The Plant Cell, Vol. 2, 741-753, August 1990 O 1990 American Society of Plant Physiologists Function of the apetala-1 Gene during Arabídopsis Floral Development Vivian F. Irish’ and lan M. Sussex Department of Biology, Osborn Memorial Laboratories, Yale University, New Haven, Connecticut 0651 1 We have characterized the floral phenotypes produced by the recessive homeotic apetala 1-1 (apl-1) mutation in Arabidopsis. Plants homozygous for this mutation display a homeotic conversion of sepals into bracts and the concomitant formation of floral buds in the axil of each transformed sepal. In addition, these flowers lack petals. We show that the loss of petal phenotype is due to the failure of petal primordia to be initiated. We have also constructed double mutant combinations with apl and other mutations affecting floral development. Based on these results, we suggest that the AP 1 and the apetala 2 (AP2) genes may encode similar functions that are required to define the pattern of where floral organs arise, as well as for determinate development of the floral meristem. We propose that the AP 1 and AP2 gene products act in concert with the product of the agamous (AG) locus to establish a determinate floral meristem, whereas other homeotic gene products are required for cells to differentiate correctly according to their position. These results extend the proposed role of the homeotic genes in floral development and suggest new models for the establishment of floral pattern. INTRODUCTION Vegetative meristems form a reiterated series of vegetative leaves with associated axillary buds and have the capacity for indefinite growth (McDaniel, 1980). ln contrast, floral meristems are strictly determinate and form a species- specific pattern of floral organs. In response to various cues, cells in the vegetative shoot apical meristem can initiate floral development. Once the floral signal is per- ceived, florally determined apical meristems are competent to carry out a floral program of development, even when isolated from the rest of the plant (Hicks and Sussex, 1970; Singer and McDaniel, 1986). These results indicate that after floral evocation floral development of the apical meristem is independent of the rest of the plant. Therefore, the development and differentiation of floral organs depend on localized interactions within the apical meristem. The biochemical nature of the substances involved in these interactions is unknown, but certain conclusions regarding their action have been inferred from a variety of surgical manipulations of flower primordia. Bisections of young floral buds result in the formation of two nearly complete flowers, indicating that cells within the developing meris- tem can reassess their position and differentiate accord- ingly (Cusick, 1956; Hicks and Sussex, 1971). These ex- periments suggest that cells within the apical meristem differentiate in response to a combination of as yet un- characterized signals, which results in the organization of floral organs in a stereotypic pattern. To define the gene products involved in specifying where and how floral organs develop, we are studying mutations that specifically alter floral development. A number of such mutations have been isolated in Arabidopsis thaliana (Koornneef et al., 1982; Haughn and Somerville, 1988; Komaki et al., 1988; Bowman et al., 19; Kunst et al., 19). These include a number of homeotic floral muta- tions in which one organ type is replaced by another that normally develops in a different location. Extensive anal- yses of homeotic mutations in Drosophila have provided a genetic and molecular framework for understanding how cells correctly differentiate according to their position (Lewis, 1978; Akam, 1987; lngham, 1988). The discovery of the homeobox, a DNA-binding motif present in many of the Drosophila homeotic genes, has provided the basis for models in which these genes function to regulate transcrip- tion (Laughon and Scott, 1984; McGinnis et al., 1984). The conservation of similar homeobox sequences in vertebrate genes suggests that a common mechanism of transcrip- tional control is an integral part of positional specification in animal systems (Shepherd et al., 1984; McGinnis, 1985). The identification of a different DNA-binding motif in both the homeotic deficiens A gene of Antirrhinum and the homeotic agamous gene of Arabidopsis indicates that tran- scriptional control may be involved in positional specifica- tion in plants as well (Sommer et al., 1990; E. Meyerowitz, personal communication). An analysis of floral-specific genes in Arabidopsis, where both genetic and molecular studies are feasible (Meyerowitz and Pruitt, 1985), will ’ To whom correspondence should be addressed. 742 The Plant Cell facilitate a better understanding of how organ position and identity are specified during floral development. We have characterized the phenotype of a homeotic floral mutation in Arabidopsis that appears to affect some of the earliest stages of floral development. This mutation, apefala7-7 (ap7-7), is recessive and maps to a single locus on chromosome 1 (Koornneef et al., 1983). We present a detailed description of the mutant phenotype and of the development of mutant flowers. We also describe double mutant studies in which we combined ap7-7 with other mutations affecting floral development. These genetic anal- yses suggest that the AP7 wild-type gene product is required to specify the location at which floral organ pri- mordia will develop, as well as for the appropriate differ- entiation of specific cell types. RESULTS Wild-Type Morphology Wild-type plants (ecotype Landsberg erecfa) germinate in 3 to 5 days and produce a rosette of about eight to 10 leaves. At about 16 days post-germination, the conversion from vegetative growth to flowering becomes apparent with the appearance of the rapidly bolting floral meristem. Anthesis first occurs at about 21 days. By 28 days, the mature plant is in full flower and secondary floral branches appear: usually three or four axillary floral stems branch off from the main stem, each subtended by a cauline leaf, or bract, as illustrated in Figure 1. On each axillary floral branch two secondary cauline leaves arise at transverse positions relative to the primary cauline leaf. Later in de- velopment inflorescences develop in the axils of these cauline leaves and other inflorescences can arise from the basal rosette. The Arabidopsis inflorescence is a raceme, and flowers arise in a helix around the axis. Each flower arises directly from the axis without a subtending bract. Flowers develop acropetally, with older flowers near the base of the inflo- rescence and progressively younger flowers toward the apex. Arabidopsis flower morphology has been extensively described and is similar to that of other crucifers (Vaughan, 1955; Polowick and Sawhney, 1986; Bowman et al., 19; Hill and Lord, 19). The wild-type flower consists of four whorls of morphologically distinct organs, as shown in Figures 2A and 2B. We would like to emphasize that we use “whorl” to indicate a position in the flower rather than the organ that develops at that location. In wild-type flow- ers, the first, or outer, whorl is composed of four sepals that alternate with the second whorl of four petals. The third whorl consists of six stamens: two short outer sta- mens on the transverse plane and four long inner stamens on the median plane. The central pistil has two carpels that compose the fourth whorl. The apetala 7 Mutant Phenotype The apefala 7 (AP7) locus maps at 103.5 centimorgans on chromosome 1 (Koornneef et al., 1983). A number of recessive mutations at the AP7 locus, all with apparently similar phenotypes, have been isolated and briefly char- acterized (Koornneef et al., 1982). Here we describe the phenotype of plants homozygous for the ap7-7 mutation. The ap7-7 mutation affects the development of flowers; we do not observe any disruption of normal vegetative growth or in the formation of floral axillary branches (data not shown). ap7-7 flowers show a transformation of the first-whorl organs into bract-like structures based on both epidermal morphology and the development of flower buds in the axil of each bract-like organ. Figures 2C and 2D show that this pattern of development is reiterated in the axillary second- ary flowers so that tertiary buds can be formed in the axils of each secondary bract-like organ. Secondary and tertiary flowers can be incomplete or irregular, with mosaic organs or with fewer organs than normal. The secondary flowers are oriented with respect to the primary flower axis. This is most clearly seen in the orientation of the septum of each secondary pistil, which lies on a median plane with respect to the primary flower axis (Figure 2D). In wild-type plants, floral internodes are compressed, whereas in ap7-7 flowers, the internodes between the media1 and the lateral first-whorl organs elongate. In general, the lateral first- Figure 1. Diagram of Wild-Type Arabidopsis Morphology. Open circles represent floral rneristems, and closed circles are inflorescence meristems. The Arabidopsis apetala-1 Gene 743Figure 2. Morphology of Wild-Type and ap1-1 Mutant Flowers.(A) Wild type. Bar = 100 ^m.(B) Diagram of wild type. The adaxial sepal is adjacent to the inflorescence axis (indicated by small circle).(C) Homozygous ap1-1 mutant flower. Bar = 200 ^m.(D) Diagram of ap1-1 flower.whorl organs appear opposite each other and the abaxialand adaxial bract-like organs are located more apically.The transformation of first-whorl organs into more leaf-like bracts can be seen at the cellular level in Figure 3.Wild-type sepals have characteristic elongated epidermalcells and a few simple trichomes on their abaxial surface.Wild-type cauline leaves, on the other hand, have irregu-larly shaped epidermal cells and stellate trichomes on theirabaxial surface. In ap1-1 plants, the epidermal cells of thefirst-whorl organs appear similar to those of cauline leavesand do not have any of the epidermal features associatedwith sepal tissue. This bract-like epidermal phenotype isalso characteristic for the first-whorl organs of the second-ary ap1-1 flowers.In addition to the aberrant development of first-whorlorgans as bract-like structures, ap1-1 mutants also lackpetals. Very rarely in ap1-1 mutants some petal epidermalcells differentiate as part of a mosaic organ. Occasionally,on secondary and tertiary flower buds, organs developthat are mosaics of leaf-like and stamen-like tissue; theseorgans appear to arise from a region in the meristem wherethe first- and third-whorl primordia are closely apposed.744 The Plant CellFigure 3. Epidermal Cell Phenotypes of Wild-Type and apl-1Tissues.(A) Wild-type sepal tissue.(B) Wild-type cauline leaf tissue.