Genetic control of male fertility in Arabidopsis thaliana: structural analysis of premeiotic developmental mutants

Sex Plant Reprod (1994) 7:17-28

9 Springer-Verlag 1994

A. M. Chaudhury 9 M. Lavithis. P. E. Taylor S. Craig 9 M. B. Singh 9 E. R. Signer 9 R. B. Knox E. S. Dennis

Genetic control of male fertility in Arabidopsis thaliana: structural analysis of premeiotic developmental mutants

Received: 3 August 1993 / Revision accepted: 27 September 1993

Abstract We have taken a mutational approach to identify genes important for male fertility in Arabidopsis thaliana and have isolated a number of nuclear male/ sterile mutants in which vegetative growth and female fertility are not altered. Here we describe detailed developmental analyses of four mutants, each of which defines a complementation group and has a distinct developmental end point. All four mutants represent premeiotic developmental lesions. In ms3, tapetum and middle layer hypertrophy result in the degeneration of microsporocytes. In ms4, microspore dyads persist for most of anther development as a result of impaired meiotic division. In ms5, degeneration occurs in all anther cells at an early stage of development. In ms 15, both the tapetum and microsporocytes degenerate early in anther development. Each of these mutants had shorter filaments and a greater number of inflorescences than congenic male-fertile plants. The differences in the developmental phenotypes of these mutants, together with the non-allelic nature of the mutations indicate that four different genes important for pollen development, have been identified.

Key words Male-sterile mutants 9 Arabidopsis Pollen development 9 Microsporogenesis

A. M. Chaudhury 9 S. Craig 9E. S. Dennis CSIRO Division of Plant Industry, GPO Box 1600, Canberra, ACT 2601, Australia M. Lavithis 9P. E. Taylor 9M. B. Singh 9R. B. Knox ([~) School of Botany, University of Melbourne, Parkville, Victoria 3052, Australia E.R. Singer Massachusetts Institute of Technology, Cambridge, MA 02138, USA

Introduction Male reproductive processes in angiosperms include the specification of the male organs, the filaments and anthers, and development of the male gametophytes, the haploid pollen grains. Both of these processes are critical for male fertility. While a number of genes have been isolated from various plants based on their relative abundance in the male organs (Koltunow et al. 1990; Mascarenhas 1990; Theerakulpisut et al. 1991), only mutants that are impaired in male fertility or are malesterile (ms) mutants conclusively define genes that are essential for male fertility. Normal microspore development takes place inside the anther lobes which consist of inner archesporial cells and a four-layered anther wall (for a review of angiosperm gametophytic development see: Knox 1984; Kaul 1988). During normal microsporogenesis, archesporial cells give rise to microsporocytes surrounded by nutritive tapetal cells. Microsporocytes enter meiosis to form tetrads of haploid microspores. These microspores, within a callose special wall, are released by callose dissolution. The microspore then completes exine synthesis, enlarges, makes the inner cellulosic intine and undergoes mitosis to form a tube nucleus and a generative cell that divides to form the two sperm cells. Finally, the endothecium walls thicken and the anther dehisces at the stomium, releasing viable pollen grains. Male-sterile mutants defective in male gametophyte development have also been identified in a number of other plants. In tomato, there are 27 mapped male sterile mutants (Rick 1948) that are altered in different stages of microspore development and that have high penetrance of the phenotype and high female fertility. In Arabidopsis thaliana, only two sporophytic mutants have been reported that are defective in pollen development. They include ms1 (Van der Veen and Wirtz 1968; Chaudhury et al. 1992) that has been characterized and used extensively for genetic mapping studies. Another mutant, apt, has been isolated, based on the resistance of the plant to 2,6-diaminopurine, which lacks the enzyme

18

adenine phosphoribosyl transferase (Moffat and Sommerville 1989). This mutant was further characterized and found to have a postmeiotic defect in microspore development (Regan and Moffat 1990). Because of the complexity of the developmental steps leading to functional pollen, and the evidence that a large number of genes are required for fertility in a number of other species of higher plants (Kaul 1988), it is likely that a large number of genes is involved in male development in A. thaliana. We set out to identify more male-sterile mutants in A. thaliana in order to define further the developmental pathway leading to male fertility. The advantages of Arabidopsis as a model angiosperm have been described (Meyerowitz and Pruitt 1985). Because of the small genome size and the development of molecular genetic methodologies in A. thaliana, the isolation of male-sterile mutants may facilitate the cloning and molecular characterization of ms genes, thus allowing a molecular definition of the pathway of male development. In this paper, we describe the isolation and characterization of four male-sterile mutants impaired in premeiotic development.

