifically anther dehiscence. We therefore looked for tir1 or afb auxin receptor mutants (Dharmasiri et al., 2005) that were altered only in anther dehiscence by analysing the phenotype of tir1-1, afb1-3, afb2-3 and afb3-4 mutants (Cecchetti et al., 2008).
We first measured the percentage of dehiscent anthers in mutant and wild-type flowers at various developmental stages. We also analysed this percentage in the triple mutant tir1 afb2 afb3, which is defective in anther dehiscence (Cecchetti et al., 2008). We observed early anther dehiscence at stage 12 initial (12i) before rapid filament elongation (compared to stage 13 in the wild-type) in approximately 10% of afb1-3 and afb2-3 (Figure 1a and Figure S1a) and in 35% of tir1 afb2 afb3 flowers (Figure 1a) but not in tir1-1 and afb3-4 (Figure S1a). To assess whether pollen maturation is also premature in afb1-3 and afb2-3, we performed in vitro germination assays on pollen grains from stage 12 and 13 flowers. Wild-type and afb1-3 stage 12 pollen grains did not germinate after 24 h of culture, whereas 5% of afb2-3 pollen grains at the same stage developed pollen tubes (Figure S1b,c). At stage 13, wild-type and afb1-3 pollen grains showed comparable percentages of germination, but afb2-3 pollen grains showed a significantly higher one (Table S1). Filament length in afb1-3 stamens was comparable to that of wild-type, but afb2-3 filaments were slightly but significantly shorter than those of wild-type (Table S1). These results suggest afb1-3 as the most suitable mutant for the present study as it is only defective in anther dehiscence.
To assess whether expression of AFB1 in wild-type plants is compatible with the early anther dehiscence of the afb1-3 mutant, we performed an in situ hybridization analysis on flower buds at various developmental stages. As shown in Figure 1(b–g), the AFB1 mRNA signal is faint at pre-meiotic stage 8 (Figure 1b) but strong at meiotic stage 9, localized in the tapetum, endothecium and tetrads (Figure 1c). At stage 10, before endothecium lignification, the AFB1 mRNA signal is still present in tissues surrounding the theca (Figure 1d) but is undetectable at stage 11 during endothecium lignification, while tapetum degeneration occurs (Figure 1e). The signal is absent at all stages in procambial cells (Figure 1b–e) and in negative controls (Figure 1f,g). These data indicate that AFB1 mRNA is detectable specifically in tissues surrounding the theca before endothecium lignification.
Thus afb1-3 mutant lines were used for subsequent analysis. The triple mutant tir1 afb2 afb3 was also used, although it is also defective in pollen maturation and filament elongation (Cecchetti et al., 2008), because it shows a higher percentage of early dehiscent anthers (see above) compared to afb1-3 (Figure 1a), but expresses AFB1.
Auxin controls the timing of endothecium lignification via the MYB26 gene
Endothecium lignification is the main event at the start of the dehiscence process, and occurs at stage 11 of flower development in wild-type anthers, during tapetum degeneration. In a previous study (Cecchetti et al., 2008), we showed that early anther dehiscence occurs in tir1 afb1 afb2 afb3 quadruple mutants due to precocious endothecium lignification at the end of stage 10, before tapetum degeneration. To verify whether early anther dehiscence in afb1-3 and tir1 afb2 afb3 mutants is also due to precocious endothecium lignification, we performed a comparative histological analysis at various developmental stages (see Methods S2). As shown in Figure 2(a), at stage 10, no endothecium lignification is observed in wild-type anthers, which show lignified thickenings at stage 11 (Figure 2b). In contrast, endothecium lignification is clearly visible in afb1-3 and tir1 afb2 afb3 anthers from stage 10 (Figure 2a,b), but not earlier than stage 10 (Figure S2).
It has been shown by Yang et al. (2007) that the transcription factor MYB26 is required for endothecium lignification. To assess whether MYB26 expression in afb1-3 and tir1 afb2 afb3 is consistent with the precocious endothecium lignification in these mutants, we analysed the level of MYB26 transcript in flower buds at various developmental stages by quantitative RT-PCR.
