The Plant Cell, Vol. 5, 1217-1229, October 1993 O 1993 American Society of Plant Physiologists Anther Development: Basic Principles and Practical Applications Robert B. Goldberg,' Thomas P. Beals, and Paul M. Sanders Department of Biology, University of California, Los Angeles, California 90024-1606 INTRODUCTION Male reproductiveprocessesin flowering plants take place in the stamen (Esau, 1977).This sporophyticorgan system contains diploid cells that undergo meiosis and produce haploid male spores, or microspores. Microspores divide mitotically and differentiate intomulticellularmalegametophytes, or pollengrains, that contain the sperm cells. Figures 1and 2 show that the stamenconsistsof two morphologicallydistinct partsthe anther and the filament. The filament is a tube of vascular tissue that anchors the stamen to the flower and serves as a conduit for water and nutrients. By contrast, the anther contains the reproductive and nonreproductivetissues that are responsible for producingand releasing pollengrains so that pollination and fertilization processes can occur within the flower. Figure 1showsthat anther development can be divided into two general phases. During phase 1, the morphology of the anther isestablished,cell and tissue differentiationoccur, and microspore mother cells undergo meiosis.At the endof phase 1,the anther contains most of itsspecializedcellsandtissues, and tetrads of microsporesare presentwithin the pollen sacs (Figure 1). Duringphase2, pollengrains differentiate, the anther enlarges and is pushed upward in the flower by filament extension, and tissue degeneration, dehiscence, and pollen grain releaseoccur (Figure 1). The cellular processesthat regulate anther cell differentiation, establish anther tissue patterns, and cause the anther to switch from a histospecification program (phase l)to acelldegenerationand dehiscence program (phase 2) are not known. The developmentalevents leadingto anther formation and pollen release are exquisitely timed and choreographed (Koltunow et al., 1990; Scott et al., 1991). Cell differentiation and dehiscence events occur in a precise chronological order that correlates with floral bud size (Koltunow et al., 1990; Scott et al., 1991). This permitsthe mechanisms responsible for cell-type differentiation, tissue degeneration, and cellspecific gene activation within the anther to be explored with relative Base. During the past few years, there has been an explosive burst of interest in anther biology, both as a system to dissect plant developmentalprocessesat the molecular and genetic levels(Koltunowet al., 1990;Gasser, 1991; McCormick, To whom correspondenceshould be addressed. 1991; Scott et al., 1991; Chaudhury et al., 1992; Dawe and Freeling, 1992; Spena et al., 1992; van der Meer et al., 1992; Worrall et al., 1992; Aarts et al., 1993; Preusset al., 1993)and for practicalgenetic engineering studies to improvecrop plants (Mariani et al., 1990, 1992; Denis et al., 1993; Schmülling et al., 1993). In this review,we outlinethe major events that occur during anther development, using our studies of the tobacco anther as a focal point (Koltunow et al., 1990). We concentrate primarily on processes concerned with anther cell-type differentiation and dehiscence, because others in this volume have concentratedon the male gametophyte and have reviewed pollen grain biology and development (Mascarenhas, 1993,this issue; McCormick, 1993,this issue). STAMEN INlTlATlON IS CONTROLLED BY HOMEOTIC GENE INTERACTIONS As shown in Figure 2 and represented schematically in Figure 1, stamen primordiaare initiatedas awhorl of small bumps at specific locations on the floral meristemsurface. The number of stamen primordiavaries in different plant species, but their locationon the floral meristemisfixed in what is referred to as the thirdwhorl (Coen, 1991; Coenand Meyerowitz, 1991). Intobacco, for example, five stamen primordiaare initiatedin the third whorl (Figures 1 and 2) over a 1- to 2-day period (Koltunowet al., 1990).Stamen primordiaappear after the sepal and petalprimordiahave initiatedbut priorto the initiation of carpel primordia(Figure2). A short time after stamenprimordia initiation occurs, the anther and filament compartments differentiate, two bilaterallysymmetrical theca are established within the anther, major anther cell types and tissues form, and the anther takes on its characteristic paddlelike shape (Figures 1 and 2). Stamen primordia initiation does not depend uponthe presente of sepal or petal primordia, because surgical removal of these primordiadoes not preventthird-whorlprimordiafrom becoming stamens (Hicks and Sussex, 1971). In addition, homeoticmutationsthat affect the identity of petal and sepal primordiado not affect stamen primordiainitiationand differentiation (Coen, 1991; Coen and Meyerowitz, 1991). Thus, 1218 The Plant Cell Sepal Petal Carpel Sepal Stamen Petal Phase 1| Stamen Cell Specification Tissue Differentiation Meiosis and Spore Formation Stages -6 to -1 '/ -Filament Stage -7 Stage +1 Phase 2| Pollen Grain and Sperm Cell Differentiation Anther Enlargement and Filament Extension Tissue Degeneration Dehiscence and Pollen Release Stages + 2 to+10 CCC PG Stage +1 Stage +11 Stage +12 Figure 1. A Generalized Overview of Anther Development. Schematic representations of anther developmental stages and cross-sections are based on scanning electron and light microscopy studies of tobacco anther development (Koltunow et al., 1990; Drewset al.,1992). The dashed line through the stage 1anther drawn in the phase 1portion of the figure represents the cross-section plane for anthers drawn schematicallyin phase 2. The verticallines drawn through the endothecium at stages 11and 12 represent fibrous cell wall bands. C, connective; CCC, circular cell cluster; E, epidermis; En, endothecium; PG, pollen grain; PS, pollen sac; St, stomium; T, tapetum; Td, tetrads; Th, theca; V, vascular bundle. stamen primordia specification occurs independently of the presence of other floral organ primordia and vice versa. What processes control stamen morphogenesis and the differentiation of diverse stamen cell and tissue types are not known. Genetic and molecular studies have shown that stamen primordia specification is controlled by the interaction of several homeotic genes (Figure 1), including APETALA3 (APS), PISTILLATA (PI),and AGAMOUS(AG) in Arabidopsis and their counterparts DEFICIENS (DEF), GLOBOSA (GLO), and PLENA (PLE) in snapdragon (Schwarz-Sommer et al., 1990; Coen, 1991, 1992; Coen and Carpenter, 1993, this issue; Okamuro et al., 1993, this issue). Mutations in any one of these genes result in the loss of stamen primordia and the conversion of third-whorl primordia to adifferent floral organ type (SchwarzSommer etal., 1990; Coen, 1991,1992; Coen and Meyerowitz, 1991). These homeotic genes also play a role in the specification of other floral primordia and are discussed in detail by Okamuro etal. (1993, thisissue) and Coen andCarpenter (1993, this issue). Related homeotic genes have also been identified in tobacco (Hansen et al., 1993), petunia (Angenent et al., 1992; van der Krol and Chua, 1993, this issue), tomato (Pnueli et al., 1991), and corn (Schmidt et al., 1993; Veil et al., 1993, this issue), indicating that the same genes control stamen primordia specification in distantly related plants. Homeotic genes controlling