Tomato Male sterile 1035 is essential for pollen development and meiosis in anthers.

Male fertility in flowering plants depends on proper cellular differentiation in anthers. Meiosis and tapetum development are particularly important processes in pollen production. In this study, we showed that the tomato male sterile (ms10(35)) mutant of cultivated tomato (Solanum lycopersicum) exhibited dysfunctional meiosis and an abnormal tapetum during anther development, resulting in no pollen production. We demonstrated that Ms10(35) encodes a basic helix-loop-helix transcription factor that is specifically expressed in meiocyte and tapetal tissue from pre-meiotic to tetrad stages. Transgenic expression of the Ms10(35) gene from its native promoter complemented the male sterility of the ms10(35) mutant. In addition, RNA-sequencing-based transcriptome analysis revealed that Ms10(35) regulates 246 genes involved in anther development processes such as meiosis, tapetum development, cell-wall degradation, pollen wall formation, transport, and lipid metabolism. Our results indicate that Ms10(35) plays key roles in regulating both meiosis and programmed cell death of the tapetum during microsporogenesis.

Introduction

Pollen development is one of the most fundamental processes in the plant life cycle (Wilson and Zhang, 2009). Through pollen, plants deliver genetic material and expand genetic diversity by producing recombinant progeny in the subsequent generation (Deveshwar ). Pollen development involves an exquisite pathway supported by cellular changes and the regulation of an enormous number of genes (Honys and Twell, 2004; Wilson and Zhang, 2009; Feng ).

In Arabidopsis (Arabidopsis thaliana) and rice (Oryza sativa), anther development has been well studied and many of the genes involved have been identified. In Arabidopsis, after the floral structures are successively generated, anther cells initiate specification and differentiation to form a bilaterally symmetrical structure with four lobes. Archesporial cells in each lobe generate five distinct cell layers (from outer to inner: epidermis, endothecium, middle layer, tapetum, and the sporogenous cell) (Smyth ; Wilson and Zhang, 2009). At the meiotic stage, meiocytes (pollen mother cells), which are developed from sporogenous cells, undergo meiotic cell divisions and are separated from the tapetal cell layer. Tapetal cells become vacuolated and initiate programmed cell death (PCD)-triggered degradation. During meiosis I, homologous chromosomes replicate, pair, synapse, and undergo recombination by exchanging DNA (Zickler and Kleckner, 1999). After that, homologous chromosomes are aligned and pulled towards opposite poles by spindle organization. Finally, dyads are produced (Ma, 2005). For example, in prophase I, rice HOMOLOGOUS PAIRING ABERRATION IN RICE MEIOSIS 1 (PAIR1), PAIR2, and PAIR3 are important for chromosome pairing and synapsis, respectively (Nonomura , 2006; Yuan ). Arabidopsis SWITCH1 (SWI1) and rice MEIOTIC RECOMBINATION PROTEIN8 (REC8) are essential for chromatid cohesion and bivalent formation (Mercier ; Shao ). In metaphase I, Arabidopsis MULTIPOLAR SPINDLE1 (MPS1) plays a role in organization of the spindle and chromosomal segregation (Jiang ). In anaphase, rice POLLEN SEMI-STERILITY1 (PSS1), which encodes a kinesis-1 like protein, is necessary for meiotic chromosome pulling by the spindle (Zhou et al., 2001). Mutations of these genes related to meiosis cause defective meiocyte development and male sterility.

During meiosis, tapetal cells surrounding meiocytes produce various enzymes, lipids, starch, pollen wall materials, and other molecules required for pollen development (Goldberg ; Zhang ). For instance, Arabidopsis EXTRA SPOROGENOUS CELLS/EXCESS MICROSPOROCYTES1 (EMS1/EXS) and TAPETAL DETERMINANT1 (TPD1) are important for tapetal specification and maintenance of tapetal cell fate, and these mutants show extra meiocytes and no tapetal cells (Canales ; Yang ; Zhao ). Mutations in Arabidopsis DYSFUNCTIONAL TAPETUM1 (DYT1) and rice UNDEVELOPED TAPETUM1 (UDT1) cause abnormal tapetal development such as vacuolated tapetal cells, resulting in male sterility (Jung ; Zhang ). In addition, EMS1/EXS and TPD1 are also required for cytokinesis after chromosomal segregation (Canales ; Yang ; Zhao ), suggesting that sporophytic cells and gametophytic cells coordinate with each other by cell-to-cell communication.

Tomato flowers contain five sepals that alternate with five petals, in addition to five stamens and a style formed by two fused carpels. The stamens, which house pollen production, sit inside the petals. A single tomato stamen consists of two elongated compartments, and the individual stamens are fused together to form an anther cone called the androecium, which surrounds the style. In the cultivated tomato, the stigma is completely covered under the staminal tube (Brukhin ). Tomato pollen development is quite similar to that of Arabidopsis and rice (Brukhin , Wilson and Zhang, 2009; D. Zhang ). Tomato stamen primordia are initiated at the early stage of anther development followed by archesporial cell differentiation. Sporogenous and parietal cells are differentiated from archesporial cells. These cells give rise to microspore tetrads and tapetum, respectively, after going through meiosis. Finally, the microspores mature and become pollen grains (Rasmussen and Green, 1993; Brukhin ). In tomato, over 50 male-sterile mutants have been reported, and they can be divided into three classes (functional, structural, and sporogenous) based on their developmental defects (Gorman and McCormick, 1997). Sporogenous male-sterile mutants can be further classified into five groups (pre-meiotic, meiotic, tetrad, microspore, and not determined) according to the stage at which pollen development aborts or breaks down (Rick and Butler, 1956; Gorman and McCormick, 1997). For example, the functional male-sterile mutant positional sterile-2 (ps-2) is defective in pollen dehiscence. Pre-meiotic mutants such as male sterile (ms) 3 and ms15 display pollen mother cell (PMC) collapse or abortion prior to the meiotic prophase. Meiotic mutants such as ms5 and ms10 (allelic to ms10) show defects in tapetal tissue (Rick and Butler, 1956; Gorman and McCormick, 1997). While many tomato male-sterile mutants are available, the only known underlying gene was the polygalacturonase gene responsible for the ps-2 male-sterile mutant (Gorguet ).

The ms10 mutant was described previously as a spontaneous mutant with defects in tapetum development and degeneration (Rick, 1948; Zamir ; Corral-Martínez ). Because of its stable male sterility and lack of growth defects, it has been widely used for F1 hybrid breeding (Georgiev, 1991; Kumar and Singh, 2005). In addition, the ms10 mutant also has been used for anther culture to generate haploid plants, because a callus can easily be induced from its anthers (Zamir ; Corral-Martínez ). Here, we demonstrated that the ms10 mutant is defective in chromosome segregation at anaphase I during meiosis, as well as in tapetum development, causing male sterility. Using a map-based cloning approach, we found that Ms10 encodes a basic helix–loop–helix (bHLH) transcription factor. RNA sequencing (RNA-seq)-based transcriptome analysis revealed that Ms10 regulates 246 genes related to meiosis, tapetum development, lipid metabolism, cell wall modification/degradation, and pollen wall biosynthesis. These results demonstrated that Ms10 serves as a master regulator of pollen development in tomato.

