SEARCH

SEARCH BY CITATION

Keywords:

  • Arabidopsis;
  • At4g35420;
  • DFR-like gene;
  • male sterility;
  • pollen wall development

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • • 
    Arabidopsis contains only one functional dihydroflavonol 4-reductase (DFR) gene, but several DFR-like genes encoding proteins with the conserved NAD(P)H binding domain. At4g35420, named DRL1 (Dihydroflavonol 4-reductase-like1), is a closely related homolog of the rice anther-specific gene OsDFR2 reported previously.
  • • 
    Two T-DNA mutants (drl1-1 and drl1-2) were found to have impaired pollen formation and seed production. Histological analysis revealed defective microspore development after tetrad release in both mutants. Microspore walls were found to rupture, releasing the protoplasts which eventually degenerated. The DRL1 promoter is anther-specific in closed flower buds. Promoter-GUS analysis in transgenic Arabidopsis revealed expression in tapetum, tetrads, and developing microspores, but not in mature anthers. Enhanced yellow fluorescent protein (EYFP)-localization analysis demonstrated that DRL1 is a soluble cytosolic protein that may also be localized in the nucleus.
  • • 
    Restoration of male fertility and seed formation was only achieved by a native promoter-DRL1 construct, but not by a 35S-DRL1 construct, demonstrating the importance of spatial and temporal specificities of DRL1 expression.
  • • 
    DRL1 may be involved in a novel metabolic pathway essential for pollen wall development. DRL1 homologs were identified as anther- and floral-specific expressed sequence tags from different species, suggesting that DRL1 may have a conserved functional role in male fertility in flowering plants.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Dihydroflavonol 4-reductase (DFR) is a NADPH-dependent enzyme converting dihydroflavonols to their corresponding leucoanthocyanidins, which are necessary for the formation of anthocyanins and proanthocyanidins (condensed tannins) in higher plants. The biosynthesis pathway of flavonoids is one of the best characterized metabolic pathways in nature. Most key enzymes in the pathway have been determined and their corresponding genes have been cloned and functionally characterized (Winkel-Shirley, 2001). Subsequently, numerous plant sequences generated from different sequencing projects have been annotated as homologs of flavonoid structural genes. In Arabidopsis, there are at least five genes encoding DFR-like proteins (Yuan et al., 2007). However, TT3 represents the only DFR-encoding gene in Arabidopsis (Winkel-Shirley, 2001) and the tt3 mutants are completely deficient in the accumulation of anthocyanin pigments and condensed tannins (Shirley et al., 1995). Other DFR-like genes have been demonstrated to have different biochemical and physiological functions in Arabidopsis. For example, BAN is a structural gene encoding anthocyanidin reductase, which converts anthocyanidins to their corresponding 2, 3-cis-flavan-3-ols for the biosynthesis of proanthocyanidins (Xie et al., 2003). On the other hand, BEN1 was recently demonstrated to be involved in regulating the concentrations of several brassinosteroids, including typhasterol, castasterone, and brassinolide (Yuan et al., 2007). However, the biochemical and physiological functions of the other DFR-like genes in Arabidopsis remain unknown.

We have previously reported a rice DFR-like gene (OsDFR2), which was identified by a cDNA subtraction strategy using anthers from wild-type (AN-N) and thermosensitive male-sterile (AN S-1) mutant indica rice lines (Yau et al., 2005). In situ hybridizations revealed that the expression of OsDFR2 was anther-specific during early microsporogenesis. In the AN S-1 mutant, OsDFR2 expression was down-regulated fourfold under sterility-inducing conditions. These findings suggested that the OsDFR2-derived metabolites, which may be flavonoids, are likely to be necessary for anther development and male fertility in rice. In fact, a specific flavonoid class, flavonols, has been demonstrated to be essential for pollen germination in plant species such as maize and petunia (Coe et al., 1981; Mo et al., 1992; van der Meer et al., 1992). Mutations in the chalcone synthase gene, which encodes the key branch-point enzyme for flavonoid biosynthesis, resulted in the formation of nonfunctional white pollen in these plants. Flavonols were found to be deficient in the mutant pollen and their fertility was completely restored by exogenous application of flavonols (Mo et al., 1992). In Arabidopsis, however, flavonoids do not appear to be critical for male fertility since the different transparent testa (tt) mutants are completely fertile, producing yellow or pale-brown seeds.