stamen specification encode transcription factors related to those in yeast and mammals and contain a conserved MADS box region that probably represents the DMAbinding domain (Coen and Meyerowitz, 1991). DEF (AP3) and GLO (PI) encode transcription factors that interact as heterodimers with their target DMAsequence motifs and, with the PLE (AG) gene product, activate genes required for stamen specification inthe third whorl (Coen, 1992; Schwarz-Sommer et al., 1992; Trobneret al., 1992). In situ hybridization studies have shown that these homeotic genes are expressed in the predicted pre-third-whorlterritories ofthe floral meristemjust prior tostamen primordia initiation andthat they Anther Development 1219 continue to be expressed throughout stamen development in both the anther and filament (Schwarz-Sommer et al., 1990, 1992; Coen and Meyerowitz, 1991; Coen, 1992). Thus, the floral homeotic genes DEF (AP3), GLO (PI), and PLE (AG) set off a cascade of events that cause third-whorl primordia to follow a stamen differentiation pathway (Figure 1). Although progress has been made in identifying genes required for stamen primordia specification, several crucial questions remain unanswered. First, which genes control the stamen primordia identity genes, and howdo these upstream genes activate DEF(APS), GLO(PI), and PLE(AG) in the proper third-whorl territories? That is, what processes regulate the position and number of stamen primordia on the floral meristem? Several homeotic genes have been identified in Arabidopsis and snapdragon that play a role in regulating floral meristem identity (Schwarz-Sommeret al., 1990; Coen, 1991,1992; Coen and Meyerowitz, 1991; Coen and Carpenter, 1993, this issue), including SOUAMOSA and FLORICAULA in snapdragon (Coen et al., 1990; Schwarz-Sommer et al., 1990), and their counterparts APETALA1 (Mandel et al., 1992) and LEAFY (Weigel et al., 1992) in Arabidopsis (Okamuro et al., 1993, this issue). It is not yet known, however, which genes and processes function in the interval between the activation offloral meristem identity genes and those involved in stamen primordia specification. Second, and most important, it is not known what target, or downstream, genes are activated transcriptionally by the DEF (AP3), GLO (PI), and PLE (AG) proteins within the stamen primordia. That is, what are the genes whose products are required to carry out stamen morphogenesis and tissue differentiation programs after the stamen primordia have been specified? Among these downstream genes should be those that (1) regulate the differentiation of the stamen into anther and filament compartments, (2) control the shape of the anther and its size, (3) initiate cell and tissue specification programs required forthe development of a mature anther, and (4) activate anther cell-specific gene sets. Clearly, the identification of the downstream genes and their products should provide insight into the detailed mechanismsresponsible for anther morphogenesis and differentiation. STAMEN DEVELOPMENT IS INFLUENCED BY NUCLEAR AND MITOCHONDRIAL GENOME INTERACTIONS Figure 2. Scanning Electron Micrographs of Tobacco Stamen Primordia Development. Floral buds and anthers harvested at the designated stages were photographed in the scanning electron microscope(Koltunowet al.,1990; Drews et al.,1992). Magnification factors and stage numbers are marked on each micrograph. Numbers have not been assigned to stagesprior to -7, which are collectively referred to as pre -7 (Figure 1). A, anther; C, carpel; F, filament; P, petal; S, sepal; T, theca. Several homeotic-like mutants have been identified in tobacco that affect stamendevelopment (Rosenberg and Bonnett, 1983; Kofer et al., 1990,1991,1992). These mutants were identified originally as male-sterile progeny of crosses between different tobacco species (Sand, 1968; Hicks et al., 1977) and, more recently, in plants regenerated from tissue culture after fusion of protoplasts from different male-sterile cultivars (Kofer et al., 1990, 1991, 1992). Unlike male-sterile homeotic mutants in Arabidopsis and snapdragon that arecaused by nuclear gene lesions, tobacco homeotic-like mutants are inherited cytoplasmically and are associated with alterations in specific mitochondrial DNAfragments (Kofer et al., 1991,1992). This implies that mitochondrial genes play a role in the stamen differentiation pathway. The phenotypes of tobacco homeotic-like mutants vary,but they correlate with the presence of a specific cytoplasm and mitochondrial genome (Kofer et al., 1991,1992). For example, in some mutants, the third-whorl primordia emerge and then abort just prior to carpel primordia initiation (Rosenberg and Bonnett, 1983). In other mutants, the third-whorl primordia develop into petal-like structures or antherless stamens whose filaments have stigmatoid structures at their tips (Hicks et al., 1977; Rosenbergand Bonnett, 1983). The varied effects of the mitochondrial genome on stamen development suggest that the nuclear/mitochondrial gene interactions occur after stamen primordia specification (Hicks etal., 1977; Rosenberg and Bonnett, 1983). If so, this suggests that the altered cytoplasms interfere with the activation and/or function of the downstream genes required for stamen differentiation and morphogenesis. The question of how mitochondrial and nuclear gene interactions affect the stamen differentiation pathway remains unresolved. 1220 The Plant Cell ANTHER CELL TYPES CAN BE TRACED TO SPECIFIC FLORAL MERISTEM LAYERS At the end of phase 1 (Figure 1), a differentiated anther has several highly specialized cells and tissues that are responsible for carrying out nonreproductive functions (e.g., support and dehiscence) and reproductive functions (e.g., spore and pollen formation). Schematic representations of differentiated stage 1 tobacco anther cross-sections are shown in Figures 1 and 3, actual bright-field photographs are shown in Figure 4, and anther cell types and their functions are listed in Table 1. The nonreproductive tissues include the epidermis, endothecium, tapetum, circular cell cluster, connective, stomium, and vascular bundle (Table 1). Each of these tissues and cell types carries out specialized tasks (Esau, 1977). For example, the stomium and circular cell cluster are involved in dehiscence (Bonner and Dickenson, 1989), the tapetum plays a role in pollen wall formation (Esau, 1977; Pacini et al., 1985), and the connective is responsible for anchoring the four pollen sacs together into a single anther structure (Weberling, 1989). In addition to these diploid sporophytic tissues, the anther also contains haploid microspores that fill the pollen sacs and differentiate into pollen grains (McCormick, 1991, 1993, this issue; Scott et al., 1991; Mascarenhas, 1993, this issue). Cytochimera cell lineage studies performed with the Jimson weed (Datura stramonium), a solanaceous relative of a tobacco, showed that the floral meristem consists of three "germ" layers, designated L1, L2, and L3 (Satina et al., 1940), which give rise to different anther tissues following stamen primordia initiation (Satina and Blakeslee, 1941). Thus, once specified, the developmental fate of L1, L2, and L3 layer STAGE 10 Phase 1 Anther Primordium Stage -7 Phase 2 Inner Tapetum Microspores Connective Vascular Bundle Outer Tapetum Endothecium Stomium Epidermis Anther Stage +1 Circular Cell Cluster Dehiscence Figure 3. Cell Lineages and Major Events That Occur during Anther Differentiation and Dehiscence. Stages, timing of events, and anther schematic representations were taken from Koltunow et al. (1990) for tobacco anther development. Similar events take place during anther development in other plants (Esau, 1977). Cell lineages derived from the L1, L2, and L3 primordia layers were inferred from the histological observations of Satina and Blakeslee (1941), Esau (1977), and Koltunow et al. (1990). Anther Development 1221 0) V) (0 CM 0) (A CO -6 -2 -1 8 11 st En Figure 4. Bright-Field Photographs of TobaccoAnther Development. Anthers at the designated stages were fixed, embedded in paraffin, and sliced into 10-nm transverse sections through the plane drawn schematically in Figure 1. The sections were stained with toluidine blue and photographed with bright-field illumination. Pictures aretaken from Koltunow et al. (1990), except for stage 11. A, archesporial cells; C, connective; CCC,circular cell cluster; E,epidermis; En,endothecium; MMC, microspore mother cells; Msp, microspores; P, parietal cells; PS, pollen sac; Sp, sporogenous cells; St, stomium; T,tapetum; TDS, tetrads; V, vascular bundle. derivatives is fixed. Figure 3 shows a schematic representation of a tobacco stamen primordium and the anther cell lineages that are inferred from histological analysis to be derived from the L1, L2, and L3 layers (Satina and Blakeslee, 1941; Esau, 1977; Koltunow et al., 1990). In most cases, individual tissues and cell types arederived from a single germ layer. For example, the L1 layer gives riseto the epidermis and the stomium; the L2 layer gives rise to the archesporial cells, microspore mother cells, endothecium, and middle-wall layers that lie between the epidermis and tapetum; and the L3 layer gives riseto the connective, vascular bundle, and circular cell cluster adjacent to the stomium (Figure 3). By contrast, both the L2 and L3 layers contribute to tapetum formation (Figure 3). Tapetal cells along the upper (inner) portion of the pollen sacs are specified from the LS-derived connective tissue, whereas those that line the lower (outer) portion of the pollen sacs are specified from the L2-derived archesporial lineage (Figure 3; Esau, 1977). The underlying mechanisms responsible for the differentiation of specific anther cell types after stamen primordia initiation occurs are not known. CELL TERRITORIES MAY PLAY A ROLE IN THE ANTHER DIFFERENTIATION PROCESS Figures 3 and 4 show that cells destined to become major anther tissues and cell types form at precise times and positions during anther development. For example, archesporial cells destined to differentiate into microsporangia (pollen sacs) and surrounding tapetum and endothecium tissues arise simultaneously in each corner of the anther primordium (Figures 3 and 4). By contrast, vascular tissues differentiate within the Table 1. Functions of Anther Cell and Tissue Types Cell or Tissue Type8 Major Function6 Connective Circular Cell Cluster0 Endotheciumd Epidermis Microspore Middle Layerd Stomium Tapetum Vascular Bundle Join anther theca together; connect anther to filament; structure, support, and morphology Dehiscence Structure and support; dehiscence Structure and support; prevent water loss; gas exchange; dehiscence Pollen grain and sperm cell development Structure and support; dehiscence Dehiscence Pollen wall components; nutrients for pollen development; enzymes for microspore release from tetrads Connection between anther, filament, and flower; nutrient and water supply 8 Cells present in a stage 1 tobacco anther (Figures 1 to 4). "Taken from Esau (1977), Weberling (1989), and Koltunow et al. (1990). c Also referred to as the crystal-containing idioblasts (Trull et al., 1991), intersporangial septum (Bonner and Dickenson, 1989), and hypodermal stomium (Horner and Wagner, 1992). d Collectively referred to as the anther wall. 1222 The Plant Cell center of the anther primordium and establish a connection with the filament (Figures3 and 4). Other tissues, such as the circular cell cluster and stomium, form at the boundaries between each microsporangialpair later indevelopment (Figures 3 and 4). Thus, specific regions,or territories, are established early in anther development within which unique histodifferentiation events occur in a precise chronological order. No information exists about whether the differentiation of anther cells and tissues is induced (or inhibited) by signals originating from contiguous regions of a territory, or is controlledautonomouslyby sequesteringmorphogeneticfactors during divisions of the L1, L2, and L3 layers, or both. The territory-baseddifferentiationof four microsporangiawith identicaltissue patternsand the differentiationof tapetalcellswithin each territory from two different cell lineages (Figure 3) suggest that cell-cell communication processes may play an important role in the anther histospecificationprocess. For example, it is conceivable that a gradient of morphogens, synthesizedby archesporial cells and interpreted by contiguous L2 and L3 cells, triggers the differentiationof tapetal and endothecial cells (Figures 3 and 4). Other morphogens appearing later at the boundaries of the microsporangia could be perceived by neighboring L1-derived epidermal cells and L3-derived connective cells and cause them to differentiate intothe stomium and circular cellcluster, respectively (Figures 3 and 4). lnterpretation of putative intercellular signalswould lead ultimatelyto the transcriptionalactivationofgenes responsible for establishing the differentiated fate of each cell type within the territory. Although speculative, this model leads to testable predictions. Severa1lines of evidence suggest that intercellular signals play an important role in determiningthe fate of cells within each floral meristem germ cell layer. These studies are reviewed by Sussex (1989)and by Hualaand Sussex (1993,this issue). Genetic chimera experimentshave indicatedthat the position of L1, L2, and L3 cells within the floral meristem is the primary factor in specifyingtheir fate rather than priorcell ancestry (Sussex, 1989; Hualaand Sussex, 1993,this issue). For example, rare divisions that enable L2 cells to penetrate or invade the L1 meristem layer cause these cells to follow an epidermalpathway.Conversely, penetrationof L1cells into the L2 layer causes these cells to have a fate identical to that of surroundingL2cells. In addition, recent studies usinggene markersspecific for each germ layer have shown that the L3 layer influencesthe developmentof cells derivedfrom the L1 and L2 layers (Szymkowiak and Sussex, 1992). These experimentssuggestthat cellswithin each meristemgerm layer are conditionallyspecified(Davidson, 1991) and that the ultimate fate of these cells depends upon their position in the floral meristem. That is, position-dependent communication processes probablyoccur between adjacent meristematiccells, and these signaling events influencefloral organ cell specification pathways. The extent to which cell-autonomous and position-dependent cell interaction processes play a role in the differentiationof specialized cell types derivedfrom each anther primordiumgerm layer (Figure 3) remains a major unanswered question. ANTHER DEHISCENCE INVOLVES THE PROGRAMMED DESTRUCTION