post-fixed in either 1% or 2% OS204in buffer for 2 h, rinsed and dehydrated through a graded series of ethanol before embedding in LR white resin (London Resin Company, UK). To ensure adequate infiltration, anthers were dissected from flowers and left in resin at l0 ~ C with agitation for up to 3 weeks prior to polymerization. Sections, 0.5-2 gm thick, were cut and stained with toluidine blue (0.5% in 1% borax). Alternately, tissues were rinsed and dehydrated through a graded series of acetone before embedding in Spurr's (1969) resin. Sections, 2 gm thick, were stained with 0.5% toluidine blue. Sections, 100 nm thick, were stained with uranyl acetate and lead citrate prior to observation with either a Siemens Elmiskop 102 TEM at 60.kV or a JEOL 1200EX TEM at 80 kV. For fluorescence microscopy, flower buds were embedded in Historesin (Reichert-Jung, Germany) (Block and Debrouwer 1992). Briefly, the unfixed material was pretreated with spermidine, dehydrated to 90% acetone and embedded in a watermiscible glycol methacrylate resin at 5~ C. Sections, 6-10 gm thick, were stained with either the DNA specific stain, DAPI (Sigma; 20 mg/ml in 0.2 M McIlvaine's buffer), Calcofluor White, specific for newly synthesized cell walls, or Auramine O which is specific for sporopollenin, cuticles, and lipids (Heslop-Harrison 1979). Preparations were viewed by incident fluorescence microscopy (Olympus system) using UV or violet exciter filters. GA 3 treatment Gibberellin was applied as GA 3 onto the tip of 4-week-old A. thaliana plants as described (Nester and Zeevaart 1988). Silique phenotype was screened 5 days after GA 3 was applied.

Materials and methods Plant lines, growth conditions, mutant isolation, and genetic methods Male sterile mutants of A. thaliana (L.) Heynh. were isolated in the ecotype 'Columbia'. Plant were grown at 22~ C under continuous fluorescent illumination supplemented with incandescent light (150 g E m s1 PAR). For mutant isolation, seeds were mutagenized with 0.3% (v/v) ethylmethane sulfonate for 16 h at room temperature and were grown in soil mix that has been described (Last and Fink 1988). M2 seeds were produced by self-fertilization of the M1 plants. For the isolation of the mutants, 200 000 M2 seeds were planted and were allowed to flower and form siliques. Self-sterile plants were initially identified by the continued inability of the siliques to elongate. These sterile plants were crossed with pollen from isogenic male-fertile plants. If the siliques elongated due to outcrossing, the plant was a presumptive male-sterile mutant. F1 seeds from these crosses were harvested and F1 plants were screened for self-fertility. F2 seeds were harvested from the F1 plants, and were grown up to study the segregation of the malesterility phenotype.