(C)ap7-7 first-whorl organ tissue.Bar = 15 /urn.Rarely, we observe mosaics of leaf-like and petal-liketissue, as illustrated in Figure 4. In all cases, the sectorsof leaf, stamen, or petal tissue are large, contiguouspatches running longitudinally from the base to the apexof the organ. Although the sector boundary is not alwaysstraight, we see a relatively discrete transition betweencell types, with only one or a few cells of intermediatephenotype spanning the boundary (Figure 4C).Table 1 shows that the development of first-whorl or-gans as bract-like structures with associated axillary flowerbuds is most consistent and complete in the more basalflowers of the inflorescence. Figures 5A and 5B illustratethat the first-whorl organs on more apical flowers can beaborted or can develop into small structures that havesome of the epidermal features of cauline leaves. None-theless, secondary flowers can develop in the axils ofaborted organs. The most common pattern of develop-ment of the more apical flowers is for one or both of thelateral first-whorl organs and their associated axillary budsto develop, whereas abaxial and adaxial first-whorl organsand their associated axillary buds tend to abort (Table 1;Figure 50). Whereas the extent of axillary bud formationvaries with the position of the primary flower on the inflo-rescence, the petal phenotype is constant throughout theplant. That is, we never see petals formed, and the occur-rence of the rare mosaic organs containing petal tissueappears to be random with respect to the position of theflower along the inflorescence.Development of Wild-Type and ap1 Mutant FlowersThe ontogeny of wild-type flowers has been describedboth morphologically and histologically (Vaughan, 1955;Bowman et al., 19; Hill and Lord, 19; Kunst et al.,19). We will briefly summarize the stages of wild-typefloral development. Figure 6A illustrates how wild-typeflower primordia arise in a helical sequence as outgrowthsof cells on the flank of the apical meristem. In the youngflower primordium, the abaxial sepal is initiated first,quickly followed by the development of a ridge of cellsdestined to become the adaxial sepal. The lateral sepalsare then initiated and begin to grow out; the whorls ofpetal and stamen primordia are also apparent at this stage(Figure 6B). The sepals grow to enclose the developinginner primordia and begin to differentiate the elongatedcells and stomata characteristic of the mature sepal epi-dermis. Petal development is retarded while stamen pri-mordia continue to grow. At this time, the remaining tissueof the meristem elongates and develops a cleft in thecenter (Figure 6C). This central tube will give rise to thepistil. As the stamens grow, filament and anther primordiadifferentiate; anthers develop locules prior to the expan-sion of the filaments (Figure 6D). By this stage, petalprimordia begin to elongate, and the beginning of differ-entiation of the stigmatic papillae on the pistil becomesFigure 4. ap1-1 Mosaic Organs.(A) Mosaic of stamen and bract tissue. Locules and cellularmorphology characteristic of stamens is toward the top; bract-like epidermal cells appear in the lower part of this mosaic organ.Tip of the organ is to the right. Bar = 15 Mm.(B) Mosaic of bract and petal tissue. Epidermal cells with char-acteristic petal cell morphology occupy approximately the top halfof this organ. Tip of the organ is to the left. Bar = 75 Mm.(C) Closeup of organ pictured in (B). The mosaic boundary isindicated with a dotted line. Bar = 10 Mm.The Arabidopsis apetala-1 Gene 745apparent. At this point, nectaries at the base of the sta-mens also develop. The petals then rapidly expand toreach the tip of the pistil. Finally, the filaments of thestamens elongate as the pollen matures.The apical meristem of ap1-1 plants appears similar tothe wild type in size and shape (Figure 6E). However,deviations become apparent as soon as individual ap1-1flower primordia begin to develop. First-whorl organs ap-pear in an order similar to that of wild-type sepals but donot grow to enclose the developing bud. These primordiaare more rounded and grow away from the central floralmeristem (Figure 6F). As these first-whorl organs mature,stipules are often apparent at the base (Figure 6G). Sti-pules are characteristic of leaves and normally do notappear on sepals. Primordia do not appear in the secondwhorl; it appears that the cells that reside in the spacesalternate with the first-whorl primordia are eventually re-cruited into the growing bract-like first-whorl organs. Sta-men and carpel primordia develop normally. Small out-growths that appear to be nectaries are variably presentat the base of the stamens. As the stamen primordia grow,development of buds in the axils of the first-whorl organsbecomes apparent. These buds develop in a manner sim-ilar to that of the primary flowers, but because they areinitiated later, they appear younger than the primary flower(Figure 6H).Double Mutant StudiesWe have generated plants that are homozygous for bothap1-1 and other recessive mutations affecting floral devel-opment. These mutations either affect carpel number orresult in homeotic transformations of different floral organs.Our observations are based on examination of the basalflowers formed on the inflorescence of each double mutantcombination because the transformation elicited by theapt-1 mutation is most complete in these flowers. We findTable 1. Number of First-Whorl Organs and Axillary Buds thatDevelop in apefa/a 1 Mutant Flowers*Abaxial Adaxial LateralFirst-whorl organ developmentBasal flowers\"0 (n = 22)17625Apical flowers (n = 1 2)6012First-whorl axillary bud developmentBasal flowers'30 (n = 22)201941Apical flowers (n = 1 2)2015\"The development and location (abaxial, adaxial, and the twolateral positions) of bract-like first-whorl organs and axillary budswere scored in both basal and apical flowers.bc Flowers 1 to 4 on the main inflorescence. Flowers 10 to 20 on the main inflorescence.746 The Plant CellFigure 5. ap7-7 Aborted and Irregular Organs.