As shown in Figure 2(c), the expression of MYB26 was negligible in wild-type flower buds until stage 10 when it peaked before tapetum degeneration, and then decreased at stage 11 when endothecium lignification occurs. In contrast, in both afb1-3 and tir1 afb2 afb3 flower buds, the level of MYB26 mRNA was significantly higher than in the wild-type before stage 10, i.e. just before the precocious endothecium lignification, but the peak of MYB26 expression at stage 10 was lower in both afb1-3 and tir1 afb2 afb3.
To assess whether MYB26 is expressed early specifically in anthers of afb1-3 and tir1 afb2 afb3 mutants, we generated and analysed wild-type (Col-0 and Ws), afb1-3 and tir1 afb2 afb3 plants harbouring the GUS reporter gene driven by the MYB26 promoter (see Methods S3). As shown in Figure 3(a–c), MYB26:GUS activity was undetectable before stage 10 in wild-type anthers. At stage 10, staining was observed in the tapetum, middle layer and endothecium (Figure 3a,d), and also in stamen filaments and in the pistil.
The same GUS staining pattern, but more intense, was observed at early stage 11, before endothecium lignification (Figure 3a,e), whereas GUS staining was mainly detectable in the stamen filament from late stage 11 (Figure 3a). In agreement with the above described quantitative RT-PCR data, MYB26:GUS activity was detectable at stage 9 (meiosis) in the afb1-3 background, mainly in the anther, where it was localized in tapetum, middle layer, endothecium, tetrads and procambium (Figure 3a,c). At stage 10, MYB26:GUS activity was still visible in the tapetum and middle layer, and less in endothecium (Figure 3d), while no GUS activity was detectable in the procambium. In addition, GUS staining is almost undetectable in the afb1-3 pistil from stages 10–12 (Figure 3a), possibly explaining why the transcript level of MYB26 was higher in whole flower buds from the wild-type line compared to those of afb1-3 at stages 10 and 11 of development. In tir1 afb2 afb3 anthers, MYB26:GUS activity was detectable at pre-meiotic and meiotic stages: at pre-meiotic stage 8, it was quite diffuse in the theca (Figure 3b), while at meiosis (stage 9), it was localized mainly in the tapetum and to a lesser extent in the middle layer, endothecium and procambium (as in the afb1-3 background) (Figure 3a,c); MYB26:GUS activity in the tir1 afb2 afb3 background at stage 10 was still very high in tissues surrounding the theca, mainly in tapetum, but was low in the filament and the pistil (Figure 3a). In contrast to what was observed in MYB26:GUS afb1 anthers, GUS staining was still detectable at stage 11 in the anther tissues, as shown in Figure 3(a,e).
Thus, in afb1-3 and tir1 afb2 afb3 mutants, which showed early endothecium lignification, there was an enhanced activity of the MYB26 promoter specifically at pre-meiotic and meiotic stages in the endothecium and in tissues surrounding the theca, suggesting that auxin controls the timing of endothecium lignification via MYB26.
To confirm this hypothesis, we treated wild-type inflorescences with 50 μm NAA in planta (Cecchetti et al., 2008), and analysed the MYB26 transcript level in flower buds at various developmental stages. As shown in Figure 3(f), MYB26 mRNA levels were comparably low in NAA- and mock-treated flower buds up to stage 10. In contrast, MYB26 mRNA levels were severely reduced in NAA-treated compared to mock-treated flower buds at stages 10 and 11, confirming that auxin exerts negative control over MYB26 expression. To verify the effects of NAA treatments on the activity of the MYB26 promoter in anthers, we treated MYB26:GUS inflorescences with NAA and analysed GUS staining at various develpmental stages. In agreement with quantitative RT-PCR data, no GUS staining was observed in NAA- and mock-treated anthers prior to stage 10, whereas only faint GUS staining was observed in all NAA-treated anthers at stages 10 and 11 (Figure 3g,h). In contrast, at the same stages, MYB26 promoter activity was clearly detectable in mock-treated (Figure 3g,h) MYB26:GUS anthers similar to untreated anthers shown in Figure 3a,d.