Results Mutant isolation

In wild-type A. thaliana, flowers predominantly self-fertilize. The ovary harboring the fertilized embryos elongates, giving rise to the seed pod or silique. However, if fertility is impaired due to a mutation, the ovaries do not elongate. If the mutation causing sterility is due to a defect specific to male development, the short pistils can be pollinated by outcrossing with pollen from another self-fertile plant, leading to elongation. Because the inflorescence, possessing flowers at different stages of development, has a flowering period of at least 3 weeks, such outcrossings can be performed on different flowers of the inflorescence to verify that the sterility is a stable trait and was not caused by environmental changes. When M2 plants produced an inflorescence, at least five flowers were allowed to form before screening. This Complementation analyses of the male-sterile mutants was necessary because the early formed flowers are ofCrosses were performed between the Ms~ms heterozygote con- ten sterile in a mutagenized population. Forty plants structed from each of the male-sterile mutants with each of the ms~ms homozygote male-sterile plants. At least 50 F1 plants re- were identified that continued to make short siliques. In 19 of these, siliques could elongate with normal seed-set sulting from these crosses were scored for male sterility. when they were crossed with pollen from an isogenic wild-type plant. These 19 plants were designated putaLight and electron microscopy tive male-sterile plants. The other 21, presumed to be For transmission electron microscopy (TEM) and bright-field mi- either female-sterile or both male- and female-sterile, croscopy, whole flower buds were vacuum-fixed in either 3% were not rescued or studied further. Of the 19 putative paraformaldehyde and 2% glutaraldehyde, 2% paraformalde- male-sterile plants, six were tested for developmentalhyde and 1% glutaraldehyde, 3% glutaraldehyde, or 2% glu- stage pollen arrest and of these, four which shared pretaraldehyde for 4-24 h at room temperature or 4~ C. The fixative included 0.03 M or 0.05 M PIPES buffer or 0.025 M Na phosphate meiotic defects are described here and the others will be buffer (pH 7). Following rinsing in the same buffer, tissues were described in a subsequent paper.

19 Table 1 Frequency of male-sterile plants in the F2 progeny of crosporocytes are enveloped by four rows of anther wall selfed MS~ms lines cells: tapetum, middle layer, endothecium, and epidermis (Fig. 1A), as observed in 0.5-mm buds. Following Genotype MaleMaleX2 P meiosis, the tetrads of microspores are held within the sterile fertile callose wall (Fig. 1B) which is later degraded to liberate ms3/MS3 59 193 0.33 0.5 free microspores (Fig. 1C). Tapetal cells are active up to ms4/MS4 35 99 0.16 0.75 this point but degenerate soon afterwards, at the vacuoms5/MS5 66 189 0.16 0.75 late period of microspore development (Fig. 1D). ms15/MS15 71 209 0.17 0.75 All four ms mutants, when observed by light microscopy, showed microsporocyte arrest prior to or during meiosis (Fig. 1F-Q). Accordingly, an ultrastructural Table 2 Femalefertility in male-sterile mutants study was undertaken to indicate more precise timing of Mutant Seeds/pod Seeds/pod developmental arrest in these four premeiotic mutants. selfed (mean__SD) outcrossed (mean_+SD) Ultrastructural analysis Wild type 51 _+0.64 53 _+0.21 of microspore development ms3 0 54-t-0.11 ms4 0 56__0.18 Wild type ms5 0 56_+0.87 msl5 0 57_+0.22 Figure 2A-C shows cross-sections of an anther at early prophase of meiosis. The microsporocyte has a large nucleus, containing a prominent peripheral nucleolus Genetic analyses and paired chromosomes, and the nuclear envelope is enveloped by two layers of rough endoplasmic reticuEach of the four putative male-sterile mutants, designatlure. The cytoplasm is rich in proplastids, mitochondria, ed ms3, ms4, ms5 and ms15, was rescued by pollination and electron-lucent vesicles. The tapetal cells are large, with pollen from isogenic self-fertile plants. F1 plants binucleate and rich in mitochondria, with some proplasfrom each of the ms mutants were grown and tested for tids and vacuoles with prominent nucleoli. The nuclei male sterility based on both their ability to form pollen are enveloped by several layers of rough endoplasmic and silique length. Each of the mutations was recessive reticulum. The D N A of microsporocyte nuclei fluoto wild-type, as deduced from the fertility of F1 plants. resces strongly after DAPI staining and Calcofluor F2 progeny of these F1 plants was scored for segregaWhite staining resulting in bright fluorescence of the tion of ms mutations (Table 1). Each of the mutations outer tangential wall of the tapetum and moderate fluobehaved as a single mendelian locus. rescence of the cell walls of endothecium and epidermis (Fig. 2D, E). Following Auramine O staining, tile cuticle around the epidermal cells fluoresces strongl~y, along Genetic complementation with lipids in the microsporocytes (Fig. 2F). between the ms mutations Pollen from MS/ms heterozygous plants of the four male-sterile mutants (Table 2) was used to pollinate the same, or a different ms/ms male-sterile plant to determine whether the four mutants belong to the same or different complementation groups. If the male-sterile plants contain mutations in the same complementation group, it would be expected that 50% of the progeny from the cross would be male-sterile. If, on the other hand, the male-sterile plants belong to different complementation groups, all the progeny would be fertile. In the case of the four putative mutants, all the progeny proved to be fertile, so that each mutant is a member of a different complementation group. Light microscopic analysis of microsporocyte and microspore development Light microscopic analyses of the various stages of wildtype pollen development are shown in Fig. 1. Mi-