(A) Reduced first-whorl organ that has differentiated a long, hair-like structure at its tip. Bar =15 ^m.(B) Aborted first-whorl organ with an undeveloped axillary bud.Arrow indicates a mature stomate. Bar = 2 ^m.(C) Flower from the more apical part of an ap1-1 inflorescence.Only one secondary flower has developed. Bar =100 ^m.both epistatic and additive interactions, and in the case ofap1 ap2 double mutants, we see a synergistic effect.apt clvlMutations at the clavata 1 locus (CLVT, Koornneef et al.,1983; Okada et al., 19) alter the carpel number fromtwo to four, resulting in a fat, club-shaped pistil, as shownin Figure 7A. We have combined ap1-1 with the clv1-1allele, and the resulting flowers display a nearly additivephenotype. As in ap1-1 alone, petals are missing andbracts with axillary flowers are formed, but both the pri-mary and secondary flowers have four carpels instead oftwo (Figures 7B and 7C). However, the loss of the CLV1function appears to result in a greater tendency towardcarpelloidy of organs in other whorls of the flower. Forinstance, mosaic carpelloid bracts are occasionally ob-served in secondary ap1-1 clv1-1 flowers (Figure 7C). Thistype of mosaic organ has not been observed in plantshomozygous for ap 1 -1.ap1 pi and ap1 ap3Figure 8 illustrates that mutations at both the pistillata (PI;Koornneef et al., 1983) and the apetala 3 (APS; Bowmanet al., 19) loci have similar phenotypes. Plants homozy-gous for ap3-1 display a transformation of second-whorlorgans into structures with the epidermal characteristicsof sepals, whereas third-whorl organs differentiate as car-pels that can fuse to the central pistil (Bowman et al.,19). Homozygous pi-1 flowers also have two outeralternate whorls of sepal-like organs surrounding a centralenlarged pistil. The loss of stamens has been attributed tothe lack of development of the third whorl (Bowman et al.,19). Closer histological and developmental examinationindicates, however, that primordia do arise in the region ofthe meristem that normally gives rise to the stamens butdifferentiate into carpelloid organs that fuse with the centralpistil (Hill and Lord, 19). Therefore, the phenotypic ef-fects of the ap3-1 or pi-1 mutations appear to be nearlyidentical.ap1-1 in combination with either ap3-1 or pi-1 results inflowers with very similar phenotypes (Figure 8). In theprimary flower, four bracts with their associated axillarybuds are formed, but organs derived from the secondwhorl do not appear. In both ap1-1 ap3-1 and ap1-1 pi-1flowers, the pistil is enlarged and appears to result fromthe fusion of third- and fourth-whorl organs. This patternof development is also seen in the secondary flowers ofboth double mutant combinations. In terms of the second-whorl organs, then, ap1 appears to be epistatic to bothap3 and pi mutations because second-whorl organs donot develop.Jhe Arabidopsis apetala-1 Gene 747Figure 6. Development of Wild-Type and ap7-7 Mutant Flowers.(A) to (D) Wild-type development.(E) to (H) ap7-7 development.(A) Top view of a wild-type inflorescence. Developing flower buds 1 through 9 can be seen. Bar = 10 ^m.(B) Young wild-type bud. Sepals are beginning to close over the floral meristem, and the stamen (st) and petal (p) primordia are visible assmall outgrowths. Bar = 10 ^m.(C) and (D) Progressively later stages of development. Some sepal and petal tissue has been removed. Petals begin to elongate as thecentral gynoecium matures. Bar = 50 ^m in (C) and 100 ^m in (D).(E) Top view of an ap7-7 inflorescence. Developing flower buds 1 through 6 and 8 and 9 can be seen. Bar = 10 Mm.(F) A young ap7-7 bud of comparable age to the bud shown in (B). Bar =10 ^m.(G) A slightly later stage of development, where stamen and carpel primordia are apparent. First-whorl organs have stipules (sp). Bar =10/*m.(H) Axillary buds develop later than the primary flower but are apparent before the primary flower is mature. Bar = 50 ^m.ap1 ap2 and ap7 agBoth the first- and the second-whorl organs are variablyaffected by mutations in the apetala 2 (AP2) locus (Komakiet al., 1988; Bowman et al., 19; Kunst et al., 19). Anumber of ap2 alleles with different phenotypic effectshave been generated (Komaki et al., 1988; Bowman et al.,19; Kunst et al., 19). Not only do the phenotypes varyfrom allele to allele, but the ap2 mutant phenotype variesacropetally within a single homozygous plant. For somealleles, environmental effects such as daylength or tem-perature affect the severity of the resulting mutant phe-notype (Komaki et al., 1988; Bowman et al., 19). Wehave used the well-characterized ap2-7 allele (Bowman etal., 19) to generate the double mutant combination withap7-7.Plants homozygous for ap2-7 show an acropetal changein the mutant phenotype. More basal flowers display atransformation of first-whorl organs into leaf-like structureswith stigmatic papillae at their tips. Figure 9A shows thatthe second-whorl organs develop as white stamenoid pet-als with infoldings that appear to be rudimentary locules.Stamens and carpels develop normally. The degree offirst-whorl carpelloidy and second-whorl stamenoidy in-creases acropetally although the first whorl never showsa complete transformation to carpels.Both organ initiation and organ identity are disrupted inap7-7 ap2-7 double mutants. These plants are normal invegetative growth and upon bolting develop normal caulineleaves and associated axillary buds. However, floral mer-istems that would normally give rise to a flower differentiatein an indeterminate pattern characteristic of an infloresc-ence (Figures 9B, 9C, and 9D). Lateral structures arise offthis transformed meristem in a helical phyllotaxy charac-teristic