OF SPEClFlC CELL TYPES One of the major mysteriesof anther development is how a differentiated anther switchesfrom a histodifferentiation program (phasel)to a cell degeneration and dehiscence program (phase 2) that leads ultimately to pollen release and stamen senescenceat flower opening. The dehiscence program begins after the formation of tetrads, results in the sequential destruction of specific anther cell types, and is coordinated temporally with the pollen differentiationprocess (Figure 1). In tobacco, the entire dehiscence and cell degeneration process occurs over a l-week period (Koltunow et al., 1990) and involves three major cell types that are formed during phase 1from different anther primordiumcell layers(Figure 3).These include the stomium, the endothecium, and the circular cell cluster (Table 1; Figures 1 and 4). The latter tissue, which is common in solanaceous anthers, is also referred to as the hypodermal stomium (Horner and Wagner, 1992), the intersporangialseptum (Keijzer, 1987; Bonner and Dickenson, 1989) and, becauseit containscalcium oxalatecrystals, the crystalcontaining idioblasts (Trull et al., 1991; Horner and Wagner, 1992). The onset of the dehiscence program sets off an ordered series of events within the anther, including (1) the formation of fibrous band thickeningson the endothecial cell walls, (2) degenerationof the circularcellcluster and merging of the two pollen sacs in each theca into a single locule, (3) breakdownof the tapetum and connective, and (4) rupture of the anther at the stomium and pollen release (Figures 1and 4; Bonner and Dickenson, 1989). Male-steriletobacco anthers that lack pollengrains and tapetal cells undergoa normaldehiscenceprocess(Koltunow et al., 1990; Marianiet al., 1990), indicating that dehiscence is not set into motion by signals derived from either the tapetum or differentiatingpollengrains. The dehiscence program probably requiresthe activation of manygenes, includingthose thatencode hydrolyticenzymes requiredfor cell death, such as RNases, proteases, and cellulases(Keijzer, 1987; Bonner and Dickenson, 1989;de1Campillo and Lewis, 1992). One marker for the dehiscence process in tobacco is the TA56 thiol endopeptidase gene (Koltunow et ai., 1990).Asshown in Figure5, TA56mRNAaccumulatesfirst in the circular cell cluster prior to its destruction, then in the stomium, and finally in the connective. Although the role of the TA56thiol endopeptidasegene inthe dehiscenceprocess is not known, it marks the cell-specific degeneration events that occur prior to stomium rupture and pollen release from the anther (Figures 1 and 4; Koltunow et al., 1990). Recent experimentswith chimeric genes containingTA56 5'sequences showed that the sequential accumulation of TA56 mRNA in specific anther cell types is controlled primarily at the transcriptionalleve1(T. F! Beals and R. B. Goldberg, unpublished results). Thus, the onset of the dehiscence program results in the transcriptionalactivationof genes that are inactive duringthe histodifferentiationphaseof anther development(Figure 1). Which cellular events trigger the dehiscence program, activatedehiscencespecificgenes, andcoordinatepollengrain Anther Development 1223 DEVELOPMENTAL STAGE m R N A - I I Z 3 4 5 6 7 8 9 10 II 12 TA13 TA36 TA29 TA32 TA26 TA20 TA39 TA55 TA56 CCC W MS Figure 5. Temporaland Spatial Regulation of mRNAs during Tobacco Anther Development. Data represent a summary of RNA gel blot and in situ hybridization studies performed in our laboratory with anther-specific cDNAs (Koltunow et al., 1990; T.P. Seals and R.B.Goldberg, unpublished results). Proteins encoded by these mRNAs and their cell specificities are summarized in Table 2. The colors present in each bar correspond with the cell and tissue types in which each mRNAis localized at that period of anther development. The tapering in each bar shows periods when the mRNA accumulates or decays, and the bar thickness approximates the relative mRNA prevalences. C, connective; CCC,circular cell cluster; E, epidermis; MS, microspore; S, stomium; T, tapetum; V,vascular bundle; W, wall. The wall includes the endothecium and middlelayer (see Table 1). Figure adapted from Koltunow et al. (1990). differentiation with the sequential destruction of specific anther tissues are not known. GENE EXPRESSION IS TEMPORALLY AND SPATIALLY REGULATED DURING ANTHER DEVELOPMENT RNA-excess DNA/RNA hybridization experiments with singlecopy DNA indicated that there are ~25,000 diverse genes expressed in a stage 6tobacco anther (Figure 4; Kamalay and Goldberg, 1980). The vast majorityof these genes encode rare mRNAs that constitute less than 0.001% of the mRNA mass when averaged over the entire anther mRNA population. Comparison of floral and vegetative organ system RNApopulations showed that ~10,000 diverse anther mRNAsare undetectable in the nuclear RNA and mRNA populations of other organs and are anther specific (Kamalay and Goldberg, 1980,1984). Although it is not known how many additional anther-specific genes are expressed during phase 1,when major histospecification events occur (Figure 4), these results indicate that the differential regulation of a large number of genes is required to specify and maintain the differentiated state of cells and tissues within the anther. Experiments conducted in several laboratories have identified cDNA clones that represent prevalent mRNAs that are present exclusively, or at elevated levels, in the anther (Koltunow et al., 1990; Smith etal., 1990; Evrard etal., 1991; McCormick, 1991,1993, this issue; Nacken et al., 1991; Scott et al., 1991; Shen and Hsu, 1992). Table 2 lists several tobacco antherspecific mRNAs studied in our laboratory. Anther-specific mRNAs have been shown to encode lipid transfer proteins, protease inhibitors, thiol endopeptidases, glycine-rich and proline-rich polypeptides with properties of cell wall proteins, pectate lyases, polygalacturonases, and chalcone synthase isoforms (Table2; Koltunow et al., 1990; McCormick, 1991; Scott et al., 1991). Many anther-specific mRNAs, such as TA20,are found exclusively in sporophytic tissues (Table 2; Koltunow et al., 1990). Others are pollen grain specific or, like TA39 (Table 2), are present in both sporophytic and gametophytic cells (McCormick, 1991; Scott et al., 1991). Experiments conducted with anther-specific cDNAs have demonstrated the striking degree to which different temporaland cell-specific gene expression programs occur during anther development (Koltunow etal., 1990; McCormick, 1991, 1993, this issue; Scott et al., 1991). Figure 5 presents a Table 2. mRNA8 '" TA13 TA20 TA26 TA29 TA32 TA36 TA39 TA55 TA56 Gene Markers for Tobacco Protein Encoded Glycine-rich Unknown Unknown Glycine-rich Lipid transfer Lipid transfer Unknown Unknown Thiol endopeptidase Anther Cell Types Cell Specificity Tapetum Connective, stomium, endothecium, middle layer Tapetum Tapetum Tapetum Tapetum Tapetum, microspore Tapetum, connective, endothecium, middle layer Connective, circular cell cluster, stomium a Taken from Koltunow et al. (1990) and T.P. Beals and R.B. Goldberg (unpublished results). b Gene markers for anther cells in other plants have been reviewed recently (Gasser, 1991; McCormick, 1991; Scott et al., 1991). 