ms3

Developmental studies show that microsporocytes in 0.5 mm buds have become irregular in shape and their cytoplasm is pyknotic and electron-opaque in the youngest anthers studied. Microsporocyte wall profiles are irregular, and organelle or plasma membranes cannot be detected, showing that the microsporocytes have undergone autolysis (Fig. 3). Cells of the tapetum and middle layer have become highly vacuolate and markedly hypertrophied. Absence of fluorescence after DAPI staining indicates that the D N A of their nuclei has been degraded (Fig. 3C). Nuclei of most anther wall cells fluoresced as in wild type, suggesting that the cells of the epidermis and endothecium are not degraded (Fig. 3C). Following Calcofluor White staining, the cellulosic walls in the locule showed intense white fluorescence (Fig. 3D). Auramine O fluorescence occurred in both the anther cuticle and the degenerating microsporocytes (Fig. 3E). In mature anthers at anthesis,

Fig. I A - Q Light micrographs (LM) from bright-field microscopy. A - E Pollen development in male-fertile anthers. A Early microsporocyte stage. Microsporocytes are weakly stained and enveloped by the anther wall. B Tetrad stage. Tetrads (7) are surrounded by a tapetal layer (t). The epidermal (ep), endothecial (en), and middle layers are differentiated. C Microspores (m), after their release from tetrads. Tapetal cells have not yet degraded. D Microsporocytes are irregular in shape and possess vacuoles. E Tricellular pollen grains (p) prior to release from the anther. Note that both anther locules have ruptured at the stomium (st). F - I Pollen development in the male-sterile mutant ms5. F Anther at early microsporocyte stage appears similar to male-fertile (A). G At the next stage of development, mierosporocytes persist and are weakly stained when compared with the tapetal cells. H The tapetal cells enlarge and microsporocytes appear to degrade. I

Remnants of microsporocytes appear in the locule, surrounded by vacuolate cells of the anther wall. Systemic cell death appears to occur in all the cells of the anther. J - M Pollen development in the male-sterile mutant ms4. J Anther at the early microsporocyte stage appears similar to male-fertile (A). K Microsporocyte undergoing meiosis. L Dyads of microspores are heavily stained. Small vacuoles appear in tapetal cells. M Dyads of microspores persist and are surrounded by vacuolate tapetal cells. N - Q Pollen development in the male-sterile mutant ms15. N Anther at the early microsporocyte stage. O Microsporocytes have granular cytoplasm and nuclei with irregular shape. P Microsporocytes persist and are weakly stained. Tapetum appears heavily stained. Q Both microsporocytes and tapetum degrade and a thickened wall layer delineates the locule. The contents of anther wall cells also degenerate. Bar 20 gm

Fig. 2 A - F Micrographs from sectioned anthers of wild-type. A TEM of an anther from a bud (0.5 mm length) showing the vacuolated cells of the epidermis (ep), endothecium (en), middle layer (rnl), and tapetum (t). Each epidermal cell possesses chloroplasts, a single nucleus, and a granular cytosol. Starch granules are seen in the chloroplasts of endothecial cells. Cells of the middle layer partly interdigitize. The tapetal cytoplasm is more electronopaque than other layers of the anther wall. Each tapetal cell is binucleate and possesses rough endoplasmic reticulum, plastids, mitochondria, and vacuoles. Bar 1 gin. B The microsporocyte (m) possesses a single heterochromatic nucleus with a nucleolus at the periphery, endoplasmic reticulum, plastids, mitochondria, and

vesicles. Each microsporocyte is enveloped by an electron-lucent region. Bar 1 gm. C LM showing the four pollen sacs of an anther containing microsporocytes. Bar 40 pro. D DAPI staining showing sharp and bright nuclei of microsporocytes and cells of the anther wall Bar 20 ram. E Calcofluor White stains anther cell walls and in particular, the brightly fluorescent outer tangential wall layers of the tapetum at early microspore stage. Microspores are also diffusely stained. Bar 40 gm. F Auramine O staining of a section, from the same stage as in Fig. 2 D, shows that the cuticle (cu) and the cytoplasm of all cells in the anther except t]he epidermis are stained. Bar 40 pm