of Arabidopsis inflorescences, not in the cruciformpattern of an Arabidopsis flower. These lateral structuresdifferentiate as a normal bicarpellate pistil associated withone or a few stamens, or more often as a fusion of severalcarpelloid structures that may or may not have associatedstamens. Up to 30 lateral structures can differentiate onone transformed floral meristem. The most basal lateralstructures of an ap1-1 ap2-1 flower can themselves de-velop as an inflorescence, occasionally subtended by aleaf-like organ (Figure 9D). As the flower matures, inter-748 The Plant Cell Figure 7. clv7-7 and ap7-7 clv7-7 Floral Phenotypes. (A) clv7-7 flower. (B) apl-7 clvl-7 flower. (C) Mosaic ap7-7 clv7-7 organ composed of bract-like and carpel- like tissue. Stigmatic tissue at the tip of the organ and ovules at the margin are apparent. Bar = 100 pm. nodes elongate between lateral structures in a pattern characteristic of an inflorescence. Because an ap7-7 ap2-7 floral meristem develops as an indeterminate inflorescence when neither apl nor ap2 flowers alone have a similar indeterminate phenotype, these two mutations interact synergistically . Mutations at the agamous (AG) locus (Koornneef et al., 1983) cause a homeotic transformation of third-whorl or- gans as well as affecting the determinate nature of floral development (Bowman et al., 19). In homozygous ag-7 flowers, sepals and petals develop normally, but the third whorl differentiates as six petals. Figure 1 OA shows that, instead of forming a terminal whorl of carpels, the ag-7 mutant repeats the abnormal floral program of develop- ment so that four sepals develop followed by two more whorls of petals. This indeterminate pattern of develop- ment can be reiterated three or four times. Figure 10B shows that plants doubly mutant for ap7-7 and ag-7 do not develop petals. Four bract-like first-whorl organs develop and have buds in their axils. Next, a variable number of green, leaf-like organs arise. The inde- terminate nature of the ag-7 phenotype is still apparent in the ap7-7 ag-7 double mutants. The pattern of organ development described above reiterates with the appear- ance of another whorl of bracts with axillary buds, followed by a number of leaf-like organs. The buds formed in the axils of the ap7-7 ag-7 bracts also repeat this indetermi- nate pattern of development. DISCUSSION We are studying the homeotic mutations of Arabidopsis thaliana in an effort to understand how the wild-type products of these genes interact to allow the cells of the meristem to realize their appropriate developmental fates. To this end, we have characterized the phenotype of the homeotic ap7-7 mutation and of double mutant combina- tions between ap7-7 and other mutations involved in floral development. These genetic analyses suggest roles for the AP7 wild-type gene product in the establishment of both organ position and organ identity. AP1 1s Required To Promote Sepal Differentiation and Suppress Axillary Bud lnitiation The first-whorl primordia in an ap7-7 flower develop as bracts, as characterized by their leaf-like epidermal mor- phology, the presence of stipules, and the formation of buds in their axils. In ap7-7 flowers, as well as in all the double mutant combinations generated, the axillary buds develop the same phenotype as that of the primary flOWer. This observation indicates that a new floral meristem is 748 The Plant CellFigure 7. clv1-1 and ap7-7 clvT-1 Floral Phenotypes.(A) clv1-1 flower.(B)ap7-7 clv1-1 flower.(C) Mosaic ap1-1 clv1-1 organ composed of bract-like and carpel-like tissue. Stigmatic tissue at the tip of the organ and ovules atthe margin are apparent.Bar = 100 Mm.nodes elongate between lateral structures in a patterncharacteristic of an inflorescence. Because an ap7-7 ap2-1floral meristem develops as an indeterminate inflorescencewhen neither apt nor ap2 flowers alone have a similarindeterminate phenotype, these two mutations interactsynergistically.Mutations at the agamous (AG) locus (Koornneef et al.,1983) cause a homeotic transformation of third-whorl or-gans as well as affecting the determinate nature of floraldevelopment (Bowman et al., 19). In homozygous ag-1flowers, sepals and petals develop normally, but the thirdwhorl differentiates as six petals. Figure 10A shows that,instead of forming a terminal whorl of carpels, the ag-7mutant repeats the abnormal floral program of develop-ment so that four sepals develop followed by two morewhorls of petals. This indeterminate pattern of develop-ment can be reiterated three or four times.Figure 10B shows that plants doubly mutant for ap7-7and ag-7 do not develop petals. Four bract-like first-whorlorgans develop and have buds in their axils. Next, avariable number of green, leaf-like organs arise. The inde-terminate nature of the ag-7 phenotype is still apparent inthe ap7-7 ag-7 double mutants. The pattern of organdevelopment described above reiterates with the appear-ance of another whorl of bracts with axillary buds, followedby a number of leaf-like organs. The buds formed in theaxils of the ap7-7 ag-7 bracts also repeat this indetermi-nate pattern of development.DISCUSSIONWe are studying the homeotic mutations of Arabidopsisthaliana in an effort to understand how the wild-typeproducts of these genes interact to allow the cells of themeristem to realize their appropriate developmental fates.To this end, we have characterized the phenotype of thehomeotic ap7-7 mutation and of double mutant combina-tions between ap7-7 and other mutations involved in floraldevelopment. These genetic analyses suggest roles forthe AP1 wild-type gene product in the establishment ofboth organ position and organ identity.AP1 Is Required To Promote Sepal Differentiation andSuppress Axillary Bud InitiationThe first-whorl primordia in an ap7-7 flower develop asbracts, as characterized by their leaf-like epidermal mor-phology, the presence of stipules, and the formation ofbuds in their axils. In ap7-7 flowers, as well as in all thedouble mutant combinations generated, the axillary budsdevelop the same phenotype as that of the primary flower.This observation indicates that a new floral meristem isThe Arabidopsis apetala-1 Gene 749Figure 8. pi-1, ap3-1, ap1-1 pi-1, and ap1-1 ap3-1 Floral Phenotypes.