Materials and methods

Plant material and plant growth

A tomato male-fertile parent (T-1082) and male-sterile ms10 (2–517), which was backcrossed to T-1082 six times, were obtained from the National Institute of Horticultural and Herbal Science (Suwon, Korea). T-1082 and the backcrossed ms10 were used in all experiments except the mapping experiment in which the original ms10 was crossed to T-1082. Seedlings were grown in 50-plug trays containing sterilized soil in a growth chamber maintained under 18h light (265 mE m–2 s–1) at 27 °C and 6h darkness at 18 °C and 60% humidity. At the eight-leaf stage, the seedlings were transplanted to a greenhouse in the farm of the College of Agriculture and Life Science at Seoul National University (Suwon, Korea).

Microscopy

Fluorescein diacetate (FDA) was used to check pollen viability according to the protocol of Kim . FDA-stained pollen was examined using an Axiophot microscope (Zeiss, Oberkochen, Germany). For ultrastructure and transmission electron microscopy (TEM) analysis, floral buds were infiltrated with Spurr’s resin according to the protocol of Kim . TEM images were observed using a JEM1010 transmission electron microscope (Jeol, Tokyo, Japan) at 80kV. For scanning electron microscopy (SEM), pollen grains of mature flowers were mounted and coated with palladium-gold in a sputter coater (BAL-TEC/SCD 005; Balzers, Lichtenstein) and examined using a field emission scanning electron microscope (SUPRA 55VP; Carl Zeiss, Germany) with an acceleration voltage of 15kV.

4′,6-Diamidino-2-phenylindole (DAPI) staining analysis of meiotic processes

For the observation of meiotic chromosomes, a modified PMC spreading protocol was used (Kwon and Kim, 2009). Briefly, floral buds around the meiotic stage were fixed in Carnoy’s fixative solution (ethanol:acetic acid=3:1, v/v) for 48h. Fixed buds were rinsed twice in distilled water and then once in 10mM citrate buffer (pH 4.5). Samples were incubated at 37 °C for 3h in a digestion mix containing 2% (w/v) cellulase RS, 1% (w/v) pectinase, and 0.5% (w/v) pectolyase Y23 (Sigma, St Louis, MO, USA) dissolved in 10mM citrate buffer. After digestion, cells were fixed in 60% acetic acid on a heated slide. After air drying, fixed cells were stained and mounted with 2 μg ml–1 of DAPI solution in Vectashield anti-fade mounting medium (Vector Laboratories, Burlingame, CA, USA).

DNA extraction

Genomic DNA was extracted from two to three young leaves using a hexadecyltrimethyl-ammonium bromide method (Jeong ). Leaf tissue was fragmented using TissueLyserII (Qiagen, Haan, Germany). DNA concentrations were measured with a Nanodrop spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) and diluted to a final concentration of 20ng μl–1 in TE buffer (pH 7.0) for further experiments.

Bacterial artificial chromosome (BAC) alignment and Ms10 -linked marker development

The Ms10 gene is known to be located between the PEROXIDASE-2 (PRX-2) and ANTHOCYANIN ABSENT (AA) genes, which are around the 69–78 cM region on chromosome 2 (Tanksley and Rick, 1980; Tanksley ). Tomato BAC clones corresponding to the 69–78 cM region were aligned and assembled by Seqman software (DNA Star; DNASTAR, Madison, WI, USA). In order to develop Ms10 -linked markers, primer sets were randomly designed within the 69–78 cM region from the assembled BAC clones. The designed primers were tested for polymorphism between parental DNA (ms10 and T-1082) and F1 DNA derived from a cross between ms10 and T-1082 plants using high-resolution-melting (HRM) analysis (Rotor-Gene 6000 thermocycler; Corbett Research, Sydney, Australia) according to a previously described method (Jeong ).

Genetic analysis and map-based cloning of Ms10

Fine mapping of Ms10 was performed with an F2 population derived from a cross between ms10 mutant and T-1082 plants, and was facilitated by the assembled BAC sequence described above and the tomato genome sequence (Tomato Genome Consortium, 2012). A population of 1100 F2 plants was scored for male sterility and subsequently genotyped with HRM markers (Supplementary Table S1 at JXB online). Linkage analysis of molecular markers was conducted using the Carthagene 1.0 program (de Givry ). Ms10 was positioned to an ~80kb region on chromosome 2 flanked by markers 762K and 843K. Putative genes in the 80kb region were predicted using the FGENESH program (http://linux1.softberry.com/), the tomato Unigene database from SGN (http://solgenomics.net), and the BLASTP interface of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/).

Total RNA isolation and reverse transcription (RT)-PCR

Floral buds at different stages, leaves, stems, and fruits were collected from ms10 and T-1082 plants and quickly frozen in liquid nitrogen. Total RNA was isolated using Trizol extraction buffer (Ambion, Carlsbad, CA, USA) according to the manufacturer’s protocol. cDNA was synthesized from 2 μg of total RNA using reverse transcriptase (Promega, Madison, WI, USA). cDNA (200ng) was used for RT-PCR. Amplified PCR products were separated on a 1% agarose gel and stained with ethidium bromide. The primer sequences used for RT-PCR are listed in Supplementary Table S1.

Rapid amplification of cDNA ends (RACE)

To identify the transcription start site of the Ms10 gene, 5ʹRACE-PCR was performed using a SMARTer™ RACE cDNA Amplification kit (Clontech Laboratories, Mountain View, CA, USA). RNA was extracted from T-1082 anthers and cDNA was synthesized according to the manufacturer’s instructions. Sequencing analysis was performed at the National Instrumentation Center for Environmental Management (NICEM, Seoul National University, Seoul, Korea).

Genome-walking PCR

To identify the mutated region of the ms10 gene, genome-walking PCR was performed using a Genome Walker kit (Clontech Laboratories, Mountain view, CA, USA) according to the manufacturer’s manual. Sequence information for gene-specific primers 1 and 2 is provided in Supplementary Table S1. The fragment amplified by genome-walking PCR was cloned into a pGEM-T vector (pGEM®-T Easy Vector Systems, Promega, Seoul, Korea) and sequenced at the NICEM.