Fig. 3A-E Micrographs from sectioned anthers of ms3. A TEM micrograph of an anther from a young bud (0.5 mm length) showing irregularly shaped microsporocytes (m) bounded by cell walls and enveloped by lightly stained walls (w). The microsporocyte cytoplasm appears pyknotic and electron-opaque. The tapetum (t) and the middle layer (ml) are highly vacuolated and hypertrophied. Bar 2 txm. B LM of anther at the same stage as in Fig. 3 A. The epidermis (ep) and the endothecium (en) are vacuolate and possess large nuclei. The middle layer (ml) and the tapetum appear

devoid of cytoplasmic contents and the degenerated microsporocytes are densely stained. Bar 20 gm. C Neither the microsporocytes nor the tapetum shows any fluorescence after DAPI staining (arrow). The epidermal and endothecial nuclei fluoresce brightly. Bar 40 gm. D Calcofluor White shows highly fluorescing cell wall material of degenerating microsporocytes within the locule. Bar 40 g. E Auramine O shows intense fluorescence of both the cuticle (cu) and the degenerating microsporocytes. Bar 20 gm

23

Fig. 4A-D Micrographs from sections of ms4. A TEM of dyads from anthers of a bud (0.5 mm length) showing incomplete wall formation (arrows). Each dyad appears to be surrounded by an electron-lucent region. Bar 2 gin. B LM of a bud (0.3 mm length) showing microsporocytes (m) apparently enveloped by a callosic wall. The tapetum (t) possesses vacuoles and appears to be degenerating. Bar 10 gm. C TEM of an anther of a bud (2 mm length) showing the thickened outer tangential wall of the tapetum. The tapetum is now highly vacuolate. Bar 0.5 gin. D Within the anther, only the nuclei of the epidermis and endothecium fluoresce with DAPI. Cell walls show non-specific fluorescence. Bar 20 Ixm the h y p e r t r o p h i e d tapetum and middle layer have degenerated and collapsed. The anther then consists of an intact epidermal and endothecial layer with cellular debris a r o u n d the periphery of each empty locule.

ms4

In this mutant, microsporocytes are apparently' formed normally and are observed in 0.5 m m buds (Fig. 1J) and enter meiosis, but development is arrested by the end of meiosis I. at the dyad stage (Fig. 4A). Cellular structures are visible in later stages of anther development, yet their nuclei cannot be detected with the fluorochrome D A P I (Fig. 4D), An unusual feature is that dyads persist in the locules until late in the development of the anther. Absence of Calcofluor White fluorescence shows there is no cellulosic wall surrounding the locular contents (data not shown). An unusual feature of the locule is the presence of an inner locular wall at the site of the degenerated tapetum, at the periphery of the locule. Spherical

Fig. 5

25

Fig. 5A-E Micrograph from sections of rosS. A TEM of an anther of a bud (0.5 mm length) with a tapetum (1) possessing many small vesicles. The microsporocyte nucleus is present but the cytoplasm has many vacuoles and appears to be degenerating. The epidermis (ep), endothecium (en), and middle layer (rn/) are highly vacuolate but their plasma membranes appear to be intact. Bar 9/am. B The microsporocyte nuclei (m) show diffuse fluorescence with DAPI, the tapetal nuclei show no fluorescence, but the nuclei of the other anther walt layers are brightly fluorescent. Bar 25/am. C Calcofluor White shows only diffuse fluorescence in the microsporocytes. The walls of the epidermis and endothecium are strongly fluorescent. Bar 20/am. D Auramine O staining shows fluorescence of the cuticle (cu) and the cytoplasm of both the tapetum (t) and the microsporocytes (m). Bar 20 /am. E At a later stage of development, both the microsporocytes and the outer tangental wall (s) of the tapetum autofluoresce. Bar 100/am