(A)p/-7 flower.(B)ap7-7 pi-1 flower.(C)ap3-7 flower.(D)ap1-1 ap3-1 flower.Bar = 200 jum.created in the axil with all the attributes of the primaryfloral meristem. In other words, whatever interactions arerequired to define the location and differentiation of floralorgan types in the primary meristem, these interactionsare reestablished within the axillary meristem. The forma-tion of axillary buds and the concomitant loss of petals arenot the result of a homeotic conversion of petal primordiainto buds because buds arise in the axils of the first-whorlorgans and not in the position of second-whorl primordia.A similar mutation in which axillary flowers develop withinthe primary flower has been described in another crucifer,Nasturtium officinale (Arber, 1931). In this case, petalsstill develop, demonstrating that in Nasturtium, petal dev-elopment and axillary bud formation are not exclusiveprocesses.AP1 Is Required for Second-Whorl DevelopmentIn homozygous ap1-1 flowers petal primordia do not arise.Consequently, petal development and differentiation can-not take place. However, there may be a requirement forAP1 activity for the development of petal tissue. In ap1-1ag-1 double mutant plants, phylloid organs arise in the750 The Plant CellFigure 9. ap2-1 and ap1-1 ap2-1 Floral Phenotypes.(A) ap2-1 flower from the base of the inflorescence. First-whorl organs have leaf-like trichomes and stigmatic tips; second-whorl organsare white but have some stamenoid characteristics. Bar = 200 ^m.(B) ap1-1 ap2-1 flower from the side with many lateral structures. Bar = 200 ^m.(C) Top view of an ap1-1 ap2-1 flower showing the helical phyllotaxy. Bar = 200 ^m.(D) Basal lateral structure with an inflorescence-like phenotype on an ap1-1 ap2-1 flower. Arrows indicate tertiary structures that alsoappear to have the morphological characteristics of inflorescences. Bar = 20 urn.position normally occupied by petals in the ag-1 mutantalone. These organs do not have any of the epidermalfeatures associated with petals. Thus, it is possible that inaddition to being required for second-whorl development,the/4P7 gene product is also required for the differentiationof petal tissue. We occasionally see petal-like epidermalcells in mosaic organs of ap1-1 plants. These mosaicpatches of petal tissue arise along the margins of the first-whorl organs, close to where second-whorl primordia nor-mally arise. This observation suggests that the ap1-1mutation has not disrupted the processes required forpetal cell differentiation and may only affect the formationof second-whorl primordia. One explanation for this obser-vation is that the ap1-1 allele may not be a null mutation,and, therefore, we could be seeing the phenotypic effectsof residual activity from the mutant allele.The failure of petal primordia to be initiated in an ap1-1mutant argues against any type of relay model in whichthe identity of each whorl of organs is determined by thepreceding whorl (Heslop-Harrison, 19; McHughen, 1980).Because stamens develop normally in the absence ofpetals, stamen development cannot be dependent uponpetal development. Other models in which the formationof a whorl of organs is dependent upon the biophysicalconstraints induced by the primordia of the precedingwhorl (Green, 1988) also do not seem to be valid.Figure 10. ag-1 and ap1-1 ag-1 Floral Phenotypes.(A)ag-7 flower.(B)ap1-1 ag-1 flower.Bar = 100 ^m.The occasional development of petal epidermal tissue inlarge, contiguous mosaic patches suggests two possibili-ties for how cells in a developing organ assess theirposition and differentiate accordingly. A cell could acquirea specific fate during the initiation of the organ primordiumand then in a heritable fashion pass this decision on to itsprogeny, resulting in a large clone of cells in the matureorgan. A similar model has been proposed to explain thegeneration of mosaic patches in ag-1 flowers (Bowman etal., 19). Alternatively, cells could acquire a particular fatein a nonautonomous manner late in the development of anorgan. Cells might normally read their position within adeveloping organ and then differentiate accordingly. In thecase of mosaic organs, cells might read an unstable orThe Arabidopsis apetala-1 Gene 751threshold signal and then differentiate to form a mosaicpattern. The large size of the mosaic patches would indi-cate that even though cells are individually reading theirposition, nonautonomous cell-cell interactions could serveto enhance the acquisition of one cell fate or another.Indirect support for such a model comes from studies onImpatiens flowers, in which it appears that cells are notcommitted to a particular fate until just before their finaldifferentiation (Battey and Lyndon, 1984).Genetic Interactions of apl-1 and ap2-1The apl-1 and ap2-1 mutations interact synergistically,such that doubly mutant flowers develop as indeterminateinflorescences. This indeterminate phenotype may becaused by a partial redundancy of the AP1 and AP2 wild-type functions