The annotation of OsDFR2 as a DFR-like gene was primarily based on sequence homology and no functional data are available for the rice gene. It has been demonstrated that secondary metabolic enzymes are functionally interchangeable between distantly related plants, such as Arabidopsis and maize (Dong et al., 2001). Thus, the Arabidopsis tt mutants are useful for defining the biochemical functions of structural gene homologs from diverse plant species (Yu et al., 2005; Shih et al., 2006). To this end, we over-expressed OsDFR2 in transgenic Arabidopsis tt3 mutants, but the flavonoid deficiencies were not complemented (C. H. Shih & C. Lo, unpublished), indicating that the encoded rice protein does not function as a typical DFR enzyme. It is interesting to note that OsDFR2 not only shows homology to DFR but also to cinnamoyl-CoA reductases and cinnamoyl alcohol dehydrogenases in plants (Yau et al., 2005). In addition, the rice protein also shares homology with mammalian 3-β-hydroxysteroid dehydrogenase, bacterial cholesterol dehydrogenase, and UDP-galactoase-4-epimerase. Hence, OsDFR2 may represent a NADPH-dependent reductase, dehydrogenase, or epimerase that is involved in some novel metabolic pathways critical for male reproductive development in rice.

We took advantage of the comprehensive collections of publicly available Arabidopsis T-DNA mutants to identify mutations that may be useful for us to understand the biochemical and physiological functions of OsDFR2. Two T-DNA insertion alleles were identified in At4g35420, which is the closest homolog of OsDFR2 in Arabidopsis (63% amino acid sequence identity). During propagation of the homozygous mutant lines for seed collection, we observed drastic reduction of seed formation, prompting us to launch an extensive investigation into At4g35420, which we now designate as DRL1 (dihydroflavonol 4-reductase-like1). Here, we report the characterizations of the two insertion mutants, drl1-1 and drl1-2, which showed impaired pollen development leading to different degrees of male sterility. Histological examinations revealed the degeneration of microspores following their release in the mutant anthers as well as the specific expression of the GUS reporter gene driven by the DRL1 promoter in tapetum and developing microspores. Genetic complementation experiments confirmed that DRL1 is essential for pollen development and male fertility in Arabidopsis. In addition to OsDFR2, there are a number of conserved DRL1 homologs (> 60% sequence identity at amino acid level) identified as expressed sequences derived from anther and floral tissues in diverse flowering plant species, suggesting a conserved function of a DFR-like protein in male fertility.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Mutants, plant growth, and PCR genotyping

Arabidopsis thaliana (L.) Heynh. accession Columbia (Col-0) was used as the wild-type control in this study. DRL1 T-DNA insertion lines were obtained from the French National Institute for Agricultural Research (drl1-1, stock no. Flag_019D03) and the Arabidopsis Biological Resource Center (drl1-2, stock no. CS837358). Seeds were germinated in the dark for 3 d at 4°C. All plants were grown on soil and kept under long-day conditions (16 h light : 8 h dark) at 22°C in growth chambers. PCR genotyping was used to identify homozygous insertion lines. For drl1-1 mutant, plant-specific LP primer CL208 (5′-ATTGGATGAGCTCTTGATTGC-3′) and RP primer CL209 (5′-ATCACATGACTCGTTTTTGGG-3′), and T-DNA LB primer CL183 (5′-CGTGTGCCAGGTGCCCACGGAATAGT-3′) were used. For drl1-2 mutant, plant-specific LP primer CL194 (5′-GAAGAAACTTGCGCACCTATG-3′) and RP primer CL195 (5′-CGAGCAGATAGGAATGAAACG-3′), and T-DNA LB primer CL184 (5′-TTCATAACCAATCTCGATACAC-3′) were used. For homozygous mutants, only LB and RP primers resulted in a PCR product. For wild-type plants, only the plant-specific primers resulted in a PCR product. For heterozygous plants, PCR reactions were positive for both combinations of primer sets. PCR amplifications were programmed as follows: pre-incubation (95°C for 10 min), followed by 30 cycles of denaturation (95°C for 30 s), annealing (55°C for 30 s) and extension (72°C for 1 min), followed by a final extension step at 72°C for 7 min. The precise T-DNA insertion sites in drl1-1 and drl1-2 were determined by sequencing the PCR products.

Microscopic investigations of anther development

Anthers were removed from flowers and mounted in a drop of Alexander's (1969) stain under a cover glass to study the abundance of pollen grains. For histological examinations, flower buds of different stages were fixed in 2.5% glutaraldehyde and 1.6% paraformaldehyde buffered with 0.05 m phosphate buffer (pH 6.8) for 24 h at 4°C. After fixation, the flowers were dehydrated using an ethanol series and embedded in Historesin as described previously (Yeung, 1999). Serial sections of thickness 3 µm were sliced using a microtome and stained with periodic acid-Schiff (PAS) reaction for total carbohydrates, and counterstained with amido black 10B for proteins or toluidine blue O for general histological organization (Yeung, 1984), followed by examination under a Leitz photomicroscope. In order to confirm the nature of the PAS-positive granules, the starch staining method detailed by Solís et al. (2008) was also used.