Complementation of the ms10 mutation

A 2992bp genomic sequence of Ms10 , which contained the entire coding region (1002bp) with 1.4kb of upstream sequence and 0.6kb of downstream sequence, was amplified with XbaI restriction site-tagged primers (Supplementary Table S1). The resulting fragment was digested with XbaI, purified, and cloned into the XbaI site of the pCAMBIA2300 binary vector to generate pCAMBIA2300::Ms10 . The resulting construct and pCAMBIA2300 were introduced into Agrobacterium tumefaciens strain LBA4404 and used to transform ms10 heterozygous (Ms10 /ms10 ) cotyledon explants, as described previously (Seong ). The presence of the T-DNA insert in independent primary (T0) transformants was confirmed by PCR using a NPTII-specific primer set (Supplementary Table S1) to amplify a fragment of the NPTII gene. Subsequently, independent T1 plants with the homozygous ms10 /ms10 genotype were selected using ms10 -specific primers (Supplementary Table S1). From these, F1 plants having both NPTII and Ms10 transgenes were selected using NPTII-specific and Ms10 transgene-specific primers, respectively (Supplementary Table S1).

RNA transcriptome analysis by RNA-seq

Total RNA from floral buds at stages 1 to 3 was extracted in the same manner as described above. A strand-specific RNA-seq library was constructed for the synthesis of cDNA as described by Zhong . RNA transcriptome was obtained using Hiseq 2500 (Illumina/Solexa, San Diego, CA, USA) at NICEM. The RNA-seq algorithm of the CLC Genomics Workbench 6.0 was used for relative digital expression with a 98% identity threshold (CLC bio, Prismet, Denmark). Digital expression data were normalized and transformed using CLC Genomics Workbench 6.0 internal algorithms. ITAG2.3_CDS was used as a reference genome for read mapping (http://solgenomics.net). The mean values of three biological replicates were transformed into log2 values (ms10 /T-1082). The DESeq tool of the R package (http://www.bioconductor.org/) was used to identify differentially expressed genes in ms10 compared with T-1082 with a false discovery rate of <0.05 (Anders ).

Results

The ms10 mutant does not produce pollen

The tomato male-sterile mutant ms10 has been described as being defective in the production of active pollen (Rick, 1948; Zamir ). To examine developmental defects, ms10 plants were compared with male-fertile T-1082 plants. There was no difference in development between ms10 and T-1082 plants until the flowering stage. At the flowering stage, ms10 plants showed morphological differences in flower shape. The ms10 mutant had much smaller flowers compared with T-1082 plants (83% of T-1082, Fig. 1A, D, and Supplementary Fig. S1B at JXB online). In addition, the anther cone and the style in ms10 flowers were much shorter than in T-1082 flowers (60 and 86% of T-1082, Supplementary Fig. S1B). Due to the more dramatic change in anther cone length compared with style length, the styles of ms10 flowers protruded over the anther cones (Fig. 1D and Supplementary Fig. S1B). To check whether ms10 flowers produced viable pollen grains, an FDA assay, which measures cell viability, was performed with mature pollen grains or dust particles released from anthers at the dehiscence stage. The pollen of T-1082 flowers showed green fluorescence whereas no signals were detected from ms10 mutant flowers (Fig. 1B, E), indicating that ms10 flowers did not produce viable pollen. To confirm that ms10 flowers produced no pollen, we used SEM. T-1082 anthers contained normal globular pollen grains, but ms10 anthers had no pollen (Fig. 1C, F). These results demonstrated that the male sterility of the ms10 mutant resulted from a lack of production of pollen.

Fig. 1.

Flower phenotypes and pollen production in T-1082 (male fertile) and the ms10 mutant. (A, D) Flowers of T-1082 (A) and ms10 plants (D). (B, E) FDA assay for pollen viability of T-1082 (B) and ms10 (E) plants. Pollen grains or dust particles released from anthers at the dehiscence stage were used. (C, F) SEM observation of the surfaces of anthers at the dehiscence stage from T-1082 (C) and ms10 (F) plants. The insets show a closer look at the surface of the anthers. The T-1082 anther produced pollen grains (arrow and inset). Bars, 0.5cm (A, D); 100 µm (B, C, E, F); 10 μm (insets). (This figure is available in colour at JXB online.)

Pollen development in ms10 is arrested at the tetrad stage

To determine the spatial and temporal occurrence of defects in ms10 anthers, we prepared thin sections from anthers at different stages of development and examined them using light microscopy. At the pre-meiotic and meiotic stages of T-1082 and ms10 anthers, the five different cell layers were successfully differentiated from archesporial cells (Fig. 2A, D), and sporogenous cells developed into PMCs and underwent meiosis (Fig. 2B, E). At the tetrad stage, however, dramatic morphological differences were observed between ms10 and T-1082 anthers. In T-1082 anthers, PMCs divided into tetrads after meiosis. Tapetal cells were greatly condensed and deeply stained. The middle cell layer was degenerated and almost invisible (Fig. 2C). In ms10 anthers, PMCs were crushed and failed to produce tetrads. Tapetal cells and the middle cell layer were excessively enlarged and vacuolated (Fig. 2F). At the microspore stage in T-1082 anthers, free microspores were released into anther locules (Fig. 2G). In ms10 anthers at the microspore stage, degenerated meiocytes were aggregated and gradually degraded. Vacuolated tapetal cells and middle cell layers were severely expanded (Fig. 2J). At the mitotic and dehiscence stages in T-1082 anthers, vacuolated microspores were deeply stained with toluidine blue due to the accumulation of collapsed tapetum fragments and nutrients, and the tapetum and middle cell layer had already disappeared (Fig. 2H). Finally, anthers dehisced and pollen grains were released (Fig. 2I). By contrast, in ms10 anthers at the mitotic and dehiscence stages, degenerated meiocytes continued to dwindle, while tapetal cells remained swollen and vacuolated without degeneration (Fig. 2K, L).

Fig. 2.

Histological characterization of anther development in T-1082 and ms10 flowers. Cross sections of T-1082 (A–C, G–I) and ms10 (D–F, J–L) anthers at pre-meiotic (A, D), meiotic (B, E), tetrad (C, F), microspore (G, J), mitotic (H, K), and dehiscence (I, L) stages. dMs, degenerated meiocytes; dT, degenerated tapetum; En, endothecium; Ep, epidermis; ML, middle cell layer; Msp, microspore; PMC, pollen mother cell; SC, sporogenous cell; T, tapetum; Tds, tetrads. Bars, 50 μm. (This figure is available in colour at JXB online.)