because loss of either gene product aloneis insufficient to produce an indeterminate phenotype. Wedo not know what the API and AP2 genes encode so wecannot determine the nature of this functional similarity.These genes may encode similar products, each of whichis sufficient to induce determinate growth. Functional re-dundancy of similar gene products has been demonstratedin yeast, where mutations in different members of the /•as-related gene family individually are viable, but double mu-tant combinations result in lethality (Tatchell et al., 1984).Alternatively, the AP1 and AP2 gene products may notthemselves be homologous but may each participate infunctionally similar biochemical pathways. We should em-phasize that the products of the AP1 and AP2 loci are likelyto have unique functions as well because the spectrum ofphenotypes produced by mutations at each locus isdistinct.Our interpretation of the similarity of the AP1 and AP2functions rests on our observation of ap1-1 ap2-1 flowers.However, we do not know how much residual AP1 orAP2activity remains in the double mutant combination. Theap2-1 allele is only a partial loss of function mutationbecause more severe alleles exist (Komaki et al., 1988;Bowman et al., 19). Other AP2 alleles have been gen-erated whose phenotypes bear some of the features ofapl-1 mutant flowers. Under short-day conditions, plantshomozygous for either ap2-3 or ap2-4 occasionally pro-duce flowers in the axils of the leaf-like first-whorl organs(Komaki et al., 1988). This similarity in phenotype producedby mutations at two different loci supports our suggestionthat the API and AP2 gene products participate in acommon developmental pathway.Role of the Homeotic Genes in Floral DevelopmentThe phenotypes of the homeotic floral mutants indicatethat the sequential formation of whorls of organs does notappear to depend upon interactions with either floral or-752 The Plant Cell gans or primordia in the preceding whorl. Instead, it ap- pears that the position of a developing whorl is laid down independently of the formation of a particular organ type. The development of the floral pattern must, therefore, depend on two processes: first, position must be specified within the developing flower bud, and, second, cells must respond to this information and execute a specific genetic program leading to the differentiation of particular cell types. Models in which the formation of floral organs depends upon the establishment of concentric rings of organ-determining substances (Meyerowitz et al., 19) or in which cells read their position relative to the meristem and not to other whorls (Holder, 1979) are both based upon this premise. However, it is not clear how this posi- tional information is established. The AP1 and AP2 gene products appear to act in concert to promote both determinate growth and cruciform phyl- lotaxy. The AG gene product is also required for determi- nate development because the ag-1 mutation causes in- determinate growth of the flower. We propose that inter- actions between the AP1 and AP2 functions together with the AG gene product are involved in establishing a deter- minate floral field. The interactions observed in apl-1 ap3-1 and apl-1 pi-1 combinations suggest that the AP1 function is required before the action of AP3 and PI. Similarly, the AP2 function has been postulated to act earlier than AP3 and PI (Kunst et al., 19; Meyerowitz et al., 19). Based on this interpretation, we suggest that the AP3 and PI genes may be activated in certain regions of the floral meristem in response to positional information established by the action of the APl, AP2, and AG gene products. Our model of homeotic gene action in Arabidopsis differs from that of others (Bowman et al., 19; Kunst et al., 19; Meyerowitz et al., 19) in that we propose a specific role for the AP1, AP2, and AG gene products in establishing position within the developing flower. How- ever, we cannot distinguish whether these gene products are required only for the establishment of position or whether they also play a role in the cellular interpretation of those signals. Other gene products such as AP3 and Pl appear to be involved only in the interpretation step and may encode cell-autonomous functions required for the perception or tranduction of positional signals. METHODS Plants were grown at 22OC to 24\"C, under a 16-hr day/d-hr night regime, with a combination of fluorescent and incandescent light (175 pmol.m-' . sec-' at pot-top). Seeds were planted in a 9:3:1 vermicu1ite:soil:sand mix and watered daily. The mutant alleles described here are all in the Landsberg ecotype carrying the erecta mutation. Mutant alleles were a gift from Maarten Koorn- neef (Department of Genetics, Wageningen Agricultura1 University, The Netherlands). Crosses were performed manually with ap7-7 plants as the pollen donor. Because ag-7 plants are both male and female sterile, heterozygous ag-l/+ plants were used to construct the ap7-7 ag-7 double mutant line. In all cases, the resclting F1 plants were allowed to self, and double mutant plants were isolated from the F2 population. Morphological characteriza- tion of both single and double mutant flowers was based on light microscopic examination of at least 20 flowers and scanning electron microscopy of between five and 1 O flowers. For scanning electron microscopy, flowers or buds were fixed in FAA (3.7% formaldehyde, 50% ethanol, 5% acetic acid) for 30 minutes to 1 hr and dehydrated in a graded ethanol series. Dehydrated material was critical point dried in liquid C02. Individual flowers were mounted on scanning electron microscope stubs and dissected with glass needles. Specimens were sputter coated with gold- paladium in an SPI sputter coater. Specimens were examined in an ISI SS40 scanning electron microscope with an accelerating voltage of 5 kV. ACKNOWLEDGMENTS This work was supported by grant DCB-8807561 from the Na- tional Science Foundation to V.F.I. We thank members of the Sussex and Nelson labs for comments and suggestions, and lain Dawson for critical reading of the manuscript. Received April 20, 1990; revised May 22, 1990. REFERENCES Akam, M.E. (1 987). The molecular basis for metameric pattern in the Drosophila embryo. Development 101, 1-22. Arber, A. (1931). Studies in floral morphology. II. On some normal and abnormal crucifers: With a discussion on teratology and atavism. New Phytol. 