To provide further evidence that auxin acts via MYB26, we crossed afb1-3 and ms35/myb26 single mutants and analysed three independent double knockout afb1 myb26 lines (Figure S3a) phenotypically and histologically to assess their fertility and the timing of endothecium lignification (see Methods S3). As shown in Figure 4(a), afb1 myb26 flowers had indehiscent anthers like ms35/myb26 flowers (Dawson et al., 1999). Histological analysis revealed that, at stage 11, endothecium expansion had already taken place in wild-type anthers, but only partially in afb1 myb26 anthers (Figure 4b); the lignified thickenings that were observed at late stage 11 in wild-type (Figure 4b,c) were absent in afb1 myb26 anthers at this and subsequent stages (Figure 4b–d).
Thus, endothecium lignification does not occur in the afb1 myb26 double mutant, confirming that auxin acts through MYB26 to regulate this event.
Jasmonic acid controls stomium opening, but is not involved in the timing of endothecium lignification
Sanders et al. (2000) showed that stomium opening was delayed in mutants defective in the JA biosynthetic gene OPR3, but no data on the timing of endothecium lignification were provided.
To assess whether JA plays a role in endothecium lignification, we performed a comparative histological analysis of anthers from opr3 and wild-type flowers at various developmental stages. As shown in Figure 2(a), no endothecium lignification was observed either in wild-type or opr3 anthers before tapetum degeneration (stage 10), whereas lignified thickenings were clearly visible at stage 11 in anthers of both genotypes (Figure 2b). We also compared the transcript level of MYB26 during stamen development in opr3 and wild-type flower buds by quantitative RT-PCR from stages before 10–12, and did not observe any significant difference (Figure 2c). These data indicate that JA has no effect on the timing of endothecium lignification.
We then crossed the afb1-3 auxin-perception mutant and the opr3 JA-defective mutant. Flowers from three independent double knockout afb1 opr3 lines (Figure S3b) were phenotypically and histologically analysed to assess stomium opening and the timing of endothecium lignification. As shown in Figure 5(a), afb1 opr3 flowers had indehiscent anthers like opr3 flowers and unlike afb1-3 flowers (see Figure 1a), indicating that stomium opening does not occur in the absence of JA.
Histological analysis of afb1 opr3 flowers showed lignified thickenings in the endothecium of approximately 10% of double mutant anthers at stage 10 of development (Figure 5b), as previously observed in afb1-3 single mutant anthers, indicating that the early endothecium lignification caused by defective auxin perception also occurs in the absence of JA. In addition, in afb1 opr3 flower buds, the level of MYB26 transcript is higher than in wild-type before stage 10, as described for the afb1-3 single mutant, whereas the peak at stage 10 is slightly higher than that for the wild-type (Figure 5c). Thus, MYB26 expression is enhanced at early stages in afb1 opr3 flowers showing precocious endothecium lignification, as previously observed in afb1-3 flower buds.
These data indicate that endothecium lignification is controlled by auxin and is independent of JA, whereas auxin has no effect on stomium opening in the absence of JA.
Auxin controls JA production
It has been shown that the JA biosynthetic genes DAD1 and OPR3 are expressed in stamens during late development, and that the JA content peaks in flower buds at stages 11 and 12 (Nagpal et al., 2005) when auxin concentration decreases (Cecchetti et al., 2008; this paper). To analyse the relationship between auxin and JA biosynthesis, we compared the expression of DAD1 and OPR3 in wild-type, afb1-3 and tir1 afb2 afb3 flower buds.