The ms10 mutant is defective in tapetum development

To examine in detail the defects of tapetal cells in the ms10 mutant, we used TEM. In agreement with the light microscopic observations, there was no noticeable difference between ms10 and T-1082 at the pre-meiotic stage (Fig. 3A, E). At the meiotic stage, tapetal cells in T-1082 anthers were still well defined, but more vacuoles were generated and the nuclear membrane began disappearing, indicating that PCD-triggered cell degradation had already commenced (Fig. 3B). By contrast, ms10 tapetal cells at the meiotic stage showed abnormal morphology: the cytoplasm of ms10 tapetal cells was highly vacuolated and showed extensive lipid deposits (Fig. 3F). At the tetrad stage in T-1082, the cytoplasm was condensed and deeply stained. The nuclear membrane and cellular organelles had disappeared. Electron-dense deposits were observed in the vacuoles (Fig. 3C). At the tetrad stage of ms10 , tapetal cells were greatly expanded and vacuolated. Cellular organelles including the nucleus still maintained their structures (Fig. 3G). At the microspore stage of T-1082 tapetal cells, the cytoplasm and cell walls were diminished. Nuclei and cellular organelles were also completely absent. Instead, orbicules were distributed along the loosened tapetum cells facing the microspore (Fig. 3D). In ms10 tapetal cells at the microspore stage, tapetal cells were still enlarged and vacuolated (Fig. 3H). These observations reveal that the ms10 mutant had abnormally vacuolated tapetal cells with aborted degeneration.

Fig. 3.

TEM analysis of T-1082 and ms10 tapetum. Tapetal cells of T-1082 (A–D) and ms10 (E–H) anthers at pre-meiotic (A, E), meiotic (B, F), tetrad (C, G), and microspore (D, H) stages. At the meiotic stage, T-1082 anthers (B) contained nuclei, plastids, mitochondria, and vacuoles. Arrowheads indicate electron-dense deposits in the vacuoles. Arrows indicate plasmodesmata. In ms10 anthers, the nuclei appeared normal, but larger vacuoles were generated. Mitochondria (indicated by a circle) and lipid structures were also detected (F). At the tetrad stage, T-1082 anthers (C) showed an irregular tapetum cell wall, and a degenerated nucleus and cellular organelles. Arrowheads indicate electron-dense deposits in the vacuoles. ms10 anthers (G) had extremely vacuolated tapetum and intact nuclei and organelles. At the microspore stage, T-1082 anthers (D) exhibited degenerated tapetal cells and production of orbicules (arrowheads). ms10 anthers (H) were completely vacuolated. dMs, degenerated meiocytes; EN, endothecium; Li, lipid deposit; M, mitochondria; ML, middle cell layer; Msp, microspore; N, nucleolus; Nu, nucleus; P, plastid; T, tapetum; TCW, tapetal cell wall; V, vacuole. Bars, 2 μm.

Development of PMCs in the ms10 mutant is arrested at anaphase I during meiosis

PMCs in ms10 anthers were degenerated and failed to produce tetrads (Fig. 2). To investigate defects of meiosis in ms10 , we observed meiocytes using TEM. In T-1082 anthers, uni-nucleate PMCs (Fig. 4A) gave rise to dividing cells (Fig. 4B) and dyads (Fig. 4C), producing tetrads (Fig. 4D) during meiosis. In ms10 anthers, PMCs with a well-defined structure were observed as in T-1082 (Fig. 4E). Before long, however, the nuclei of PMCs were crushed without degeneration (Fig. 4F, arrow) and diminished steadily. Eventually only traces remained (Fig. 4G, arrow) and even these disappeared at the end of development (Fig. 4H). No dividing nucleus, dyads, or tetrads were observed in ms10 anthers. We further performed chromosome spread experiments using DAPI staining. In T-1082 anthers, chromosomes in PMCs underwent homologous chromosome pairing and synapsis at leptotene/zygotene (Fig. 4I, J) and pachytene (Fig. 4K, L), respectively. Chromosomes were condensed as bivalents at diakinesis and aligned at metaphase I (Fig. 4M, N). The aligned chromosomes then separated and moved towards opposite poles at anaphase I (Fig. 4O), forming dyads at telophase I (Fig. 4P). Tetrads were formed at the end of meiosis as a result of two nuclear segregations (Fig. 4Q). In ms10 anthers, meiosis occurred normally until metaphase I (Fig. 4R–W). Chromosomes were abnormally separated at anaphase I (Fig. 4X), failed to form dyads during telophase I (Fig. 4Y), and tetrads were never observed. These results indicated that ms10 has a defect in chromosome segregation at anaphase I, resulting in no tetrad formation during meiosis.

Fig. 4.

TEM analysis and DAPI staining for observation of meiosis in T-1082 and ms10 . (A–H) TEM micrographs of meiocyte cells from T-1082 (A–D) and ms10 (E–H) anthers at meiosis initiation (A, E), anaphase I (B, F), telophase I (C, G), and tetrad (D, H) stages. At anaphase I, the meiocyte of T-1082 anthers was divided (arrow and white circles, B), while the meiocyte of ms10 anthers failed to divide and was diminished (arrow, F). At telophase I (dyad stage), the meiocyte of T-1082 anthers was divided into a dyad (C), whereas the meiocyte of ms10 anthers was further diminished (arrow and circle, G). At the tetrad stage, the dyad of T-1082 anthers was divided into tetrads (D), whereas only remnants of the meiocyte remained in ms10 anthers (H). Nu, nucleus; Ha, haploid. Bars, 2 μm. (I–Y) Meiotic chromosomes stained with DAPI in T-1082 (I–Q) and ms10 (R–Y) at the leptotene (I, R), zygotene (J, S), pachytene (K, L, T, U), diakinesis (M, V), metaphase I (N, W), anaphase I (O, X), telophase I (P, Y), and tetrad (Q) stages. At anaphase I, T-1082 chromosomes were separated to generate two polar sets (O), while ms10 chromosomes were aggregated and not segregated clearly (X). At the telophase I stage, T-1082 had generated two different chromosome sets (P), while ms10 chromosomes were not segregated clearly and failed to form dyads (Y). At the tetrad stage, T-1082 had generated four tetrahedrally arranged chromosome sets, whereas no tetrad was produced in the ms10 mutant. Bars. 10 μm. (This figure is available in colour at JXB online.)

MS10 encodes a bHLH transcription factor

To examine the inheritance of the ms10 gene, F1 plants were developed by crossing ms10 as a female parent and T-1082 as a male parent (Supplementary Table S2 at JXB online). All F1 plants developed normal flowers with viable pollen grains. In an F2 population generated from F1 plants, male-fertile and male-sterile phenotypes segregated in a ratio of approximately 3:1, indicating that the mutation is controlled by a single recessive gene. In previous studies, it was determined that the ms10 locus is linked between the PER-2 and AA genes on chromosome 2 (Tanksley and Rick, 1980; Tanksley ). Based on this, BAC sequence information from around PER-2 and AA genes was collected, assembled, and used for marker development. Fine mapping of the locus allowed us to position Ms10 to within an ~80kb region flanked by markers 762K and 843K (Fig. 5A). A total of 13 hypothetical genes were predicted in this region (Fig. 5B). To explore their transcriptional expression in anthers, RT-PCR was performed. Among them, Solyc02g079810 was downregulated in ms10 anthers compared with T-1082 anthers (Fig. 5C). A sequence similarity search further predicted that Solyc02g079810 is similar to Arabidopsis DYT1. Given that Solyc02g079810 was expressed only in anthers, we considered Solyc02g079810 as a strong candidate gene for Ms10 . Solyc02g079810 contains a 627bp coding sequence comprising four exons and three introns with a 154bp 5′-untranslated region and a 419-bp 3′-untranslated region (Fig. 5D). The transcription start site was located 151bp upstream of the ATG site as confirmed by 5′RACE-PCR analysis. To reveal the structure of the mutation in ms10 , genome walking was performed. Compared with a T-1082-derived genomic clone containing the Solyc02g079810 gene, an ms10 -derived clone had an insertion of a retrotransposable DNA fragment (398bp) in the promoter region near the transcription start site (Fig. 5D and Supplementary Fig. S2 at JXB online). These results indicated that very weak expression of the Ms10 transcript in ms10 anthers resulted from failure of transcription initiation.