Fig. 6A-E Micrographs from sections of msl5. A LM of a bud (0.4 mm length) showing the tapetum (t) possessing many small vacuoles and microsporocytes (m) with poorly stained amorphous cytoplasm. Bar 15 gm. B TEM of a later stage reveals a thickened outer tangential wall (s) of the tapetum bordering the collapsed middle layer and endothecium (en). Globular electron-opaque material (v) in the locule is all that remains of both tapetum and microsporocytes. Bar 2 gm. C DAPI staining of a bud (0.5 mm length) shows weak diffuse fluorescence of the tapetal (arrows) and microsporocyte nuclei (arrowheads). Nuclei of the other anther wall cells fluoresce brightly. Bar 30 gin. D Calcofluor White shows predominant fluorescence of the anther walls. The microsporocyte walls are weakly fluorescent. Bar 20 /am. E Auramine O shows fluorescence of the cuticle (cu), the walls of the anther wall layers, and, to a lesser extent, the material surrounding the microsporocytes. Bar 20 gm

26 Table 3 Vegetative,floral, and prefloral growth in malesterile mutants

Mutant

Rosette leaves per plant

Bracts per plant

Flowers per inflorescence

Length of inflorescence (ram)

Wild type

8+_0.13 8 + 0.03 9 +_0.04 9 -I-0.07 8_+0.13

4.2-t-0.83 4.1 _+0.62 3.9 _+0.76 4.2 + 0.58 4.3_+0.61

38_+1.76 95 _+0.42 96 + 0.77 97 ___0.32 99__+0.41

380_+2.87 520 + 1.20 517 _+0.53 567 + 0.97 568__+1.30

ms3 ms4 ms5 ms15

electron-opaque bodies form this wall (Fig. 4C), with deposition also occurring between the microsporocyte cells. After the formation of this wall, the tapetum and dyads degenerate, although the epidermis and endothecium remain intact. Unusually, the middle layer also remains intact at this stage, in contrast to wild type (unpublished data). At anthesis, the anthers possess an intact epidermis and endothecium, a degenerated middle layer, and a locule that is empty and bordered by the thickened wall. We conclude that the microsporocytes are persistent, perhaps due to the presence of the remarkably thickened noncellulosic cell walls.

ms5

In this mutant, as in ms4, the microsporocyte cells initially appear normal. Ultrastructural analysis shows that in 0.5-mm buds, the anther tissues, including the anther wall cells, appear degraded (Fig. 1F-I; Fig. 5). Initially, the microsporocyte nuclei are present, but the cytoplasm possesses many vacuoles and appears to be degenerating (Fig. 5A). The epidermis, endothecium, and middle layer are all highly vacuolated. The microsporocytes show weak, diffuse fluorescence with DAPI (Fig. 5B) and Calcofluor White (Fig. 5C). Auramine O staining shows fluorescence of both the cuticle and cytoplasm of the tapetum and microsporocytes (Fig. 5D). At a later stage of development, both the degenerating microsporocytes and the inner tangential wall of the tapetum give strong autofluorescence (Fig. 5E). By anthesis, the profiles of both the epidermal and endothecial cells remain polyhedral, but the locule is empty. ms15

In this mutant, microsporocytes are formed normally (Fig. IN). Early in microsporocyte development at the 0.5-mm bud stage, the tapetum has numerous vacuoles (Fig. 6). In anthers from 0.5-mm buds, DAPI staining shows weak, diffuse fluorescence of the tapetal and microsporocyte nuclei, but nuclei of the other anther wall cells fluoresce normally (Fig. 6C). Calcofluor White staining shows weak fluorescence of the microsporocytes, but normal fluorescence of cell walls of the anther wall (Fig. 6D). Auramine O staining shows fluorescence