1224 The Plant Cell schematic representationof several different tobacco antherspecific gene expressionprogramsthat have been identified recently inour laboratory(Koltunowet al., 1990;T.P. Beals and R.B. Goldberg, unpublishedresults). One strikingexample is the coordinated accumulation and decay of the tapetal-specific TA13, TA26, TA29, TA32, and TA36 mRNAs during early phase 2 (Figure 5; Koltunow et al., 1990). This pattern, and others shown in Figure5, probably represent only a few of the gene expression programs that occur during anther development (Evrardet al., 1991;McCormick, 1991; Scott et al., 1991). However, these patterns illustrate that (1) genes expressed in the anther are regulated by many differentcontrol programsthat are partitionedwith respectto bothcell type and developmental time, (2) anther-specific gene expressionprogramscorrelate with the differentiation and degeneration of specific anther tissues andcelltypes, and (3)anther cell-specific mRNAsprovide useful markersfor dissectingthe regulatory pathwaysthat control the specification of different cell types during anther development. The mechanisms responsible for establishing cell- and temporal-specific gene expression patterns during anther development are not known. TRANSCRIPTIONAL PROCESSES CONTROL ANTHER-SPECIFIC GENE EXPRESSION PROGRAMS Solution hybridization experiments carried out with anther mRNA and nuclear RNA populations suggested that most anther-specific genes are under transcriptional control (Kamalay and Goldberg, 1984). Run-offtranscriptionstudies andtransformation experimentswith chimericgenes containing anther-specificgene promoters demonstrated directly that the temporal and spatialgene expression programsthat operate in the anther are controlled primarily by transcriptional processes (Koltunowet al., 1990; McCormick, 1991,1993, this issue; Scott et al., 1991).For example,the tapetal-specificTA29 gene (Table 2 and Figure 5) is not transcribed detectably in other plant organs, and chimeric genes with TA29 S'sequences are active only in the tapetum (Koltunow et al., 1990; Mariani et al., 1990, 1992). Chimeric genes driven by other antherspecific gene promoters, such as TA56 (Table 2 and Figure 5),producetranscription patternsthat also reflectthe cell-and temporal-specific expression programscharacteristic of their parental, or endogenous, genes (McCormick, 1991; Scott et al., 1991; T.P. Beals and R.B. Goldberg, unpublishedresults). These results suggest that the mosaicof gene expression programs that function in the anther (Figure 5) reflects the distributionof transcriptionfactors capable of activatingunique gene sets in each specialized cell type. If so, this impliesthat anther cell differentiationevents lead directly or indirectly to the expressionof regulatory genesthat are necessaryto activate each cell-specific gene set. The identificationof anther transcriptionfactors and their correspondinggenes should provide usefulentry points intoanther cell-specific differentiation pathways(Figure 3). How genes encodinganther transcription factors are regulatedwith respectto cell type and time during anther development, activate batteries of cell-specificgene sets, and connectwith regulatory circuitsthat ultimately trace backto homeoticgenes controllingstamen primordiaspecification (Figure 1) remain major unanswered questions. CELL ABLATION STUDIES WlTH CYTOTOXIC GENES CAN BE USED TO STUDY ANTHER CELL DlFFERENTlATlONPROCESSES One powerful use of the anther cell-specific promotersidentified in the gene expression studies (Table 2 and Figure 5) isto fusethem to cytotoxic structuralgenes, such as the diphtheria toxin A-chain gene (Palmiter et al., 1987), the RNase T1 gene (Mariani et al., 1990), or the barnase gene (Mariani et al., 1990),to selectively destroy specific cell types during anther development. This strategy, illustratedin Figure 6,is an extension of the microbeam laser cell ablation approach that (1)providedthe first insightsintothe roleof cell-cell interactions in Caenorhabditiselegans cell-typespecificationand (2) laid the foundation for molecular and genetic approaches that identified genes encoding components of different cell inductionpathways (Sulstonet al., 1983;Horvitzand Sternberg, 1991). 60th of these strategies rely on the selective destruction of a specific cell type and an identificationof the effect, if any, that the loss of this cell type has on the differentiation of contiguous cells (Palmiter et al., 1987). The microbeam laser approachis not readily applicableto studies of anther cell differentiation,becausethe anther consists of many cell layers and is contained within a closed flower bud (Figures 1 to 4). Thus, specific cell targets are refractory to identificationand physical destruction by a laser beam. By contrast, ablation studies using"genetic lasers" in transgenic plants can, in theory, be applied to an unlimited number of cell types. All that is required is a set of cell-specificpromotersthat is active in a variety of anther cell types (Table 2). We fused the tobaccotapetal-specific TA29 gene promoter (Table2 and Figure5)to several cytotoxic genes to selectively destroy the tapetum during anther development (Figure 6; Koltunowet al., 1990;Mariani et al., 1990).These experiments showedthat the tapetum is critical for pollen developmentbut that it is not necessary for the function and/or degeneration of other anther cell types during phase 2 of anther development (Koltunow et al., 1990; Marianiet al., 1990). That is, the tapetum functions in a cell-autonomousmanner following its formation during phase 1 of anther development (Figures 3 and4), and tapetal cells are not requiredfor the inductionand/or completion of the dehiscence program(Figure 1). Of particular interest, of course, is how the loss of cells formed during the early stages of anther development would affect the differentiation of adjacent cell types (Figures 3 and 4). For example, how would the loss of primary parietal cells affect the differentiationof contiguoussporogenouscells within developing microsporangialterritories? These ablation studies, Anther Development 1225 Male Sterile Plants Restored Male Fertile Plants MS NOT No MS MS Normal Tapetum Development [TA29J BARNASE)' |TA29| BARNASE) |TA29| BARSTAR) Tapetal- specific Barnase Destruction of RNA in Tapetum Tapetal- specific Barnase \ Tapetal- specific Barstar Tapetum Destroyed Barnase INACTIVE Barnase/Barstar Complex Normal Tapetum Development Figure 6. Genetic Engineering for Male Fertility Control. A schematic representation of the male sterility and male fertility restoration experiments described by Mariani et al. (1990, 1992). MS, microspores; T, tapetum. however, will require promoters that are active in specific cell types of the developing anther primordium, such as the archesporial and parietal cells (Figures 3 and 4). The availability of anther primordium-specific promoters should permit athorough investigation of the lineage relationships, cell functions, and cell-cell interaction processes that occur during anther development (Figure 3). MALE-STERILE MUTANTS CAN BE USED TO GENETICALLY DISSECT ANTHER DEVELOPMENT Cell ablation studies can provide important conceptual information about the processes that control the differentiation of anther cell types. They cannot, however, identify genes that are responsible for directing cell specification events. To accomplish this will require, in part, a genetic approach analogous to that used in Drosophila