Fig. 5.

Map-based cloning of the Ms10 gene. (A) Fine genetic mapping of the Ms10 gene delimited the target gene within an interval between markers 762K and 843K on chromosome 2. Numbers in parentheses indicate the number of recombination events identified between the markers and the target gene. (B) Physical map of the region containing the Ms10 gene. Pentagons indicate predicted genes and their directions. The putative Ms10 gene (Solyc02g079810) is indicated in red. Solyc02g079760, bHLH transcription factor; Solyc02g079770, DAG protein; Solyc02g079780, glycine-rich protein; Solyc02g079790, DUF538; Solyc02g079800, unknown protein; Solyc02g079810, putative Ms10 gene; Solyc02g079820, pentatricopeptide repeat-containing protein; Solyc02g079830, pentatricopeptide repeat-containing protein; Solyc02g079840, terpene synthase; Solyc02g079850, pleckstrin homology; Solyc02g079860, NPH3; Solyc02g079880, translation initiation factor; Solyc02g079890, terpene synthase. (C) RT-PCR for Solyc02g079810 from ms10 and T-1082 plants. S1–S5, different stages of anther development (S1, meiosis and tetrad stage; S2, young and vacuolated microspore stage; S3, mitosis and maturation stage; S4, dehiscence stage; S5, opened flower stage); St, stem; Lf, leaf; Fr, fruit. Actin was used as a loading control. (D) The structure of the Ms10 gene encoding a bHLH transcription factor. Black boxes indicate exons. The ATG start site is denoted as position 1 and the other numbers indicate nucleotide distance from the ATG site. A transposon insertion flanked by direct repeat nucleotide sequences (AGAAGA/TCTTCT) was found between the transcription start site (TSS) and the ATG in ms10 . (E) Flower morphology and pollen viability in ms10 transgenic plants complemented with the wild-type Ms10 gene. Three independent T1 transgenic plants with the Ms10 gene showed normal flowers and had viable pollen. Bars, 0.5cm (flower images); 200 μm (pollen viability images). (This figure is available in colour at JXB online.)

To confirm that Solyc02g079810 was the Ms10 gene, we used Agrobacterium-mediated transformation to introduce the wild-type Solyc02g079810 gene, expressed from its native promoter, into plants having the heterozygous genotype (Ms10 /ms10 ) because the homozygous ms10 mutant cannot produce seeds. We generated a total of six T0 plants and performed systemic analysis to select T1 transgenic plants with the homozygous ms10 genotype. We first selected a total of 46 F1 plants with the ms10 /ms10 background using ms10 -specific primers (Supplementary Fig. S3 at JXB online) and further screened these plants with NPTII-specific and transgene-specific primers, respectively (Supplementary Fig. S3 and Supplementary Table S3 at JXB online). Among the T1 plants containing the Ms10 transgene in the ms10 /ms10 background, eight plants (generated from four individual T0 plants) had normal flowers with viable pollen grains (Fig. 5E and Supplementary Table S3) and produced normal fruits with seeds. These results demonstrate that the male sterility of ms10 resulted from the loss of function of the Solyc02g079810 gene.

Ms10 encodes a putative transcription factor with a bHLH domain (Fig. 6A). A BLAST search with Ms1035 protein sequence showed that the Ms1035 protein had the highest similarity to a bHLH protein from Solanum tuberosum (91%) and had 47 and 37% similarity to AtDYT1 and OsUDT1, respectively, both of which are required for tapetum development (Jung ; Zhang ). Amino acid sequence alignment showed that a bHLH domain is highly conserved among these proteins (Fig. 6A). To gain insights into the phylogenetic relationship between Ms1035 and other bHLH homologues related to male sterility, phylogenetic analysis was performed. The results showed that Ms1035, StbHLH, AtDYT1, and OsUDT1 were within the same clade (Fig. 6B).

Fig. 6.

Comparison of Ms1035 and its homologues. (A) Amino acid sequence alignment of Ms1035 and homologous proteins from potato (StbHLH), Arabidopsis (AtDYT1), and rice (OsUDT1). Alignments based on the conserved region of the Ms1035 protein were generated using ClustalW of MegAlign (DNASTAR). The conserved bHLH domain is indicated by the outlined box. Black shading indicates identical residues. (B) Phylogenetic tree of Ms1035 and other bHLH proteins from several plant species. The phylogenetic tree of deduced amino acid sequences was generated with the neighbour-joining method using MEGA5 software (http://www.megasoftware.net). Bootstrap values (%) are from 1000 replicates, indicated above the nodes. Ms1035 is underlined. At, Arabidopsis thaliana; Bn, Brassica napus; Ca, Capsicum annuum; Mt, Medicago truncatula; Md, Malus domestica; Os, Oryza sativa; Ps, Pisum sativum; Sl, Solanum lycopersicum; St, Solanum tuberosum; Ta, Triticum aestivum. GenBank accession numbers are given in parentheses. (This figure is available in colour at JXB online.)

Comparison of global gene expression between ms10 and T-1082 anthers by RNA-seq analysis