As shown in Figure 6(a–d), OPR3 and DAD1 are expressed in wild-type flower buds up to stage 10, and their transcript levels increase at stage 11 mainly in Col-0; at stage 12, OPR3 and DAD1 transcript levels slightly decrease in Col-0 flower buds only. In afb1-3 flower buds (Figure 6a,c), the transcript levels of DAD1 and OPR3 are comparable to those of the wild-type up to stages 10 and 11, but are significantly higher at stage 12 (2.5 and 1.3 times higher than wild-type for DAD1 and OPR3, respectively). In tir1 afb2 afb3 flower buds (Figure 6b,d), DAD1 and OPR3 transcript levels are comparable to those of the wild-type up to stage 10 but are significantly higher at stage 11 (3 and 2.3 times higher than wild-type for DAD1 and OPR3, respectively); at stage 12, the DAD1 transcript level is slightly but significantly higher than that of the wild-type, whereas the OPR3 transcript level is comparable that of the wild-type. These data suggest that auxin exerts a negative effect on expression of these JA biosynthetic genes at specific late stages, i.e. before stomium opening (stages 11 and 12). To verify that the increase in JA biosynthetic gene transcripts corresponds to an actual increase in JA concentration in auxin mutants, we measured JA content in afb1-3, tir1 afb2 afb3 and wild-type flower buds at stages 11 and 12 (pooled together), when the peak of JA production was observed in wild-type flowers (Methods S4). As shown in Figure 6(e), JA content is comparable to that of the wild-type in afb1-3 flower buds, but is approximately two times higher in tir1 afb2 afb3 flower buds.
To confirm the negative effect of auxin on the expression of DAD1 and OPR3 at stages 11 and 12, wild-type inflorescences were treated with 50 μm NAA or mock-treated, and analysed for DAD1 and OPR3 transcript levels. As shown in Figure 6(f), mock-treated flower buds similar to untreated ones (Figure 6a,c) had comparable levels of DAD1 and OPR3 transcripts at all stages. In contrast, when treated with NAA, flower buds up to stage 10 showed DAD1 and OPR3 transcript levels that were comparable to those of mock-treated buds, but the transcript levels at stage 11 were significantly lower (1.5 and 1.8 times, respectively) than mock-treated controls; at stage 12, the OPR3 transcript level was significantly lower (1.8 times) than in mock-treated flower buds, but the DAD1 transcript level was not significantly lower.
These data support the notion that auxin negatively regulates the biosynthesis of JA in anthers before stomium opening at stages 11 and 12 of development, and suggest that JA concentration has a role in the timing of anther dehiscence.
High JA concentrations induce precocious anther dehiscence
To assess whether an increase in JA concentration at stages 11 and 12 causes precocious anther dehiscence, we treated opr3 inflorescences with various JA concentrations and measured the percentage of dehiscent anthers at various stages. Mutant opr3 inflorescences were mock-treated or treated with 1, 3.2 or 6.4 mm JA, and the number of dehiscent anthers after 48 h was analysed in flower buds at developmental stage 12 initial (12i) before rapid filament elongation, stage 12 middle (12 m) during rapid filament elongation, and stage 13, when this elongation phase is over. As shown in Figure 7, mock-treated inflorescences showed no dehiscent anthers at stages 12i and 12 m, and very few dehiscent anthers at stage 13. Dehiscent anthers were observed at stage 12i in inflorescences treated with 6.4 mm JA but not those treated with 1 or 3.2 mm JA; at stage 12 m, the percentage of dehiscent anthers increases as JA concentration increases; at stage 13, nearly all anthers were dehiscent in all JA-treated buds. This data indicate that precocious anther dehiscence may be caused by high JA concentration, suggesting that auxin controls the timing of anther dehiscence by modulating JA biosynthesis.
As auxin controls the timing of anther dehiscence by modulating JA biosynthesis, we wished to confirm whether, as previously suggested (Cecchetti et al., 2008), auxin does indeed act on the timing of anther dehiscence by controlling endothecium lignification.
We utilized afb1 opr3 double mutants, which show precocious endothecium lignification but indehiscent anthers due to the opr3 mutation (Figure 5a,b). afb1 opr3 inflorescences were treated with 1, 3.2 or 6.4 mm JA, and the percentage of dehiscent anthers was analysed at stages 12i, 12 m, 13 and compared with the results for JA-treated opr3 inflorescences.