RNA experiments

Total RNA was extracted from different organs by the Trizol method (Invitrogen, Carlsbad, CA, USA). RNA samples were DNase-I-treated (Invitrogen) and reverse-transcribed by M-MLV reverse transcriptase (Promega, Madison, WI, USA). Primers CL362 (5′-CGC GGA TCCATG GAT CAA GCA AAG GGA AAA G-3′) and CL363 (5′-GGA CTA GTTTAT GGA AGA ACA GTA GAT AA-3′) were used for amplification of DRL1, while primers HEL376 (5′-GCCCAGAAGTCTTGTTCCAG-3′) and HEL377 (5′-TTGGAGATCCACATCTGCTG-3′) were used for amplification of an actin gene. The 3′-transcript end of the drl1-1 allele was mapped by a rapid amplification of cDNA ends (RACE) procedure, as described by the manufacturer (Roche Molecular Biochemicals), followed by DNA sequencing. 3′-specific cDNA was amplified by PCR using a gene-specific primer (CL331, 5′-CCGATGTCCTCGGATTACTG-3′) upstream of the insertion site and an adapter primer (HEL373, 5′-CCACGCGTCGACTAGTACTT-3′). The PCR amplification program was the same as described earlier.

DRL1 promoter analysis

A 2.0 kb region upstream of the DRL1 initiation codon was amplified from wild-type genomic DNA using the primers CL334 (5′-CCGGTATAAATGGAATCACACC-CG-3′) and CL335 (5′-GTCAAGTTCCTCTTACAGCAGTAC-3′). Following sequence confirmation, the PCR fragment was used to construct a DRL1 promoter-GUS fusion gene in a pCAMBIA1300-derived binary vector. Wild-type plants were transformed with the Agrobacterium tumefaciens strain GV3101 harboring the fusion gene construct, using the floral dip method (Clough & Bent, 1998). Transformants were selected on MS plates containing 25 µg ml−1 hygromycin, and T1 seeds were germinated for GUS expression analysis. Different tissues collected from T1 plants were incubated with 5 bromo-4-chloro-3-indole glucoronide (X-gluc) in 50 mm phosphate buffer (pH 7.0), 0.5% Triton X-100, and 0.5 mm ferrocyanide for 12–24 h. GUS staining patterns were recorded using a Leica M125 stereomicroscope (Leica, Milton Keynes, UK) equipped with a Leica DFC500 digital camera. For histological examinations, the stained flowers were fixed and processed as detailed earlier after the X-gluc staining. The stained sections were examined without an applied cover glass in order to visualize the blue staining product.

Cellular localization of DRL1

The DRL1 coding region was cloned in frame with an enhanced yellow fluorescent protein (EYFP) gene under the control of the 35S promoter in a pCAMBIA-based binary vector (H. J. Liu, Zhejiang Academy of Science, China) to generate the fusion construct 35S-DRL1-EYFP. The DRL1 coding region was amplified with the primers CL394 (5′-GAG CTC ATG GAT CAA GCA AAG GGA AAA G-3′) and CL395 (5′-GGT ACC TGG AAG AAC AGT AGA TAA ATA G-3′) using a full-length cDNA clone (RAFL14-03-G07) obtained from the RIKEN Bioresource Center (Ibaraki, Japan) as the template. Particle bombardment was used to introduce the fusion plasmid into onion epidermal cells by a Biolistic PDS-1000/He system (Bio-Rad, Hercules, CA, USA). Gold particles (1.0 µm) were coated with the plasmid and a helium pressure of 1100 psi was employed. About 360 µg of gold particles were coated with 5 µg DNA in one bombardment. The target distance between the stop screen and onion scales was maintained at 6 cm. The bombarded samples were kept in darkness at 22°C for 15 h and subsequently examined under the Leica stereomicroscope. To induce plasmolysis, a layer of onion epidermal cells was removed from the bombarded scales and incubated in 1 m sucrose solution for 20 min.

Complementation constructs

A Kpn I/Sal I fragment containing the DRL1 coding region was released from the full-length cDNA clone. The DNA fragment was inserted between the CaMV 35S promoter and the nopaline synthase 3′-terminator in a pCAMBIA-derived binary vector (Yu et al., 2005). To generate the native promoter complementation construct, the 35S promoter region was replaced by the 2.0 kb upstream DRL1 genomic fragment (described earlier). The two complementation constructs were transformed independently into heterozygous drl1-2/+ insertion lines by the floral dip method as described earlier.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Phenotypic characterization of drl1-1 and drl1-2 insertion mutants