To investigate genes regulated by Ms10 during pollen development, comparative transcriptome profiling between ms10 and T-1082 was performed by RNA-seq. For the identification of distinct genes at three different stages (meiosis/tetrad, young/vacuolated, and mitosis/maturation stages), we analysed all expressed genes using the scatterplot of the DESeq package (Supplementary Fig. S4 at JXB online) and selected differentially regulated genes using a false discovery cut-off (5%) and >log2 fold difference (ms10 /T-1082). Finally, 246 genes including 220 genes that were downregulated (Supplementary Table S4 at JXB online) and 26 that were upregulated (Supplementary Table S5 at JXB online) in ms10 relative to T-1082 anthers (P<0.05) were discovered by statistical analysis. To investigate further putative functions of downregulated genes in ms10 , we utilized agriGO (http://bioinfo.cau.edu.cn/agriGO/), the National Center for Biotechnology Information database (http://www.ncbi.nlm.nih.gov), and the genomic database of SGN (http://solgenomics.net). The 220 genes downregulated in ms10 were classified into 14 different categories according to molecular function and biological process including transcription factor, cell modification/degeneration, transporter, pollen wall or coat protein, lipid metabolism related, and meiosis related (Fig. 7A and Supplementary Tables S6–S12 at JXB online). Furthermore, the 220 genes downregulated in ms10 were compared with 435 and 958 downregulated genes in Arabidopsis dyt1 (Feng ) and rice udt1 (Jung ). A total of 65 genes were commonly downregulated across the three organisms. Additional 41 and 15 downregulated genes in ms10 were also downregulated in Arabidopsis dyt1 and rice udt1, respectively (Supplementary Tables S4–S12). This result implies that common regulatory machinery is used for pollen development in Arabidopsis, rice, and tomato. Some representative genes involved in pollen development and substantially suppressed in ms10 are presented in Fig. 7B. To validate the RNA-seq analysis results, we tested the expression patterns of genes known to be involved in pollen development using RT-PCR. The RT-PCR results were consistent with those obtained from RNA-seq data (Fig. 8 and Supplementary Fig. S5). For example, meiosis-related genes such as Solyc03g116930.2.1 (sister chromatid cohesion), tapetum-specific genes including Solyc07g053460.2.1 (cysteine protease), and transcription factors such as Solyc08g062780.1.1 [ABORTED MICROSPORES-like (AMS-like)] were strongly expressed in T-1082 but considerably downregulated in ms10 . In addition, bHLH89/91-like (Solyc01g081100.1.1), AtTDF1-like (Solyc03g113530.2.1), AtMYB103-like (Solyc03g059200.1.1), aspartic proteinase (Solyc06g069220.1.1), endo-1,3-β-glucanase (Solyc03g046200.1.1), lipid transfer protein (Solyc06g059790.2.1 and Solyc01g095780.2.1), and arabinogalactan protein (Solyc11g072780.1.1) were also downregulated in ms10 . Transcript levels of two randomly selected genes, Solyc07g055920.2.1 (Tomato agamous-like 1) and Solyc09g074440.2.1 (Defenseless1), which were not differentially regulated in RNA-seq analysis, and those of the control Actin gene were similar between ms10 and T-1082 (Supplementary Fig. S5).

Fig. 7.

Genes downregulated in the ms10 mutant. (A) Classification of genes downregulated in the ms10 mutant by gene ontology. Each category of downregulated genes was classified according to putative molecular function and biological process. The number in each category indicates the number of downregulated genes in the ms10 mutant and the percentage indicates the number of genes in that category relative to the 220 annotated downregulated genes. (B) Differential expression patterns of representative genes involved in pollen development between T-1082 and ms10 anthers. Heat maps show log2-scaled reads per kilobase per million reads (RPKM). The intensities of the colours (from 0 to 1: light to dark shading) increase with increasing expression differences as indicated at the upper-right. S1, S2, and S3 indicate different anther developmental stages (S1, meiosis and tetrad stage; S2, young and vacuolated microspore stage; S3, mitosis and maturation stage). Three biological replicates of anthers at each stage were prepared for RNA-seq. (This figure is available in colour at JXB online.)

Fig. 8.

Expression of pollen development marker genes in floral buds of ms10 and T-1082 plants. RT-PCR of AMS-like (A), MS1-like (B), bHLH89/91-like (C), AtTDF-like (D), AtMYB103-like (E), LeGRP92 (F), aspartic protease (G), cysteine protease (H), sister chromatid cohesion (I), endo-1,3-β-glucanase (J), lipid transfer protein (K, L), TomA108 (M), TA29 (N), arabinogalactan protein (O), and pLAT52 (P). The y-axis indicates gene expression relative to tomato Actin. The results are averages of three independent biological experiments. Error bars show the standard error value of three replicates. RNA samples were collected from floral buds at stages 1–5 of anther development. AMS-like, Solyc08g062780.1.1; MS1-like, Solyc04g008420.1.1; bHLH89/91-like, Solyc01g081100.1.1; AtTDF1-like, Solyc03g113530.2.1; AtMYB103-like, Solyc03g059200.1.1; LeGRP92, Solyc02g032910.1.1; aspartic proteinase, Solyc06g069220.1.1; cysteine protease, Solyc07g053460.2.1; sister chromatid cohesion, Solyc03g116930.2.1; endo-1,3-β-glucanase, Solyc03g046200.1.1; lipid transfer protein, Solyc06g059790.2.1 (K) and Solyc01g095780.2.1 (L); TomA108, Solyc01g009590.2.1; TA29, Solyc02g078370.1.1; arabinogalactan protein, Solyc11g072780.1.1; pLAT52, Solyc10g007270.2.1.

Ms10 regulates transcription factors involved in anther development

Transcriptional regulation is important for controlling the expression patterns of genes to produce normal pollen (Wilson and Zhang, 2009). Among the 220 downregulated genes in the ms10 mutant were 24 transcription factors including bHLH, MYB, NAC, and zinc-finger types (Fig. 7 and Supplementary Table S6). For example, Solyc03g113530.2.1 and Solyc08g062780.1.1 are similar to Arabidopsis DEFECTIVE IN TAPETAL DEVELOPMENT AND FUNCTION1 (AtTDF1) and AtAMS, which are important for tapetum development and degeneration (Sorensen ; Zhu ). Solyc04g008420.1.1 and Solyc01g081100.1.1 are similar to Arabidopsis MALE STERILITY1 (MS1) and bHLH91, respectively, which are involved in tapetum degeneration by regulating PCD-triggered cell death (Vizcay-Barrena and Wilson, 2006; Xu ).

Ms10 positively regulates genes for lipid metabolism, cell wall modification/degradation, pollen wall/coat proteins, and transporters

The gene ontology annotation analysis showed that Ms10 regulated 20 genes for lipid metabolism, 11 related to energy metabolism, 48 for cell wall modification/degeneration, 14 for pollen wall/coat proteins, and 23 for transporters (Fig. 7 and Supplementary Tables S7–S11). This regulation could be direct or indirect. Among the genes exhibiting altered expression in the ms10 mutant, a lipid-related gene, Solyc03g051960.2.1 (fatty acid CoA reductase) showed high similarity to Arabidopsis MS2, which is essential for pollen wall formation (Chen ). Solyc06g072780.1.1 and Solyc02g068400.2.1 had high similarity to rice MICROSPORE AND TAPETUM REGULATOR1 (MTR1) and Arabidopsis QUARTET3 (QRT3), respectively. OsMTR1 encodes an arabinogalactan protein known to regulate male sporophytic and reproductive development (Tan ). AtQRT3 encodes a pectin lyase involved in dissolving the PMC callose wall and microspore separation (Preuss ). Cell wall modification/degeneration genes such as proteolytic enzyme families induce PCD in tapetum cells (Niu ). Solyc06g069220.1.1 had high similarity to rice aspartic proteinase 65, which is involved in tapetum degradation (Huang ).