and C. elegans to identify genes involved in both position-dependent and cell-autonomous differentiation pathways(Horvitz and Sternberg, 1991; Kramer et al., 1991). Male-sterile mutants have been identified in many plants, including Arabidopsis (Kaul, 1988). Although most male sterility genes and their wild-type alleles have not been cloned and studied (Aarts et al., 1993), phenotypic and histological analyses suggest that some male sterility mutations affect the differentiation and/or function of many anther cell types, including those in the stomium, tapetum, endothecium, and the archesporial and sporogenous layers of the anther primordium (Kaul, 1988; Chaudhury, 1993, this issue). Arabidopsis malesterile mutants are easy to identify because they flower for a longer time period, grow taller, and remain in a green or nonsenesced state longer than their male-fertile, wild-type counterparts (Feldmann, 1991; Chaudhury et al., 1992; Forsthoefel et al., 1992; Aarts et al., 1993; Preuss et al, 1993). Because large numbers of mutagenized plants can be screened readily (Feldmann, 1991; Chaudhury et al., 1992; Forsthoefel et al., 1992), genetic studies in Arabidopsis have the potential to dissect gene pathways that control both the 1226 The Plant Cell histodifferentiation program (phase 1) and the dehiscence and cell degeneration program (phase 2) of anther development (Figure 1). Recently, we screened a large number of Arabidopsis lines that were treated with either T-DNA (Feldmann, 1991; Forsthoefel et al., 1992) or ethyl methanesulfonate (EMS) (Koornneef et al., 1982)for male-sterile mutants. Wewere particularly interested in anther mutants that failed to dehisce and those that had defects in the anther differentiation pathway. Figure 7 shows two male-sterile mutants that were identified in our EMS screen. One mutant, designated as undeveloped anther, has stamens with normal filaments but anthers that lack pollen and have an undeveloped, club-shaped appearance (Figure 7A). The selective effect of this mutation on anther development suggests that distinct genes control anther and filament differentiation following stamen primordia specification. By contrast, the other mutant, designated as late dehiscence, has stamens that undergo a normal differentiation pathway and contain anthers that are morphologically similar to their wild-type counterparts (Figure 7B). Late dehiscence anthers, however, fail to dehisce on time (Figure 7B) and release their pollen grains after the stigma is no longer receptive to pollination. These male-sterile mutants and others with analogous phenotypes should permit the isolation of genes controlling anther developmental processes in the near future. ANTHER GENE EXPRESSION STUDIES HAVE LED TO THE GENETIC ENGINEERING OF CROP PLANTS FOR MALE FERTILITY CONTROL The anther plays a prominent role in crop production because it is responsible for carrying out male reproductive processes necessary for generating seeds that will produce the next plant Figure 7. Arabidopsis Male-Sterile Mutants That Are Defective in Anther Development. (A) undeveloped anther. (B) late dehiscence. A, anther; F,filament; OV,ovary; P,petal; S, sepal; SL, style; ST,stigma. generation (Figure 1). One of the major factors contributing to increases in crop productivity over the past few years has been the breeding of hybrid varieties in crops such as corn and oilseed rape (Brewbaker, 1964; Feistritzer and Kelly, 1987). Crosses between inbred plant lines give rise to hybrid progeny that, at times, exhibit heterosis or hybrid vigor (Brewbaker, 1964). Although the molecular basis of hybrid vigor is not known, hybrids with a heterotic phenotype are more resistant to disease, less susceptible to environmental stress, and generate higher yields than their inbred parents (Brewbaker, 1964; Feistritzer and Kelly, 1987). Increased crop yields are particularly important to keep the worlďsfood supply sufficient to meet the needs of an ever-expanding population (Klein, 1987). The reproductive biology of higher plants makes it difficult to perform directed crosses between inbred lines on a scale large enough for hybrid seed production. Because stamens and carpels (Figures 1 and 2) are present within the same flower (e.g., tobacco, oilseed rape, Arabidopsis) or are contained within separate flowers on the same plant (e.g., corn), a high level of self-pollination can occur within individuals of each parental line. Toovercome this problem and ensure that crosses occur berween inbred lines, several breeding strategies have been used, including (1) manual and/or mechanical removal of stamens from the flowers of one parental line, (2) use of self-incompatibility alleles that prevent self-pollination, and (3) employment of dominant cytoplasmic or nuclear male sterility genes that disrupt pollen development (Feistritzer and Kelly, 1987). Each strategy has limitations that prevent its use as a general approach for hybrid seed production in all crop plants. For example, many crop plants do not have selfincompatibility alleles, manual removal of stamens is labor intensive and impractical for crops with small bisexual flowers (e.g., oilseed rape), and male sterility genes require a restorer system. An important practical spin-off of anther cell differentiation and gene expression studies has been the development of a genetically engineered system for male fertility control that is applicable to a wide range of crop plants, including oilseed rape, tomato, lettuce, and corn (Mariani etal., 1990,1992). This system, which is illustrated schematically in Figure 6, takes advantage of the tapetal-specific transcriptional activity of the tobacco TA29 gene (Table2 and Figure 5; Koltunowet al., 1990) and an RNase/RNase-inhibitor defense system utilized by the bacteria Bacillus amyloliquefaciens (Hartley, 1989). The genetically engineered male fertility control system was developed as a collaborative effort between our laboratory and Plant Genetic Systems in Gent, Belgium (Mariani et al., 1990,1992). In brief, barnase is an extracellular RNase that is produced by B. amyloliquefaciens to protect itself from microbial predators (Hartley, 1989). Barstar, on the other hand, is a barnasespecific RNase inhibitor that is used intracellularly by B. amyloliquefaciens to protect itself from the cytotoxic effects of barnase (Hartley, 1989). As illustrated in Figure 6, a cytotoxic chimeric gene consisting of the TA29 gene promoter and barnase gene coding sequences leads to the selective ablation of anther tapetal cells and the production of pollenless, male- Anther Development 1227 sterile plants (Mariani et al., 1990). By contrast, an anticytotoxic chimeric gene containingthe same TA29gene promoter andthe barstar gene coding region does not affect tapetal cell development and is a dominant suppressor of the TA29/barnase gene (Figure 6). When male-sterileplantscontaining the TA29/barnasegene are crossed with male-fertile plants containing the TA29/barstar gene, progeny are produced that contain both genes, and these progeny are malefertile (Marianiet al., 1992). The presente of barstar within the tapetal cells of F, progeny restores malefertility by forming acomplex with barnasethat prevents barnase cytotoxic activity from destroying the tapetal cells within the anther (Figure 6). Becausethe TA29 gene promoter is regulated correctly