DRL1 (At4g35420) contains six exons and five introns, and an open reading frame of 981 bp encoding a protein with 326 amino acids. We have obtained two independent T-DNA insertion mutants for DRL1. In drl1-1, the insertion occurred at 127 bp upstream of the predicted stop codon, while in drl1-2 the insertion was found within intron 3 (Fig. 1a). The mutants were indistinguishable from wild type in the heterozygous state, and hence the mutation alleles are both recessive. On the other hand, homozygous lines for the two insertion mutants showed different degrees of male sterility. Flowers of drl1-1 and drl1-2 homozygous plants are of normal size and shape, and filaments of stamens elongate as in the wild type (data not shown). However, no pollen grains were found on the surfaces of anthers and stigma in the mutant flowers (data not shown). Most mature anthers of drl1-1 homozygous plants were empty, although some had a little pollen, which stained purple with Alexander's stain (Fig. 1b), indicating that they are viable. However, the drl1-1 pollen appeared irregular in shape and less intact compared with the wild-type pollen. Consequently, the drl1-1 plants displayed a partial fertility phenotype with small number of seeded siliques (Fig. 1c). On the other hand, mature anthers of drl1-2 homozygous plants were completely devoid of pollen (Fig. 1b) and the mutant plants showed undeveloped siliques without seed formation (Fig. 1c). Hence, the drl1-2 homozygous mutants are completely sterile and this mutation allele could only be propagated in the heterozygous state. Back-crossing of the drl1-2 mutant stigmas with wild-type pollen grains resulted in normal seed fertility (Fig. 1d), indicating that the female reproductive tissues were functional in the homozygous plants. Anther dehiscence was normal in both mutant alleles (data not shown), suggesting that defects in pollen development were the major cause of male sterility.

image

Figure 1. Characterization of the drl1 T-DNA mutants. (a) Diagrammatic representation of the two insertion alleles, drl1-1 and drl1-2. (b) Alexander's staining of wild-type and mutant Arabidopsis anthers. Most drl1-1 and all drl1-2 mature anthers are empty. Some drl1-1 mature anthers contain a little pollen, which stained purple but appeared irregular in shape. Bars, 100 mm. (c) Comparison of silique formation in wild-type and drl1 mutant plants. Note the short siliques (arrows) in the mutants and that they are seedless. Some seeded siliques (arrowheads) that are partially elongated are formed in the drl1-1 mutants. (d) Formation of normal silique (arrow) in a drl1-2 mutant plant after cross-pollination with wild-type pollen.

Download figure to PowerPoint

Flowers of different stages from wild-type and mutant plants were fixed, embedded, and sectioned for histological examination of reproductive development. In wild-type anthers, meiotic division proceeded normally with the formation of microspore tetrads (Fig. 2a). The microspores then started to expand and separated from one another with a distinct cell wall (Fig. 2b). Afterwards, they continued to increase in size and the cell walls thickened (Fig. 2c). As development progressed, large vacuoles and sculptured exine walls were readily observed in the microspores (Fig. 2d). Microgametogenesis then began as prominent starch granules accumulated in the microspore cytoplasm (Fig. 2e). In the mature pollen grains during the time of dehiscence, the generative cell was embedded within the tube cell and the tapetum was completely degenerated (Fig. 2f).

image

Figure 2. Arabidopsis wild-type anther development. (a) Tetrad formation; the tetrads are surrounded by a tapetum (*). (b) Microspores enlarge and separate from one another. A thin cell wall (arrow) enveloping each microspore can be seen. (c) Microspores take on a rounded appearance with a prominent wall (arrow). (d) Microspores enlarge further as a result of the process of vacuolation (v). Sculptured exine walls (arrows) can be readily observed at this stage. (e) At the time of microgametogenesis, microspore cytoplasm increase in density and prominent starch grains (arrow) begin to accumulate. (f) Mature pollen grains at the time of dehiscence. The tapetum has degenerated and is no longer visible. Tapetum (*). All bars, 10 mm.

Download figure to PowerPoint

In the drl1-2 homozygous mutants, the initial anther development was normal up to the completion of meiosis with the formation of microspore tetrads (Fig. 3a). Although the microspores rounded up and separated from one another, they remained small (Fig. 3b). Moreover, a distinct sculptured cell wall was absent in these microspores. As development progressed, the walls began to rupture, releasing the protoplasts into the anther locules (Fig. 3c). The protoplasts then increased in size with the accumulation of starch granules (Fig. 3d) and gradually occupied the entire space of the anther locules (Fig. 3e). As the flowers matured, the protoplasts degenerated rapidly, leaving remnants of cytoplasm and starch granules (Fig. 3f). Similar to the wild type, the mutant tapetum remained intact during the course of anther development and it became degenerated in the mature anthers. Pollen development in the drl1-1 mutant was essentially the same as it was observed in the drl1-2 mutant. However, some drl1-1 mature anther locules were found to contain a few pollen grains (not shown), which apparently resulted in the partial fertility phenotype in the mutant plants (Fig. 1c). It is important to note that embryo sac sections from both mutant flowers appeared normal (photo not shown), further confirming that the female reproductive functions were not affected by the drl1 mutations.