Ms10 regulates genes involved in meiosis

We demonstrated that the ms10 mutant was defective in chromosome separation during meiosis (Fig. 4). To identify meiosis-related genes regulated by Ms10 , we compared downregulated genes in ms10 with known Arabidopsis meiocyte-specific genes (Yang ). We found a total of nine genes related to meiosis (Fig. 7 and Supplementary Table S12). For example, Solyc03g116930.2.1 was similar to a protein involved in sister chromatid cohesion (Jin ). Solyc08g077860.2.1 encoded a subtilisin-related meiotic serine protease, which is expressed during meiosis and late microsporogenesis in tomato (Riggs ). Together, these results imply that Ms10 is a master regulator controlling several genes involved in anther development.

Discussion

Ms10 is important for controlling meiosis and tapetum development

Proper meiosis and development of sporophytic cell layers including the tapetum are essential for successful pollen development in plants (Ma, 2005; Yuan ; D. Zhang ). In the ms10 mutant, duplicated chromosomes were not separated to form dyads (Fig. 4X). This failure of dyad formation at anaphase I resulted in the degradation of meiocytes in ms10 . Similar defects were found in Arabidopsis dyt1 and rice udt1 mutants. In the dyt1 mutant, meiocytes were not able to complete cytokinesis, resulting in failure of tetrad formation (Zhang ; Feng ). Transcript analysis showed that expression of the meiosis-specific gene ROCK-N-ROLLER/AtMER3 (RCK/AtMER3), which is implicated in sister chromatid cohesion, was significantly reduced in dyt1 (Chen ; Zhang ), suggesting that DYT1 regulates the expression of prophase I-related genes. In the rice udt1 mutant, meiocytes did not produce tetrads due to incomplete meiosis, and PAIR1 involved in sister chromatid cohesion was downregulated (Jung ; Wang ). Although these two homologues were not differentially expressed in the RNA-seq analysis, we found that several other meiosis-related genes were downregulated in ms10 anthers (Figs 7 and 8). For example, Solyc03g116930.2.1 is homologous to yeast PRECOCIOUS DISSOCIATION OF SISTERS PROTEIN5 (PDS5), which is important for sister chromatid cohesion such as chromosome condensation, pairing, and synapsis in prophase I (Jin ). Solyc02g032910.1.1 is predicted to encode a protein highly similar to glycine-rich protein, which is important for sporopollenin deposition on meiocyte and exine formation (McNeil and Smith, 2010). Another downregulated gene, Solyc06g076020.2.1, is similar to Heat shock protein 70 (Hsp70), which has an important role supporting cyclin-dependent kinase activity in meiosis I in animals (Eddy, 1999). In plants, Lily messages induced by meiosis 18 (LIM18), showing high similarity to eukaryotic HSP70, is specifically expressed in microsporocytes during meiosis I (Minami ). Our results suggested that the incomplete meiosis of ms10 may be due to downregulation of these genes.

Another significant defect in the ms10 mutant was abnormal tapetum development. Degeneration of the tapetum and middle cell layer was delayed and, consequently, tapetal cells were greatly expanded and vacuolated (Fig. 3). These phenotypic defects are also commonly observed in Arabidopsis dyt1 and rice udt1 mutants (Jung ; Zhang ). Other mutants impaired in tapetum development such as Arabidopsis ams and rice tapetum degeneration retardation 1 (tdr1) also exhibit male sterility (Sorensen ; Li ). These results support the idea that meiosis and tapetum development are important for pollen development, and that Ms10 and its homologues have a conserved role in the completion of meiosis and tapetum development.

RNA in situ hybridization analysis revealed that Ms10 is expressed exclusively in the meiocyte and tapetal tissues at the early stage of anther development (Supplementary Fig. S6). These transcript results are consistent with the phenotypic defects in meiocytes and tapetal cells of the ms10 mutant. To find putative anther-specific cis-acting regulatory elements in the Ms10 promoter region, we searched the PLACE database (Higo ). The Ms10 promoter contained several putative transcription binding sites and regulatory sequences. For example, pollen-specific cis-elements such as Agamous binding site, Pollen1LeLAT52, and GTGANTG10 were found near the transcription start site (Supplementary Fig. S2), suggesting that these putative cis-elements regulate the expression of Ms10 in anthers.

Ms10 encodes a bHLH transcription factor

The Ms1035 protein sequence had high similarity to StbHLH, AtDYT1, and OsUDT1. Amino acid sequence alignments showed that they contain a conserved bHLH domain at the N-terminal region. The PSORT program (http://mobyle.pasteur.fr/cgi-bin/portal.py?#forms::psort) predicted these proteins to be targeted to the nucleus. Indeed, OsUDT1 is localized to the nucleus and contains a signal peptide at the N terminus (Jung ). However, Ms1035, StbHLH, and AtDYT do not contain conventional nuclear signal peptides, and whether they are also localized to the nucleus remains to be elucidated. The phylogenetic analysis showed that Ms1035, StbHLH, AtDYT1, and OsUDT1 belonged to the same clade. Considering the conserved function of Ms10 , AtDYT1, and OsUDT1, it seems possible that StbHLH may be involved in meiosis and tapetum development in potato. A clade containing OsTDR1, AtAMS, BnbHLH, and Solyc08g062780.1.1 was very close to the clade containing Ms1035. Interestingly, OsTDR1 and AtAMS are regulated by OsUDT1 and AtDYT1, respectively, and are also involved in tapetum degeneration at the post-meiotic stage (Sorensen ; Li ), suggesting that Solyc08g062780.1.1 (AtAMS-like), which was downregulated in ms10 , is probably involved in tapetum degeneration. In addition, OsTIP2 and OsEAT1 in rice and AtbHLH89/91 and AtbHLH10 in Arabidopsis are key regulator genes of tapetal PCD (Xu ; Niu ; Fu ). Tomato Solyc01g081100.1.1 (EAT-like) belongs to the clade including these genes, suggesting that Solyc01g081100.1.1 could be involved in tapetal PCD. By contrast, genes in other clades were not related to anther development. For example, AtMYC2 and AtMYC4 are related to abscisic acid signalling (Abe ) and AtICE is involved in cold stress (Chinnusamy ). These results suggest that the bHLH subfamilies containing Ms10 and AtAMS have a conserved function and were evolutionarily separated from other bHLH subfamilies.