in a wide range of dicot and monocot crop plants (Mariani et al., 1992),the barnaselbarstar system illustrated in Figure 6 has the potential for being universally applicable for hybrid seed production in awide variety of crops (Mariani et al., 1990, 1992). This system can also be used for cell ablation studies that investigate anther cell specification processes because it overcomes the need for cell-specific promotersto ablate aspecific celltype. That is, acombination of chimeric barnase and barstar genes containing promoters with overlapping cell specificities can be used to selectively ablate single cell types (Table 2; T.P. Beals and R.B. Goldberg, unpublished results). Recently, other strategies have been developed for the productionof male-sterile plants. These include (1) the useof achimeric tapetal-specific glucanase geneto prematurelydisruptmicrosporedevelopment(Worrallet al., 1992),(2) antisense inhibition of flavonoid biosynthesis within tapetal cells to disrupt pollen development (van der Meer et al., 1992), (3) tapetal-specific expression of achimeric Agmbacferium rhizogenesm/Bgene that increasesauxin activity inthe anther and causes developmental abnormalities (Spenaet al., 1992),and (4)overexpression of a chimeric cauliflower mosaic virus 35s atp9 mitochondrial gene that disrupts anther development (Hernould et al., 1993). Clearly, knowledge gained by studying anther development at the molecular and genetic levels will reveal other approaches that can be used to genetically engineer nove1varieties of hybrid crop plants. CONCWSION The anther represents anoutstanding system to study the differentiation of plant cell types. Excellent progress has been made in describing many of the gene expression events that occur during anther development as well as genes controllingstamen primordia specification. However, much remains to be learned about the mechanisms that are responsible for anther cell specification andthose requiredfor the activation of cell-specific gene sets within specialized anther cells and tissues. Molecular and genetic tools are now available that should reveal new insight into the processes controlling anther development in the near future. ACKNOWLEDGMENTS We expressour appreciationtoAnna Koltunow,Titti Mariani,andJan Leemans for many stimulating discussions on anther development. We thank Becky Chasan for reviewing a draft of this manuscript, Margaret Kowalczykfor doingthe art work, and BonniePhanfor typing this manuscript. Research from our laboratory described in this paperwas supported by aNationalScienceFoundationgrantto R.B.G. T.P.B. was supported by a National lnstitutesof Health Postdoctoral Fellowshipand P.M.S. was supported by a New ZealandCrown Research lnstitute Predoctoral Fellowship. REFERENCES Aarts, M.G.M., Dirkse, W.G., Stlekema, W.J., andPerelra, A. (1993). Transposontagging of a male sterilitygene inArabidopsis. Nature Angenent, G.C., Busscher, M., Franken, J., MOI,J.N.M., and van Tunen, A.J. (1992).Differentialexpressionof two MADSboxgenes in wild-type and mutant petunia flowers. Plant Cell 4, 983-993. Bonner, L.L., and Dlckenson, H.G. (1989). Anther dehiscence in Lycopersiconesculentum. I. Structuralaspects. New Phytol. 113, Brewbaker,J.L. (1964). Agricultura1Genetics(Englewood Cliffs,NJ: Prentice-Hall). Chaudhury,A.M. (1993). Nucleargenescontrollingmalefertility. Plant Cell 5, 1277-1283. Chaudhury, A., Cralg, S., Bloemer, K.C., Farrell, L., and Dennls, E.S. (1992). Genetic control of malefertility in higher plants.Aust. J. Plant Physiol. 19, 419-426. Coen, E.S. (1991).The roleof homeotic genesinflowerdevelopment andevolution.Annu. Rev.PlantPhysiol.PlantMOI.Biol.42,241-279. Coen, E.S. (1992). Flower development. Curr. Opin. Cell Biol. 4, Coen, E.S., andCarpenter, R. (1993).The metamorphosisof flowers. Plant Cell 5, 1175-1181. Coen, E.S., and Meyerowitz, E.M. (1991). The war of the whorls: Genetic interactions controllingflower development. Nature 353, Coen, E.S., Romero, J.M., Doyle, S., Elllott, R., Murphy, G., and Carpenter,R. (1990).floricaule:A homeoticgene requiredfor flower development in Antirrhinum majus. Cell 63, 1311-1322. Davldson, E.H. (1991). Spatial mechanisms of gene regulation in metazoan embryos. Development 113, 1-26. Dawe, R.K., and Freellng,M. (1992).The roleof initialcellsin maize anther morphogenesis.Development 116, 1077-1083. de1Camplllo, E., andLewls,L.N. (1992).Occurrenceofthe9.5cellulaseandotherhydrolasesinflower reproductiveorgansundergoing major cell wall disruption. Plant Physiol. 99, 1015-1020. Denls, M., Delourme, R., Gourret, J.P., Marlanl, C., and Renard, M. (1993). Expressionof engineerednuclearmale sterilityinErassim napus. Plant Physiol. 101, 1295-1304. Drews, G.N., Beals, T.P., Bul, A.Q., and Goldberg, R.B. (1992). Regional and cell-specific gene expression patterns during peta1 development. Plant Cell 4, 1383-1404. 363, 715-717. 97-1 15. 929-933. 31-37. 1228 The Plant Cell Esau, K. (1977). Anatomy of Seed Plants (New York: John Wiley). Evrard, J.L., Jako, C., Saint-Guily, A., Weil, J.H., and Kuntz, M. (1991). Anther-specific, developmentally regulated expression of genesencodinganewclassof proline-richproteins.PlantMOI.Biol. Feistritzer, W.R., and Kelly, A.F. (1987).HybridSeed Productionof SelectedCerealOil andVegetableCrops(Rome:FoodandAgriculture Organization of the United Nations). Feldmann, K.A. (1991).T-DNA insertionmutagenesis inArabidopsis: Mutational spectrum. Plant J. 1, 71-82. Forsthoefel,N.R., Wu, Y., Schulz,B., Bennett,M.J., andFeldmann, K.A. (1992).T-DNAinsertionmutagenesisinArabidopsis:Prospects and perspectives.Aust. J. Plant Physiol. 19, 353-366. Gasser, C.S. (1991). Molecular studies on the differentiationof floral organs. Annu. Rev. Plant Physiol. Plant MOI.Biol. 42, 621-649. Hansen, G., Estruch,J.J., Sommer, H., andSpena, A. (1993).NTGLO: A tobacco homologof GLOBOSA floral homeotic gene of Antirrhinum majus: cDNA sequence and expression pattern. MOI.Gen. Genet. 239, 310-312. Hartley, R.W. (1989).Barnaseandbarstar:Two small proteinsto fold and fit together. Trends Biochem. Sci. 14, 450-454. Hernould, M., Suharsono, S., Litvak, S., Araya, A., and Mouras, A. (1993). Male-sterilityinductionintransgenic tobaccoplantswith anuneditedatp9mitochondrialgenefromwheat. Proc. Natl.Acad. Sci. USA 90, 2370-2374. Hicks, G.S., and Sussex, I.M. (1971). Organ regeneration in sterile culture after median bisection of the flower primordiaof Nicotiana tabacum. Bot. Gaz. 132, 350-363. Hicks, G.S., Bell, J., andSand, S.A. (1977).A developmentalstudy of the stamens in a male-steriletobacco hybrid. Can. J. Bot. 55, Horner, H.T., and Wagner, B.L. (1992).Association of four different calciumcrystals in the anther connective tissue and hypodermal stomiumof Capsicummnuum (Solanaceae)duringmicrosporogenesis. Am. J. Bot. 79, 531-541. Horvitz, H.R., and Sternberg, P.W. (1991). Multipleintercellularsig- nalingsystemscontrolthedevelopmentoftheCaenorhabditiselegans vulva. Nature 351, 535-541. Huala, E., and Sussex, I.M. (1993). Determinationand cell interactions in reproductive meristems. Plant Cell 5, 1157-1165. Kamalay, J.C., and Goldberg, R.B. (1980). Regulation of structural gene expression in tobacco. Cell 19, 935-946. Kamalay, J.C., and Goldberg, R.B. (1984). Organ-specific nuclear RNAs in tobacco. Proc. Natl. Acad. Sci. USA 81, 2801-2805. Kaul, M.L.H. (1988). MaleSterility inHigher Plants(Berlin: Springer- Verlag). Keijzer,C.J. (1987).Theprocessof antherdehiscenceandpollendis- persal.I.Theopeningmechanismoflongitudinallydehiscinganthers. New Phytol. 