image

Figure 3. drl1-2 anther development. (a) Tetrad formation appears normal; the tetrads are surrounded by a tapetum (*). (b) Microspores round up and separate from one anther. They remain small and have a dense cytoplasm. (c) A defined exine wall is not present in the enlarging microspores. Rupture of walls occurs in some microspores (arrowhead), releasing the protoplasts into the anther locule. (d) The protoplasts (arrowhead) expand with the accumulation of prominent starch granules (arrow). (e) The entire anther locule is filled with protoplasts and wall remnants. (f) As the anther matures, the microspore protoplasts have degenerated, leaving behind the starch deposits (arrow). A tapetum is no longer visible at this stage. Tapetum (*). All bars, 10 mm.

Download figure to PowerPoint

Expression analysis of DRL1 in Arabidopsis

To better understand the physiological function of DRL1 in Arabidopsis, RT-PCR experiments were conducted with RNA samples prepared from different tissues in wild-type plants. Transcripts of DRL1 were not detected in 2-wk-old seedlings, cauline leaves, or inflorescence stalk, but only in a mixture of young and mature flowers (Fig. 4), indicating that the gene expression is specific in reproductive tissues. A null mutation in DRL1 was expected in the drl1-2 mutant, as the T-DNA insertion occurred in the middle of the gene (Fig. 1a). On the other hand, the reduced fertility phenotype of drl1-1 mutant suggested that DRL1 may be partially functional in the plant. RACE experiments were performed to map the 3′-end of the drl1-1 transcript, which was found to be polyadenylated and chimeric, containing the endogenous DRL1 sequence up to the insertion site, followed by sequence derived from the left border of the T-DNA (data not shown). The wild-type DRL1 encodes a protein of 326 amino acids in length. The predicted DRL1-1 mutant protein would be truncated at the 284th amino acid with a C-terminal extension of 15 amino acids derived from the T-DNA sequence. The truncation (44 amino acids) might have reduced the stability and enzyme activity of the mutant protein.

image

Figure 4. RT-PCR analysis of DRL1 and actin transcript levels in different Arabidopsis (Col-0) tissues. 2W, 2-wk-old seedlings; R, roots; RL, rosette leaves (4 wk old); CL, mature cauline leaves; St, stem; F, mixed flowers of different stages; Si, siliques; C, control (no RNA).

Download figure to PowerPoint

Spatial and temporal expression of DRL1 was examined in transgenic plants carrying the GUS reporter gene driven by a 2.0 kb upstream region of the putative DRL1 promoter. We have analyzed > 10 independent lines and the expression patterns were found to be consistent. X-gluc staining of whole plants from the transgenic lines demonstrated that GUS activities were not present in vegetative tissues (data not shown). Consistent with the RT-PCR results, GUS activities were detected only in floral tissues but were restricted to anthers. Anther staining was first visualized in c. 0.45 mm buds (at the time of pollen meiosis) and was intensified as the buds increased in size (Fig. 5a). GUS expression in anthers gradually declined as the flowers were maturing, and no anther staining was detected in fully open flowers. Isolated anthers from young buds revealed GUS activities in the developing microspores and the tapetal region (Fig. 5b,c). Histological investigations of stained anther sections demonstrated that GUS activities were strongly expressed in both tapetum and developing microspores. Staining was first observed at the tetrad stage (Fig. 6a) and became intensified during microspore release (Fig. 6b). GUS activities started to decline as development progressed, while staining could still be detected in the maturing pollen grains and the tapetal cells (Fig. 6c). Subsequently, no staining was found in mature anther and pollen grains (Fig. 6d).

image

Figure 5. DRL1 promoter-driven GUS expression patterns in transgenic Arabidopsis plants (Col-0 background). (a) Whole mount analysis of GUS expression in different-sized flower buds. Note the anther-specific staining in the developing closed buds. GUS activities were first detected in 0.45 mm closed buds (around the meiotic stage). Staining intensified and remained strong in 1 mm buds (early post-meiotic stage). Numbers represent the approximate bud length in mm. Bar, 1 mm. (b) A detached anther showing staining in the tapetal region (arrow). (c) An anther squash showing GUS activity detected in developing pollen grains. Bar: (b, c), 100 mm.

Download figure to PowerPoint

image

Figure 6. Anther sections of transgenic Arabidopsis showing DRL1 promoter-driven GUS activity. (a) Tetrad stage showing GUS activity in both the tapetum (*) and microspore tetrads (arrow). (b) Microspore release stage showing intense GUS staining in the tapetum (*) and the microspores. (c) Tapetum degeneration stage showing GUS activities in tapetal cells (*) and pollen. (d) Late-stage anther with the absence of GUS staining. All bars, 20 mm.