As shown in Figure 7, nearly all anthers were dehiscent at stage 13 in afb1 opr3 JA-treated buds, similar to opr3 JA-treated inflorescences; in contrast, at stage 12 m, a significantly higher percentage of dehiscent anthers was observed in JA-treated afb1 opr3 inflorescences compared to opr3 inflorescences. Similarly, at stage 12i, a high percentage of dehiscent anthers was observed in all JA-treated afb1 opr3 inflorescences, whereas dehiscent anthers were observed only in opr3 inflorescences treated with 6.4 mm JA at stage 12i. These data confirm that the timing of endothecium lignification contributes to regulation of the timing of anther dehiscence.
Anther dehiscence, pollen maturation and filament elongation are late processes in stamen development. In a previous paper, we showed that auxin negatively controls anther dehiscence in Arabidopsis, and suggested that a decrease in auxin concentration triggers dehiscence by acting on endothecium lignification (Cecchetti et al., 2008).
In this paper, we further clarify the role of auxin in anther dehiscence, and establish a relationship between auxin and JA, another hormone that is known to be involved in anther dehiscence (Xie et al., 1998; Ishiguro et al., 2001; Nagpal et al., 2005).
By measuring IAA concentration in Arabidopsis anthers at various developmental stages, we found an increase in IAA concentration that begins at meiotic and pre-meiotic stages (8 and 9), leading to a peak of IAA at the initial stage of late development (stage 10), followed by a significant decrease in IAA content when endothecium lignification has occurred at stage 11. A further decrease in IAA content is observed when the anthers become bi-locular, at stage 12. These results are in good agreement with data previously obtained by monitoring DR5 activity (Cecchetti et al., 2008), and support the notion that a local auxin minimum is required to trigger late developmental stages in anthers. Similarly, a decrease in auxin content has been shown to be necessary during specification of the valve layer when Arabidopsis fruit opening takes place (Sorefan et al., 2009).
To clarify how auxin regulates endothecium lignification, we utilized the auxin receptor mutant afb1-3 because it is altered only in anther dehiscence, whereas mutants of other auxin receptors are also altered either in pollen maturation and filament elongation (afb2-3), or are not altered in anther dehiscence (afb3-4 and tir1-1). These diverse phenotypes are in agreement with data published by Parry et al. (2009), which suggest that auxin receptors have different functions in given tissues due to different affinities with different AUX/IAA proteins.
In our analysis, we compared the results obtained for afb1-3 with those for the triple mutant tir1 afb2 afb3, which expresses AFB1 but not other receptors and shows a higher percentage of early dehiscing anthers.
By histological fluorescence analysis, we determined that endothecium lignification occurs at stage 10 in both afb1-3 and tir1 afb2 afb3 mutant anthers, i.e. earlier than in wild-type anthers, where lignin deposition is observed at stage 11. These data, together with phenotypic analysis of multiple and single auxin-perception mutants (Cecchetti et al., 2008; this paper), indicate that all four auxin receptors contribute to regulation of anther dehiscence by acting on the timing of endothecium lignification, with only AFB1 acting specifically on anther dehiscence.
We showed that auxin exerts its effect by acting negatively on expression of the MYB26 gene, which encodes an MYB transcription factor that is required for endothecium lignification (Yang et al., 2007). By quantitative RT-PCR and analysis of the expression of a transcriptional MYB26:GUS fusion, we found that MYB26 is mainly expressed at stage 11 in wild-type anthers, and at a low level at late stage 10 just before and during endothecium lignification, in agreement with results obtained byYang et al. (2007), but is more actively expressed than in wild-type at early stages in afb1-3 (stage 9) and tir1 afb2 afb3 (stage 8) mutant anthers.
Furthermore, we show that NAA-treated flower buds show a severe reduction in MYB26 transcript levels at stages 10 and 11, and only faint GUS staining is detectable in anthers from NAA-treated MYB26:GUS inflorescences. These findings indicate that MYB26 transcription is induced at late stage 10 and stage 11 in wild-type anthers, and we propose that this expression is negatively controlled by auxin. As auxin content increases in anthers from stage 9 to 10 (this paper), a reduction in auxin content should occur, at least in endothecial cells, at late stage 10.