of the cuticle, anther wall layers and, to a lesser extent, material surrounding the microsporocytes. The tapeturn and microsporocytes degenerate early in anther development. The middle layer also degenerates, and the locule is bounded by a markedly thickened inner locular wall resembling that observed in ms4 (Fig. 4). In older anthers, intact epidermal and endothecial cells surround an empty locule, with very few remnants of the tapetal and microsporocyte cells. Sporophytic growth of the ms mutants Wild-type and ms mutant plants were examined for their female fertility and any differences in the growth of the rosette or the inflorescence (Table 3). All the mutants examined have undiminished female fertility as determined by the number of seeds following out-crossing. The gametophytic mutations resulting in male sterility did not change the number of rosette leaves. However, in all four of the male-sterile mutants, a significant increase was observed in the number of flowers per inflorescence (Table 3). Phenotypic rescue of short siliques of the male-sterile plants with exogenous gibberellins The gibberellin growth hormone, GA3, was added in an attempt to alter the phenotypes associated with male sterility. In tomato, the phenotype of the mutant stamenless has been found to be reversible following the addition of exogenous gibberellins (Sawhney and Greyson 1973). Both auxins and gibberellin had a pronounced effect on the development of siliques in the four male-sterile mutants tested: ms3, ms4, ms5 and ms15. The ms3 siliques became as long as pollinated siliques after the addition of gibberellins [unpollinated siliques = 0.3 cm; pollinated siliques = 1.2+0.12 cm (mean _+SD); gibberellin-treated siliques = 1.1 +_0.03, n = 10]. This resulted in the formation of parthenocarpic fruits. However, the addition of gibberellins did not result in the formation of seeds in any of the male-sterile mutants. The mutants ms4, ms5, and ms15 gave results similar to ms3.

27

Discussion We have identified four male-sterile mutants of A. thaliana, each defining a single recessive locus and some likely to define novel genes. Only two previous reports have been published of A. thaliana mutants that are impaired in microsporogenesis: msl (Van der Veen and Wirtz 1968) and apt which is resistant to 2,6-diaminopurine (Moffat and Sommerville 1989). The mutants described here are non-allelic to rnsl and are not resistant to 2,6-diaminopurine (A. Chaudhury, unpublished data). Given the complexity of the male development process and the isolation of at least 25 male-sterility genes in tomato (Rick 1948), it is likely that we have not identified all the Arabidopsis genes required for specifying male development. In each male-sterile mutant studied, female fertility was unimpaired, showing that these are genuine male-sterile mutants. The mutants ms3, ms4, ms5, and ms15 all show early developmental arrest. Since no normal tetrads are seen in their anthers, these genes are either active at the premeiotic or meiotic stages of development. This means that the defects are in diploid genes of the sporophyte, in the sporogenous cells or microsporocytes or in the tapeturn. Also since these mutants have no effect on sporophytic tissues, all these mutations specify anther-specific genes. The earliest sign of a defect in development occurs in ms3 anthers. At the earliest stage of development, the sporogenous cells appeared degenerate. However, the tapetum and middle layers were markedly hypertrophied and may be responsible for the arrest in microsporocyte development. It is possible that the functions required for microsporocyte development are entirely normal in ms3 mutants, and that the ms3 defect is solely in the integrity of the tapetum and other anther wall cells. In the mutant ms15, the initial arrest occurred in microsporocytes prior to meiosis. The tapetal cells were persistent, and very electron-opaque when viewed by TEM, and were present after the microsporocytes had degenerated. A feature of this mutant is the accumulation of cell wall material lining the empty locule of the anther. The ultrastructural appearance of this wall is remarkably similar to sporopollenin, the component of microspore exines, which does not develop until the end of the tetrad period in wild-type anthers. As in ms5, more than one cell type is affected, thus making it difficult to deduce possible functions of the gene products. In the mutant ms5, systemic cell death appears to occur in all anther cells at early microsporocyte stage. From the earliest stage observed, all cells of the anther show degraded D N A and cell walls, and highly vacuolate cytoplasm with few organalles. These data suggest that ms5 may prove to be expressed in all cells of the anther, and not just in microsporocytes or tapetum. In the mutant ms4, the first evidence of abnormal development is not seen until the dividing microsporocyte is at meiosis I. A feature of this mutant (and of