Download figure to PowerPoint

Complementation analysis

To conclusively define the role of DRL1 in male fertility, two complementation constructs in a plant binary vector were generated and transformed into the completely sterile drl1-2 mutant. In one construct, the DRL1 coding region was driven by the cauliflower mosaic virus 35S promoter. In another construct, the coding region was driven by the 2.0 kb upstream putative promoter (native promoter) sequence. The binary vectors were transformed individually to drl1-2/+ heterozygous plants using A. tumefaciens. Transformants were selected on hygromycin medium, and homozygous lines (drl1-2/drl1-2) were identified by PCR. Restoration of normal pollen development, silique elongation, and seed formation was achieved in all the transgenic homozygous lines carrying the native promoter construct (13 independent lines), but not the 35S promoter construct (37 independent lines) (Fig. 7). These results demonstrated conclusively that the DRL1 is indispensable for male fertility in Arabidopsis. Furthermore, temporal- and tissue-specific expression driven by the native promoter is essential for DRL1 to express its normal physiological functions.

image

Figure 7. Complementation of drl1-2 mutant plants. Restoration of viable pollen production (Alexander's staining) and normal silique production by ectopic expression of DRL1 cDNA driven by the 2.0 kb native promoter. In drl1-2 mutants, mature anthers are empty and siliques are undeveloped. Mutation phenotypes were not rescued when the 35S promoter was used to drive the expression. Bar, 100 mm.

Download figure to PowerPoint

Cellular localization and sequence analysis of DRL1 protein

To investigate the cellular localization of DRL1, a translational fusion of DRL1 upstream to EYFP driven by the 35S promoter was generated. The fusion construct was then introduced to onion epidermal cells by particle bombardment. As shown in Fig. 8, fluorescent signals were detected throughout the whole onion cells, including the nucleus. Plasmolysis treatments resulted in fluorescent signals contained in the shrinking cytoplasm. The same localization patterns were observed for an EYFP-only control construct. Accumulation of EYFP (29 kDa) in the nucleus is expected to occur by passive diffusion (Vom Arnim et al. 1998). However, the DRL1-EYFP fusion (65 kDa) is slightly larger than the size-exclusion limit (40–60 kDa) in the nuclear pore complexes (Görlich & Mattaj, 1996), suggesting that DRL1 may also be localized in the nucleus.

image

Figure 8. Localization of DRL1 protein. Onion epidermis cells were bombarded with the 35S-DRL1-EYFP fusion construct and the control 35S-EYFP construct. Microscopic observation indicated that DRL1 is mainly cytosolic and may also be localized in the nucleus. Sucrose (1 m) was used to induce plasmolysis. Bar, 100 mm.

Download figure to PowerPoint

DRL1 shows weak sequence identity to other DFR family proteins in Arabidopsis: 39.6% to TT3, 37.1% to BEN1, and 35.3% to BAN1. By contrast, a number of DRL1 homologs from different plants with sequence identity between 63.0 to 74.2% to DRL1 were identified in the DFCI Plant Gene Indices (http://compbio.dfci.harvard.edu/tgi/plant.html) by tblastn searches (data not shown). Interestingly, these sequences were exclusively derived from cDNA libraries of floral tissues, such as anthers, pollen, immature flowers, or inflorescence (DFCI Plant Gene Indices). Phylogenetic analysis showed that DRL1 clustered with the DRL1 homologs instead of the other Arabidopsis DFR-like proteins (Fig. 9a), suggesting a distinct functional role of DRL1 in plants. DRL1 and the dicot sequences form a separate clade within the group of conserved DRL1 homologs, while the cereal sequences were clustered in another clade. Alignment of the translated proteins of the different full-length cDNAs with DRL1 revealed the N-terminal putative NAD(P)H-binding domain and the high degree of homology throughout their entire sequences (Fig. 9b). The well-conserved N-terminal region is believed to be involved in the co-factor binding site of NAD(P)H-dependent metabolic enzymes (Lacombe et al., 1997).