30, 172-203. Battey, N.H., and Lyndon, R.F. (1 984). Changes in apical growth and phyllotaxis on flowering and reversion in lmpatiens balsa- mina L. Ann Bot. 54, 553-567. Bowman, J.L., Smyth, D.R., and Meyerowitz, E.M. (19). Genes directing flower development in Arabidopsis. Plant Cell 1, 37-52. Cusick, F. (1956). Studies of floral morphogenesis. I. Median bisections of floral primordia in Primula bulleyana Forrest. Trans. R. SOC. Edinburgh. 63, 153-166. Green, P.B. (1988). A theory for inflorescence development and flower formation based on morphological and biophysical analy- sis of Echevaria. Planta 175, 153-169. Haughn, G.W., and Somerville, C.R. (1988). Genetic control of morphogenesis in Arabidopsis. Dev. Genet. 9, 73-. Heslop-Harrison, J. (1 9). Sex expression in flowering plants. Brookhaven Symp. Biol. 16, 109-125. Hicks, G.S., and Sussex, J.M. (1970). Development in vitro of excised flower primordia of Nicotiana tabacum. Can. J. Bot. 43, 133-1 39. Hicks, G.S., and Sussex, I.M. (1971). Organ regeneration in sterile culture after median bisection of the flower primordia of Nicotiana tabacum. Bot. Gaz. 132, 350-363. Hill, J.P., and Lord, E.M. (19). Floral development in Arabidop- sis thaliana: A comparison of the wild type and the homeotic pistillata mutant. Can J. Bot. 67, 2922-2936. Holder, N. (1 979). Positional information and pattern formation in plant morphogenesis and a mechanism for the involvement of plant hormones. J. Theor. Biol. 77, 195-212. Ingham, P. (1988). The molecular genetics of embryonic pattern formation in Drosophila. Nature 335, 25-34. Komaki, M.K., Okada, K., Nishino, E., and Shimura, Y. (1988). lsolation and characterization of nove1 mutants of Arabidopsis thaliana defective in flower development. Development 104, 195-203. Koornneef, M., Dellaert, L.W.M., and van der Veen, J.H. (1 982). EMS- and radiation-induced mutation frequencies at individual loci in Arabidopsis thaliana (L.) Heynh. Mut. Fies. 93, 109-123. Koornneef, M., van Eden, J., Hanhart, C.J., Stam, P., Braaksma, F.J., and Feenstra, W.J. (1983). Linkage map of Arabidopsis thaliana. J. Hered. 74, 265-272. Kunst, L., Klenz, J.E., Martinez-Zapater, J., and Haughn, G.W. (19). AP2 gene determines the identity of perianth organs in flowers of Arabidopsis thaliana. Plant Cell 1, 1195-1208. Laughon, A., and Scott, M.P. (1 984). The sequence of a Droso- phila segmentation gene: Protein structure homology with DNA binding proteins. Nature 308, 25-31. Lewis, E.B. (1978). A gene complex controlling segmentation in Drosophila. Nature 276, 565-570. McDaniel, C.N. (1980). lnfluence of leaves and roots on meristem development in Nicotiana tabacum L. cv Wisconsin 38. Planta 175,13-22. McGinnis, W.M. (1 985). Homeobox sequences of the Antenna- pedia class are conserved only in higher animal genomes. Cold Spring Harbor Symp. Quant. Biol. 50, 263-270. McGinnis, W., Levine, M.S., Hafen, E., Kuroiwa, A., and Gehr- The Arabidopsis apetala- 7 Gene 753 ing, W.J. (1984). A conserved DNA sequence in homeotic genes of the Drosophila Antennapedia and bithorax complexes. Nature 308,428-433. McHughen, A. (1980). The regulation of floral organ initiation. Bot. Gaz. 141, 3-395. Meyerowitz, E.M., and Pruitt, R.E. (1 985). Arabidopsis thaliana and plant molecular genetics. Science 229, 1214-121 8. Meyerowitz, E.M., Smyth, D.R., and Bowman, J.L. (19). Ab- normal flowers and pattern formation in floral development. Development 106,209-21 7. Okada, K., Komaki, M.K., and Shimura, Y. (19). Mutational analysis of pistil structure and development of Arabidopsis thaliana. Cell Differ. Dev. 28, 27-38. Polowick, P.L., and Sawhney, V.K. (1986). A scanning electron microscopic study on the initiation and development Of floral organs of Brassica napus (CV Westar). Am. J. Bot. 73, 254-263. Shepherd, J.C.W., McGinnis, W., Carrasco, A.E., DeRobertis, E.M., and Gehring, W.J. (1984). Fly and frog homeodomains show homologies with yeast mating type regulatory proteins. Nature 310, 70-74. Singer, S.R., and McDaniel, C.N. (1986). Floral determination in the terminal and axillary buds of Nicotiana tabacum L. Dev. Biol. 118,587-592. Sommer, H., Beltran, J.-P., Huijser, P., Pape, H., Lonnig, W.-E., Saedler, H., and Schwarz-Sommer, 2. (1990). Deficiens, a homeotic gene involved in the control of flower morphogenesis in Antirrhinum majus: The protein shows homology to transcrip- tion factors. EMBO J. 9, 605-613. Tatchell, K., Chaleff, D.T., DeFeo-Jones, D., and Scolnick, E.M. (1 984). Requirement of either of a pair of ras-related genes of Saccharomyces cerevisiae for spore viability. Nature 309, 523-527. Vaughan, J.G. (1955). The morphology and growth of the vege- tative and reproductive apices of Arabidopsis thaliana (L.) Heynh., Capsella bursa-pastoris (L.) Medic. and Anagalla arven- sis L. J. Linn. SOC. Lond. Bot. 55, 279-300.