ms15) is the presence of a sporopollenin-like wall, resembling fused orbicules lining the mature anther locule. Although anther wall cells appear normal, microsporocytes fail to complete meiosis and development is arrested at the dyad stage; thus wild-type Ms4 product may be involved in meiotic division. A functional description of the gene products of these four mutants must await molecular cloning of the genes concerned. The delayed apical senescence, higher number of inflorescences and flowers, and the formation of terminal aberrant flowers in the male-sterile mutants are likely to be the result of lack of fertilization rather than pleiotropic effects of a particular ms mutation, because the above phenotypes are present in all ms examined. Fertilization causes the release of growth hormones such as gibberellins and auxins (Pharis and King 1985). These hormones may play an important role in fertile plants in controlling the number of flowers and inflorescences. Auxin is known to cause apical dominance; thus, in male-sterile plants, a relative reduction in the concentration of auxins may cause a reduction of apical dominance and an increase in the number of inflorescences. Musgrave et al. (1986) have reported that malesterile plants have high levels of cytokinin. While we have not measured the level of growth hormones in these mutants, we have shown that the short length of siliques could be reverted by the addition of exogenous giberellins in one mutant phenotype. It would be interesting to see if other sporophytic phenotypes associated with the male-sterile mutants can be reverted by the exogenous addition of one or more growth hormones. Acknowledgements We thank the Australian Research Council for financial support and students of 606-305 class of 1992 (University of Melbourne) who contributed significantly towards the microscopic analyses.

References Block MD, Debrouwer D (1992) In situ enzyme histochemistry on plastic-embedded plant material. The development of an artefact-free 13-glucuronidase assay. Plant J 2:261-266 Chaudhury A, Craig S, Bloemer KC, Farrell L, Dennis ES (1992) Genetic control of male fertility in higher plants. A ast J Plant Physiol 19:419-426 Heslop-Harrison J (1979) Aspects of the structure, cytochemistry and germination of the pollen of rye (Secale cereale L.). Ann Bot Suppl 1 : 1-42 Kaul MLH (1988) Male sterility in higher plants. Springer, Berlin Heidelberg New York Knox RB (1984) The pollen grain. In: Johri JL (ed) Embryology of angiosperms. Springer, Berlin Heidelberg New York pp 197271 Koltunow AM, Truettner J, Cox KH, Wallroth M, Goldberg RB (1990) Different temporal and spatial gene expression patterns occur during anther development. Plant Cell 2:1201-1224 Last RL, Fink GR (1988) Tryptophan-requiring mutants of the plant Arabidopsis thaliana. Science 240:305-309 Mascarenhas JP (1190) Gene activity during pollen development. Annu Rev Plant Phys Plant Mol Biol 41:317-338 Meyerowitz EM, Pruitt RE (1985). Arabidopsis thaliana and plant molecular genetics. Science 229:1214-1218

28 Moffatt BA, Sommerville C (1989) Positive selection for male-sterile mutants of Arabidopsis lacking adenine phosphoribosyl transferase activity. Plant Physiol 86:115-1154 Musgrave ME, Antonovics J, Siedow JN (1986). Is male-sterility in plants related to lack of cyanide-resistant respiration in tissues? Plant Science 44:7-411 Nester JE, Zeevaart JAD (1988) Flower development in normal tomato and a gibberellin-deficient mutant. Am J Bot 75:45-55 Pharis RP, King RW (1985) Gibberellins and reproductive development in seed plants. Annu Rev Plant Physiol 36:517-568 Regan SM, Moffat BA (1990) Cytochemical analysis of pollen development in wild-type Arabidopsis and a male-sterile mutant. Plant Cell 2:877-889 Rick CM (1948) Genetics and development of nine male-sterile tomato mutants. Hilgardia 18:599-633

Sawhney VK, Greyson RI (1973) Morphogenesis of the stamenless-2 mutant in tomato. I. Comparative description of the flowers and ontogeny of stamens in the normal and mutant plants. Am J Bot 60:514-523 Spurt AH (1969) A low viscosity epoxy resin embedding medium for electron microscopy. J Ultrastruct Res 26:31-43 Theerakulpisut P, Xu H, Singh MB, Pettit JM, Knox RB (1991) Isolation and developmental expressions of Bcp 1, an antherspecific cDNA clone in Brassica carnpestris. Plant Cell 3:10731084 Van der Veen JH, Wirtz P (1968) EMS-induced genic male-sterility in Arabidopsis thaliana: a model selection experiment. Euphytica 17: 371-377