image

Figure 9. Sequence analysis of Arabidopsis DFR-like proteins and DRL1 homologous proteins. (a) An unrooted phylogenetic tree of the DFR family proteins in Arabidopsis (TT3, BAN, BEN1, DRL1) and DRL1 homologs in different plant species. The tree was constructed by MEGA3.1 based on the neighbor-joining method (Kumar et al., 2004). Numbers at the nodes represent bootstrap values from 1000 replications. Scale bar represents 0.1 substitutions per site. (b) Sequence alignment of DRL1 and its closest homologs in other plants by the ClustalW method (http://www.ebi.ac.uk/clustalw). The putative NAD(P)H binding domain is boxed. Amino acid sequences of DRL1 homologous proteins were predicted from full-length contiguous cDNA sequences annotated at the DFCI Plant Gene Indices (wheat, TC236822; lettuce, TC8845; Aquilegia, TC14338; maize, TC341704). The OsDFR2 sequence was reported previously (Yau et al., 2005).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Our study revealed that a new DFR-like gene, DRL1, is essential for pollen development and male fertility in Arabidopsis. We demonstrated that DRL1 is an anther-specific gene with expression in tetrads, developing microspores, and tapetum. Previously, molecular analysis of anther development had led to the estimation of c. 3500 anther-specific transcripts in Arabidopsis (Sanders et al., 1999). Complete male sterility in the drl1-2 mutant suggested that there are no redundant genes that can replace the unique function of DRL1 in anthers. In addition, complementation of the mutant phenotypes was achieved by ectopic expression of the DRL1 cDNA driven by the native promoter, but not the 35S promoter, indicating that spatial and temporal specificities of transcription are critical for the proper expression of DRL1 physiological functions. The 35S promoter is commonly used for strong constitutive expression of transgenes in plants, including Arabidopsis. However, there have been reports demonstrating that it is not suitable for expression in certain tissue types, including the tapetum (Plegt & Bino, 1989; van der Meer et al., 1992). Tapetum is a layer of sporophytic cells derived from the secondary parietal cells and it plays important roles in the process of microspore development and maturation (Sanders et al., 1999). It secretes enzyme(s) to degrade callose synthesized in microsporocytes for release of microspores into the anther locules. The tapetum also provides nutrient and cell wall materials for the development of functional pollen. A number of genes encoding transcription factors and signal transduction components essential for tapetal functions have been identified in Arabidopsis, providing evidence for a cascade of regulatory network for different stages of tapetal formation and development.

The phenotypes of our drl1 mutants (degeneration of microspores after tetrad release and complete absence of pollen in mature anthers) are reminiscent of those observed in some male-sterile mutants defective in regulatory genes, such as ms1(Wilson et al., 2001) and ams (Sorensen et al., 2003). Thus, DRL1 may represent a primary target of some anther-specific transcription factors. Recently, a number of microarray experiments have been performed to investigate differential gene expression in male-sterile mutant flowers (Alves-Ferreira et al., 2007; Wijeratne et al., 2007; Xing & Zachgo, 2008). Examination of the publicly available data revealed that DRL1 (At4g35420) expression was down-regulated in flower buds of spl/nzz (Alves-Ferreira et al., 2007; Wijeratne et al., 2007), ems1 (Wijeratne et al., 2007), and roxy1roxy2 (Xing & Zachgo, 2008) mutants. However, the ‘down-regulation’ data observed could simply be explained by the fact that tapetum and microspores, in which DRL1 showed strong expression (Fig. 6), never developed in the spl/nzz and ems1 mutants (Schiefthaler et al., 1999; Yang et al., 1999). The tapetum of roxy1roxy2 mutants showed hypertrophic growth after degeneration of pollen mother cells in the abaxial anther lobes while sporogenous cell formation is impaired in the double mutants (Xing & Zachgo, 2008). SPL/NZZ, EMS1, and ROXY1/2 are considered early anther genes controlling sporogenous cell and tapetum formation and they are all expressed in the tapetum at the pollen mother cell stage (Schiefthaler et al., 1999; Yang et al., 1999; Xing & Zachgo, 2008). On the other hand, MS1 is a PHD-finger family transcription factor essential for postmeiotic tapetal development and pollen wall synthesis (Yang et al., 2007) and may also play a role in regulating programmed cell death in tapetum (Vizcay-Barrena & Wilson, 2006). Surprisingly, examination of two independent microarray datasets revealed that DRL1 expression was differentially up-regulated in the ms1 mutant flower buds (Alves-Ferreira et al., 2007; Yang et al., 2007). While the expression pattern of MS1 appeared to overlap with that of DRL1, the microarray data seemed to suggest that MS1 may instead function as a negative regulator of DRL1 expression. Other regulatory genes that were strongly expressed at similar developmental stages as in DRL1 included AMS encoding a MYC protein (Sorensen et al., 2003) and a few genes encoding R2R3 proteins, namely TDF1 (Zhu et al., in press), AtMYB32 (Preston et al., 2004), and AtMYB103 (Higginson et al., 2003). Further genetic and expression analyses in these mutants are necessary to identify the molecular component(s) that activate the anther-specific expression of DRL1. It is possible that some unknown positive regulators function together with MS1 (which might act as a repressor) to maintain an optimal expression level of DRL1 in anthers.