Roles of Ms10 -regulated genes during anther development

A pathway regulated by the DYT1–TDF1–AMS–bHLH89/91–MYB80 transcriptional cascade is suggested to underlie Arabidopsis pollen development (Fig. 9). The genes in this pathway are involved in early tapetum function (DYT1; Zhang ), callose dissolution (TDF1; Zhu ), and PCD-triggered cell death (AMS; Sorensen ; Xu ). DYT1 regulates tapetum differentiation during the development of microspore mother cells before meiosis (Zhang ). During meiosis, TDF1 is highly expressed in tapetum cells and meiocytes. TDF1 functions downstream of DYT1 and upstream of AMS (Zhu ). AMS is strongly expressed in tapetum cells specifically after meiosis. The ams mutant did not successfully undergo PCD, resulting in abnormal tapetal degeneration and retardation (Sorensen ; Xu ). Rice TIP2 is associated with anther cell wall specification, tapetal cell size, and PCD by regulating rice TDR1 and EAT1 as well as interacting with them (Fu ). EAT1 plays an important role in PCD by regulating proteases and interacting with TIP2 and TDR1 (Niu ; Fu ). TDR1 is involved in tapetal cell size and PCD by interacting with TIP2 and EAT1 (Li ; Fu ), indicating that TDR1–TIP2–EAT1 is formed as a consecutive regulation chain for rice anther development (Fig. 9; Fu ). Arabidopsis bHLH89/91 are homologues of rice TIP2 and EAT1, and interact with AMS, the homologue of rice TDR1 (Xu ), suggesting their involvement in tapetal PCD. However, the function of bHLH89/91 is not clear due to a lack of genetic and molecular evidence. Arabidopsis MYB80 (formerly MYB103) encoding a MYB transcription factor is also expressed in the tapetum and microspores (Zhang ). The UNDEAD aspartic protease is a direct target of MYB80, and the interaction between MYB80 and UNDEAD serves to control induction of tapetal PCD in Arabidopsis (Phan ). The homologues of these genes have been found in rice (Fig. 9) and have similar functions (Li ; Wilson and Zhang, 2009; D. Zhang ; X. Zhang ; Niu ; Fu ). In the ms10 mutant, Solyc03g113530.2.1 (AtTDF-like), Solyc08g062780.1.1 (AtAMS-like and OsTDR-like), and Solyc01g081100.1.1 (OsEAT1-like and AtbHLH89/91-like) were significantly downregulated according to the RNA-seq transcriptome and RT-PCR (Fig. 8 and Supplementary Tables S4–S12). These results suggest that this transcriptional cascade for pollen development is well conserved in Arabidopsis, rice, and tomato. In addition, the early stage of meiocyte development was controlled by Ms10 . For example, Solyc03g116930.2.1, homologous to yeast sister chromatid cohesion (PDS5), was downregulated in the ms10 mutant, which showed abnormal chromosome separation and failure of dyad formation. Similarly, Arabidopsis RCK and rice PAIR1, which are involved in sister chromatid cohesion, were downregulated in Arabidopsis dyt1 and rice udt1 mutants, respectively (Jung ; Zhang ), implying that MS10 and its homologues AtDYT1 and OsUDT1 regulate meiosis. Based on these evolutionary relationships and the conserved functions of these proteins, we have presented a model for tomato pollen development (Fig. 9).

Fig. 9.

Comparative models for Ms1035 and its homologues in Arabidopsis and rice that regulate tapetal PCD and meiosis. Arrows represent positive regulation confirmed by RT-PCR or transcriptome analysis. Rhombuses depict protein–protein interaction or protein–promoter interaction confirmed by yeast two-hybrid, quantitative chromatin precipitation-PCR, or electrophoretic mobility shift assay. Dashed lines represent predicted regulation or interaction. The Arabidopsis and rice models are based on data reported by Jung , Li , Zhang , Wilson and Zhang (2009), Xu , Phan , Niu , and Fu . SCC, sister chromatid cohesion (synaptonemal complexes); bHLH, basic helix–loop–helix transcription factor; MYB, MYB transcription factor; Cys protease, cysteine protease; Asp protease, aspartic proteinase. (This figure is available in colour at JXB online.)

In addition to the conserved genes, many genes presumably involved in pollen development in Arabidopsis and rice were also downregulated in ms10 . Cell wall modification and degradation genes such as Solyc07g064190.1.1 (pectin methylesterase-like), Solyc06g069220.1.1 (cysteine protease-like), and Solyc07g053460.2.1 (C1A cysteine proteinase-like) were highly downregulated. Another role of Ms10 could be supporting pollen wall deposition by delivering materials and nutrients to developing microspores. For example, orbicules are known to deliver nutrients to sporopollenin of microspores. In T-1082, orbicules were successfully generated along the degenerated tapetum (Fig. 3D). However, in ms10 , orbicules were not found and no normal sporopollenin structures were observed. In ms10 , sporopollenin biosynthesis-related genes such as Solyc12g010920.1.1 (long chain fatty acid reductases) and Solyc04g081780.2.1 (lipase) were downregulated.

ms10 showed a protruded stigma due to significantly reduced anther cone size. Arabidopsis dyt1 also exhibits a protruded stigma phenotype (Zhang ). Tomato Stigma exsertion (Se2.1) encoding a bHLH transcription factor is the major quantitative trait locus for the development of stamen length (Chen ). Interestingly, expression of Se2.1 (Solyc02g087860.2.1) was substantially reduced in ms10 anthers, suggesting that Ms10 may affect stamen length by regulating Se2.1. As we have shown here, many Ms10 -regulated genes in our RNA-seq data were directly or indirectly involved in anther development. Therefore, our RNA-seq data should provide a good basis for the identification and analysis of new genes involved in anther development in tomato.

Supplementary data

Supplementary data are available at JXB online.

Supplementary Fig. S1. Differences of organ length in ms10 and T-1082 flowers.

Supplementary Fig. S2. Nucleotide sequence of the Ms10 promoter.

Supplementary Fig. S3. Complementation of ms10 transgenic plants with the wild-type Ms10 gene.

Supplementary Fig. S4. Scatterplot identification of differentially expressed genes between T-1082 and ms10 anthers.

Supplementary Fig. S5. Expression patterns of genes regulated by Ms10 .

Supplementary Fig. S6. Localization of Ms10 expression in T-1082 anthers.

Supplementary Table S1. Primers used in this study.

Supplementary Table S2. Genetic analysis of the ms10 gene using an F2 population derived from 2–517 (ms10 ) and T-1082 (male fertile) plants.

Supplementary Table S3. Summary of ms10 transgenic plants complemented with the wild-type Ms10 gene.

Supplementary Table S4. 220 genes downregulated in the ms10 mutant.

Supplementary Table S5. 26 genes upregulated in the ms10 mutant.

Supplementary Table S6. Transcription factors downregulated in the ms10 mutant.

Supplementary Table S7. Lipid metabolism genes downregulated in the ms10 mutant.

Supplementary Table S8. Pollen wall or coat protein genes downregulated in the ms10 mutant.

Supplementary Table S9. Cell wall modification/degradation genes downregulated in the ms10 mutant.

Supplementary Table S10. Transporter genes downregulated in the ms10 mutant.

Supplementary Table S11. Energy metabolism-related genes downregulated in the ms10 mutant.

Supplementary Table S12. Meiosis-related genes downregulated in the ms10 mutant.