105, 487-498. Klein, R.M. (1987). The Green World. An lntroductionto Plantsand People (New York: Harper and Row). Kofer, W., Glimellus, K., and Bonnett, H.T. (1990).Modificationsof floral developmentin tobacco inducedby fusion of protoplastsof different male-sterilecultivars. Theor. Appi. Genet. 79, 97-102. Kofer, W., Glimellus, K., and Bonnett, H.T. (1991). Modificationsof mitochondrialDNAcausechangesinfloraldevelopmentinhomeoticlike mutantsof tobacco. Plant Cell 3, 759-769. 16, 271-281. 2234-2244. Kofer, W., Glimelius, K., and Bonnett, H.T. (1992). Fusionof malesteriletobaccocausesmodificationsof mtDNAleadingto changes infloralmorphoicgyandrestorationoffertilityinhybridplants.Physiol. Rlant.85, 334-338. Koitunow,A.M., Truettner,J., Cox, K.H., Wallroth, M., andGoidberg, R.B. (1990).Differenttemporalandspatiaigeneexpressionpatterns occur during anther development. Plant Cell 2, 1201-1224. Koornneef, M., Dellaert,L.W.M., andvanderVeen, J.H. (1982).EMS and radiation-induced mutation frequencies at individual ioci in Arabidopsis thaliana. Mut. Res. 93, 109-123. Kriimer, H., Cagan, R.L., andZipursky, L. (1991).lnteractionof bride of sevenlessmembrane-boundligandandthe sevenlesstyrosinekinase receptor. Nature 352, 207-212. Mandel, M.A., Gustafson-Bmwn, C., Savidge, B., and Yanofsky, M.F. (1992). Moiecular characterization of the Arabidopsis floral homeotic gene APETALA7. Nature 360, 273-277. Mariani, C., De Beuckeleer, M., Truettner, J., Leemans, J., and Goldberg, R.B. (1990). lnduction of male sterility in plants by a chimaeric ribonuclease gene. Nature 347, 737-741. Mariani,C., Gossele, V., DeBeuckeleer,M., DeBlock, M., Goldberg, R.B., De Greef, W., and Leemans, J. (1992). A chimaeric ribonuclease-inhibitor gene restoresfertilityto male sterile plants. Nature 357, 384-387. Mascarenhas, J.P. (1993).Molecularmechanismsof pollentubegrowth and differentiation. Plant Celi 5, 1303-1314. McCormick, S. (1991). Molecularanalysis of malegametogenesisin plants.Trends Genet. 7, 298-303. McCormick, S. (1993). Male gametophytedevelopment. Plant Cell Nacken,W.K.F., Huijser,P., BelWn, J.P., Saedler, H., andSommer, H. (1991).Molecularcharacterizationof twostamen-specificgenes, tap 7 andfil7, that are expressedinwildtype, but notthe deficiens mutant of Anthirrhinum majus. MOI.Gen. Genet. 229, 129-136. Okamuro, J.K., denBoer, B.G.W., andJofuku, K.D. (1993).Regulation of Arabidopsis flower development. Plant Cell 5, 1183-1193. Pacini,E., Franchi,G.G., andHesse, M. (1985).Thetapetum:itsform, function,andpossiblephylogenyinEmbryophyta.PlantSyst.Evol. Palmlter, R.D., Behringer,R.R., Quaife, C.J., Maxwell, F., Maxwell, I.H., and Brinster, R.L. (1987). Celliineageablation intransgenic micebycell-specificexpressionof atoxin gene.Celi 50, 435-443. Pneuli, L., Abu-Abeid, M., Zamir, D., Nacken, W., Schwarz-Sommer, Z., andLlfschitz, E. (1991).The MADSboxgenefamily intomato: Temporalexpressionduringfloraldevelopment,conservedsecondary structuresandhomologywith homeotic genesfromAntirrhinum and Arabidopsis. Plant J. 1, 255-266. Preuss,LI., Lemieux, E.,Yen, G., andDavls, R.W. (1993).Aconditional sterile mutation eliminates surfacecomponentsfrom Arabidopsis pollenand disrupts cell signalingduring fertilization.Genes Dev. Rosenberg,S.M., andBonnett, H.T. (1983).Floralorganogenesisin Nicotianatabacum: A comparisonof two cytoplasmic male-sterile cultivars with a male-fertilecultivar. Am. J. Bot. 70, 266-275. Sand, S.A. (1968).Geneticmodificationof cytoplasmic malesterility in tobacco. J. Hered. 59, 175-177. Satina, S., and Blakeslee,A.F. (1941).PericlinalchimerasinDafura stramoniuminrelationto developmentof leafandflower.Am.J. Bot. 5, 1265-1275. 149, 155-185. 7, 974-985. 28, 862-871. Anther Development 1229 Satina, S., Blakeslee, A.F., and Avery, A.G. (1940). Demonstration of three germ layers in the shoot apex of Datura by means of induced polyploidy in periclinal chimeras.Am. J. Bot.27, 895-905. Schmidt, R.J., Veit, B., Mandel, M.A., Mena, M., Hake, S., and Yanofsky, M.F. (1993). ldentificationand molecular characterizationof ZAG1,the maizehomologof theArabidopsisfloralhomeotic gene AGAMOUS. Plant Cell 5, 729-737. Schmiilling, T., Rohrig, H., Pilz, S., Walden, R., andSchell, J. (1993). Restorationof fertilitybyantisense RNAingenetically engineered male sterile tobacco plants. MOI.Gen. Genet. 237, 385-394. Schwarz-Sommer, Z.,Huijser, P., Nacken, W., Saedler, H., and Sommer, H. (1990). Genetic control of flower development by homeotic genes in Antirrhinum majus. Science 250, 931-936. Schwarz-Sommer, Z.,Hue, I., Huijser, P., Flor, P.J., Hansen, R., Tetens, F., Wnnig, W.E., Saedler,H., andSommer, H.(1992).Characterization of the Antirrhinum floral homeotic MADS-box gene deficiens: Evidencefor DNAbindingandautoregulationof its persistent expression throughout flower development. EMBO J. 11, Scott, R., Hodge, R., Paul, W., andDraper,J. (1991).The molecular biology of anther differentiation. Plant Sci. 80, 167-191. Shen, J.B., andHsu, F.C. (1992).Bfassicaanther-specificgenes:Characterizationandinsitulocalizationof expression.MOI.Gen.Genet. Smith, A.G., Gasser,C.S., Budelier,K., andFraley,R.T. (1990).Identificationandcharacterizationof stamen-andtapetumspecificgenes from tomato. MOI.Gen. Genet. 222, 9-16. Spena, A., Estruch, J.J., Prensen, E., Nacken,W., Van Onckelen, H., andSommer, H. (1992).Anther-specificexpressionof the rol6 geneofAgrobacferiumrhizogenesincreasesIAAcontentinanthers andaltersantherdevelopment inwholeflowergrowth.Theor.Appl. Genet. 84, 520-527. 251-263. 234, 379-389. Sulston, J.E., Schierenberg, E., White, J.G., and Thomson, J.N. (1983).Theembryoniccelllineageof CaenorhabditiselegansDev. Biol. 78, 577-597. Sussex, I.M. (1989). Developmental programming of the shoot meristem. Cell 56, 225-229. Szymkowiak, E.J., and Sussex, I.M. (1992). The interna1meristem layer (L3) determines floral meristemsize and carpel number in tomato periclinal chimeras. Plant Cell 4, 1089-1100. Trtibner, W., Ramirez,L., Motte, P., Hue,I.,Huijser,P., Wnnig, W.E., Saedler, H., Sommer, H., and Schwarz-Sommer, 2. (1992). GLOBOSA:Ahomeoticgenewhich interactswith DEFlClENSinthe controlofAntirrhinumfloralorganogenesis.EMBOJ. 11,4693-4704. Trull, M.C., Holoway, B.L., Friedman, W.E., and Malmberg, R.L. (1991).Developmentallyregulatedantigenassociatedwith calcium crystals in tobacco anthers. Planta 186, 13-16. van der Krol, A.R., and Chua, N-H. (1993). Flowerdevelopment in petunia. Plant Cell 5, 1195-1203. vander Meer, I.M., Stam, M.E., van Tunen, A.J., MOI,J.N.M., and Stuitje, A.R. (1992). Antisense inhibitionof flavonoid biosynthesis in petunia anthers results in male sterility. Plant Cell 4, 253-262. Veit, B., Schmidt, R.J., Hake, S., andYanofsky, M.F. (1993). Maize floral development: New genes and old mutants. Plant Cell 5, Weberling, F. (1989).Morphologyof FlowersandInflorescences.(Cambridge: Cambridge UniversityPress). Weigel, D., Alvarez,J., Smyth,D.R., Yanofsky, M.F.,andMeyemwitz, E.M. (1992).LEAFYcontrolsfloralmeristemidentityinArabidopsis. Cell 69, 843-859. Worrall, D., Hlrd, D.L., Hodge, R., Paul, W., Draper, J., and Scott, R. (1992).Prematuredissolutionof the microsporocytecallosewall causesmalesterilityintransgenic tobacco. PlantCell4, 759-771. 1205-1215.