Our cellular localization experiment demonstrated that DRL1 encodes a soluble cytosolic protein, which may also be localized in the nucleus (Fig. 8). Previously, two Arabidopsis flavonoid enzymes (chalcone synthase and chalcone isomerase) were shown to be localized in nuclei of some cells types, in addition to their usual cytosolic location (Saslowsky et al., 2005). Analysis of the DRL1 protein sequence revealed the putative NADPH binding domain (Fig. 9b), which is conserved in the two flavonoid reductase enzymes DFR and BAN (Yuan et al., 2007). However, since the null mutant affecting chalcone synthase is fertile with functional pollen (Burbulis et al., 1996), the male fertility-associated DRL1 is unlikely to be involved in the biosynthesis of flavonoids. Consistent with this, the closely related rice homolog OsDFR2 was unable to complement the tt3 (DFR gene) mutations in Arabidopsis (C. H. Shih & C. Lo, unpublished ). Both DRL1 and OsDFR2 can be regarded as members of the mammalian 3β-hydroxysteroid/plant DFR superfamily, which also includes some bacterial and mammalian enzymes (Yau et al., 2005). The primary metabolic enzyme UDP-galactose-4-epimerase has been suggested to be a putative common ancestor for this family (Baker & Blasco, 1992). Hence, DRL1 may encode a reductase, an epimerase, or a dehydrogenase which functions in anthers to produce some novel metabolites essential for pollen development and male fertility.

The degeneration of drl1 mutant microspores was preceded by the rupture of their abnormal walls (Fig. 3). The formation of pollen wall requires the coordinated metabolic activities in both tapetum and microspores. Since the heterozygous drl1 mutants are indistinguishable from the wild type in terms of fertility and anther development, pollen abortion in the homozygous mutants is unlikely to be caused by defects in gametophytic gene expression. Instead, the tapetal expression of DRL1 at the microspore release stage may play an important role in wall formation in the developing pollen. Primexine, the footprint of the future exine of the pollen wall, is formed at the time of tetrad formation. During microspore release, sporopollenin, the main component of exine, is secreted by the tapetum and deposited on the microspores following the primexine pattern (Paxson-Sowders et al., 1997; Piffanelli et al., 1998). Sporopollenin is a biopolymer of lipid and phenylpropanoid (p-coumaric and ferulic acid) derivatives. So far only Male Sterility2 (MS2) and Faceless Pollen1 (FLP1) have been considered as candidate genes for sporopollenin biosynthesis (Ito et al., 2007). MS2, encoding a putative fatty acyl reductase, was demonstrated to be essential for exine formation (Aarts et al., 1997). Interestingly, MS2 expression was also up-regulated in most stages of the ms1 mutant flower buds (microarray data in Alves-Ferreira et al., 2007), suggesting that MS2 and DRL1 expression may be co-regulated. On the other hand, FLP1 encodes a putative lipid transfer protein that is involved in the formation of both exine and pollen coat (Ariizumi et al., 2003). The biochemistry of sporopollenin formation is still largely unknown, and whether DRL1 is involved in this process remains to be elucidated. The phenylpropanoid pathway enzymes cinnamoyl-CoA reductase and cinnamoyl alcohol dehydrogenase are also members of the mammalian 3β-hydroxysteroid/plant DFR superfamily (Lacombe et al., 1997). Hence, DRL1 may have a role in the synthesis of phenylpropanoid or aromatic precursors, which are released from tapetum to the anther locule for sporopollenin formation in the developing microspores.

Male sterility in plants is a useful trait for different agricultural applications, such as extension of flower shelf-life and production of hybrid seeds (Mitsuda et al., 2006). In addition, generation of genetically modified crops with controlled male sterility is highly desirable as a means for gene containment (Daniell, 2002). A major concern with regard to growing transgenic crops in the open field is the uncontrolled gene flow by cross-pollination of related wild species, leading to generation of ‘superweeds’ as well as disturbance of ecosystem. The occurrence of highly conserved anther- and floral-expressed DRL1 homologs in plant species of diverse families (Poaceae, Asteraceae, Rannuculaceae, Solanaceae) strongly suggested the presence of a common biochemical pathway (or metabolite) essential for pollen development. Hence, DRL1 may be exploited as a useful molecular target for manipulation of male sterility in flowering plants. We are currently investigating the differential metabolite profiles of the drl1 mutant flowers to gain an insight into the biochemical functions of this highly conserved gene which is indispensable for male fertility in Arabidopsis.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This work is supported by the Research Grant Council of the Hong Kong Special Administrative Region, China (grant no. HKU7337/04M), to CL and a NSERC Discovery grant to ECY. The authors thank the ABRC and INRA for providing the drl1-1 and drl1-2 T-DNA mutant seeds, respectively, and the RIKEN Bioresource Center for providing the DRL1 (At4g25420) full-length cDNA clone.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References