However, we cannot rule out the possibility that MYB26 expression is under additional transcriptional control independently of auxin. We obtained further evidence that auxin controls endothecium lignification through MYB26 by generating an afb1 myb26 double mutant. Phenotypic and histological analysis of afb1 myb26 flowers showed indehiscent anthers in this double mutant due to lack of endothecium lignification, suggesting that MYB26 is required for auxin regulation of the timing of endothecium lignification. The alternative explanation of the genetic data, i.e. that AFB1 acts downstream of MYB26, is in contrast with the fact that AFB1 is expressed earlier than MYB26 in anthers, indicating that AFB1 acts via MYB26 to regulate endothecium lignification.
Interestingly, and in agreement with the results obtained by Yang et al. (2007), GUS staining in MYB26:GUS anthers was observed not only in endothecium cells but also in the tapetum. It is possible that expression in the tapetum is subsequently repressed by post-transcriptional regulation. Alternatively, MYB26 expression in the tapetum may be linked to exine biosynthesis, which takes place in this tissue. Indeed, MYB transcription factors have been linked to regulation of phenylpropanoid metabolism, which is involved either in lignin biosynthesis (Yang et al., 2007) or in synthesis of sporopollenin, from which exine is made (Dobritsa et al., 2011).
In this study, we also analysed the relationship between auxin and JA, which is also involved in anther dehiscence. We tested the role of JA in endothecium lignification using a mutant defective in the JA biosynthetic gene OPR3 that is specifically expressed in anthers (Stintzi and Browse, 2000). We found that JA is not involved in controlling endothecium lignification, as the timing of lignin deposition as well as MYB26 transcript levels in opr3 flower buds were comparable to wild-type. Ishiguro et al. (2001) proposed that JA controls stomium opening, the final event of anther dehiscence, and Sanders et al. (2000) showed delayed and defective stomium opening in opr3 mutants. However, these authors did not assess whether JA controls stomium opening by controlling endothecium lignification or whether these two developmental processes are independently controlled. We show here that the latter is the case: anthers of afb1 opr3 double mutant flowers show early endothecium lignification like afb1-3 anthers, but are indehiscent as in the opr3 mutant, which is defective in stomium opening. These findings demonstrate that auxin has no effect on stomium opening in the absence of JA. However, the early dehiscence phenotype of afb1-3 and tir1 afb2 afb3 mutants suggests that auxin controls stomium opening through JA.
We show that, during late stages of flower development, auxin negatively controls JA biosynthesis, acting on both stamen-specific JA biosynthetic genes DAD1 and OPR3. In tir1 afb2 afb3 mutant flower buds, there is an increase in the level of DAD1 and OPR3 transcripts, resulting in increased production of JA during late stamen development. The increase in JA content in these early-dehiscent mutants is consistent with the phenotype of JA-defective mutants, which show indehiscent or late dehiscent anthers. The increase in JA production was not detectable in afb1-3 flower buds. However, this mutant showed fewer early-dehiscing anthers, and a lower level of DAD1 and OPR3 transcripts than tir1 afb2 afb3, possibly due to a limited increase in JA production that may go undetected under our experimental conditions. We provide further evidence that auxin negatively controls JA biosynthesis by treating wild-type inflorescences with NAA and showing that DAD1 and OPR3 transcript levels are reduced in flower buds specifically at stages 11 and 12 of development.
Future work is necessary to clarify which gene(s) mediate the negative control of auxin on JA biosynthesis in late stages of flower development. Possible candidates may be the AS1 and AS2 transcription factors, which repress the class 1 KNOX genes, which in turn repress DAD1 (Tabata et al., 2010). Another candidate may be ARF5/MP: recently, a negative role for ARF5 on JA production has been proposed (Garrett et al., 2012), in contrast to the well-known positive control of ARF6/ARF8 (Nagpal et al., 2005; Tabata et al., 2010).
We also show here that JA has no effect on auxin synthesis and accumulation in anthers. JA production peaks in flower buds when auxin concentration decreases but before auxin biosynthetic genes are turned off. To rule out a possible positive feedback effect of JA in negatively regulating auxin synthesis, we showed that, in opr3 inflorescences, the transcript levels of the YUC2 and YUC6 auxin biosynthetic genes and DR5 activity are comparable to the wild-type (Figure S4 and Data S1).
Thus, our data indicate that auxin negatively regulates endothecium lignification and JA concentration. We also show that both JA levels and the timing of endothecium lignification contribute to determining when anther dehiscence occurs. By treatment of opr3 inflorescences with exogenous JA, we showed that a high JA level causes precocious anther dehiscence, and by comparing the timing of anther dehiscence of JA-treated opr3 and afb1 opr3 flower buds, we confirmed our previous suggestion that precocious endothecium lignification causes early anther dehiscence (Cecchetti et al., 2008).
In summary, as represented in Figure 8, we suggest that, in Arabidopsis anthers, an auxin maximum at early floral stage 10 blocks premature endothecium lignification via repression of MYB26. Upon a decrease in auxin concentration at floral stages late 10-11, repression of MYB26 is released: this allows MYB26 to trigger endothecium lignification. Auxin reduction also allows higher expression of DAD1 and OPR3, which causes an increase in JA concentration, leading to stomium breakage and allowing completion of the anther dehiscence process (stage 13).
Plant treatments and sample collection
Arabidopsis wild-type and mutants are described in Methods S5. Anthers collected for IAA measurements were severed from flower buds, immediately frozen and divided into five groups according to the stage: pre-meiotic, meiotic, uninucleate microspore, pollen mitosis II and tricellular pollen grains. The stage was assessed by squeezing one anther per flower bud in acetic orcein solution to distinguish between pre-meiotic and meiotic stages or staining with DAPI (4',6-diamidino-2-phenylindole) solution (Cecchetti et al., 2004) to distinguish between pollen mitosis I and pollen mitosis II stages. The tricellular pollen grain stage was defined as flower buds with white petals protruding past sepals corresponding to stage 12. Flower bud developmental stages were also checked by histological analysis of random flower buds.
Flower buds collected for JA measurements and for quantitative RT-PCR analysis were severed from three plants for each genotype and immediately frozen. Stages <10 (Figures 2c, 3f, 5c) and ≤10 (Figure 6a-d,f) were collected after removing all flower buds from stage 10 onwards and from stage 11 onwards, respectively, from the inflorescences. Stages 11 and 12 for JA measurement were obtained by pooling equal amounts of flower buds at stages 11 and 12. For each genotype, 20 mg (dry weight) of pooled flower buds in three replicates were collected.
In planta NAA treatment of MYB26 inflorescences was performed as described previously (Cecchetti et al., 2008). Twenty-four hours after NAA treatments, wild-type flowers at stages <10, 10, 11 and 12 or ≤10 and 11 (Figure 3f) and 12 (Figure 6f) were collected for quantitative RT-PCR analysis. JA treatments were performed by brushing inflorescences of three plants with 1, 3.2 or 6.4 mM methyl jasmonate and water/0.05% Tween (Figure 7).
Quantitative RT-PCR analysis
RNA was extracted from 50 mg flower buds at the indicated stages of development or from inflorescences, and reverse-transcribed as previously described (Cecchetti et al., 2007) (Methods S6).
SYBR Green-based quantitative assays were performed as described by Savona et al. (2012) using a Bio-Rad iCycler iQ (http://www.bio-rad.com). All quantifications were performed in triplicate.
Two-tailed and one-tailed Student's t tests were used to evaluate statistical significance. All the statistical analyses were performed using Graph Pad Prism 5 (Graph Pad Software Inc., http://www.graphpad.com).
We are grateful to Mark Estelle and Geraint Parry for kindly providing tir1 and afb single mutants, Zoe Wilson for the ms35/myb26 mutant and the proMYB26:GUS construct, and John Browse for the opr3 mutant. We thank Hannah Florence for JA measurements performed at the University of Exeter mass spectrometry facility, and Gun Lövdahl for excellent technical assistance in quantification of IAA.