Plcell v1 1 37
Plcell v1 1 37
Plcell v1 1 37
We describe the effects of four recessive homeotic mutations that specifically disrupt the development of flowers
in Arabidopsis thaliana. Each of the recessive mutations affects the outcome of organ development, but not the
INTRODUCTION
Flowers develop from groups of undifferentiated cells that These particular mutations were chosen for detailed
grow from the flanks of shoot apical meristems. The cells study because each appears to cause cells in the devel-
in these floral primordia divide and then differentiate into oping flower to misinterpret their position, and thus differ-
appropriate numbers of floral organs, in appropriate entiate into inappropriate cell types. At the same time,
places. During this process of development, each cell must none of the mutations appears to cause any abnormal
somehow determine its position relative to others, and phenotype outside of the flower. The adult phenotypes of
must differentiate accordingly. Nothing is known about the some alleles of all of them have been described before,
mechanisms by which the cells of a developing flower although not in detail (Koornneef et al., 1983; Pruitt et al.,
establish their positions and subsequently give rise to 1987; Bowman et al., 1988; Haughn and Somerville, 1988).
appropriate cell types. Several processes do not seem to In addition to a detailed description of the adult phenotype,
be involved: there is no cell migration in higher plants, and we give here a description of the development of each
the totipotency of many higher plant cells indicates that mutant type, based on scanning electron microscopy.
floral primordia do not rely on deposition of positional cues Furthermore, we show that mutant alleles of two of the
in the egg cell, as is sometimes the case in aspects of genes are temperature sensitive, and we determine the
animal development. As an approach to revealing the temperature-sensitive developmental stage. Finally, the
mechanisms by which cells in developing flowers choose phenotypes and development of double mutants are de-
an appropriate developmental fate, we are studying genes scribed as a means of understanding the interactions of
whose wild-type products seem to play a central role in the gene products.
these mechanisms (Pruitt et al., 1987; Bowman et al., As a background to the descriptions of these mutant
1988). plants, we must describe briefly the appearance of wild-
In this paper we describe four homeotic mutations in the type flowers. A mature Arabidopsis flower is typical of the
flowering plant Arabidopsis thaliana, each of which ap- flowers of plants in the Brassicaceae. It is composed of
pears to affect fundamental processes in floral develop- four concentric whorls. The first whorl is occupied in wild-
ment. The mutations are agamous (ag), apetala2 (ap2), type flowers by four sepals. We will ignore the two-whorl
apetala3 (ap3), and pistillata (pi) (Koornneef et al., 1983). interpretation of the four sepals in the Brassicaceae (see
All of them are recessive mutations in single genes. Lawrence, 1951) because there seems no compelling rea-
son to treat the two pairs of sepals differently, and because
1Permanent address: Department of Genetics, Monash we wish to avoid the old controversies about whether the
University,Clayton, Victoria 3168, Australia. lateral or the medial sepals belong to the outermost whorl
2To whom correspondence should be addressed. (Arber, 1931). Inside and alternate to the sepals are four
38 The Plant Cell
petals, which occupy the second whorl. Historically, the first appear on the rim of the cylinder at the beginning of
stamens of flowers in the mustard family have been con- stage 11. This is also the stage at which the lateral
sidered as two whorls, the outer containing the two lateral nectaries appear at the base of the lateral filaments; the
stamens, and the inner the four medial ones (see Law- development of nectaries at the base of the other filaments
rence, 1951). Because no studies have revealed funda- occurs later. Stage 12 begins when petals reach the height
mental differences between different stamens, we will for of the long stamens. Figure 1A shows a mature Arabidop-
convenience and simplicity refer to the region containing sis flower; Figure 2A shows some of its early develop-
the stamens as the third whorl. The center of the flower is mental stages. A detailed description of early flower de-
occupied by a gynoecium made up of an ovary that con- velopment in Arabidopsis is being prepared (D.R. Smyth,
tains two chambers separated by a septum. The ovary is J.L. Bowman, and E.M. Meyerowitz, manuscript in
topped by a short style and a papillate stigma. Within the preparation).
ovary there develop approximately 50 ovules, attached in
The original ap2 mutant allele (Koornneef et al., 1980), two whorls. Plants grown at 25°C have an outer whorl
which we designate ap2-1, is temperature sensitive, with consisting of four organs resembling cauline leaves, rather
different phenotypes at different temperatures (Tables 1 than four sepals (Figure 1 C). That they are leaflike is shown
and 2). At all temperatures, the effects are on the outer by the presence of stellate trichomes, which are charac-
40 The Plant Cell
teristic of leaves (sepals have few, simple trichomes), not in genuine sepals (Figure 2C). In two respects these
appearing as early as stage 7 on both sides of the organs. organs are not leaflike: they have the long (>100 ^m)
These trichomes are more dense on the abaxial surface; epidermal cells that are characteristic of sepals, but not
in genuine cauline leaves they are more dense on the leaves, on their abaxial surface. In addition, they often
adaxial. In addition, these organs senesce in a way un- have stigmatic papillae at their tips, revealing a slight
characteristic of sepals, in that they do not yellow and fall gynoecial transformation. The frequency with which stigma
off shortly after anthesis. Furthermore, the early develop- tissue is seen at the tips of these organs depends on the
ment of the organs of the ap2 outer whorl is characterized position of the flower in the inflorescence, with later flowers
by the presence of stipules, present in cauline leaves but showing papillae more often. In addition, stigmatic papillae
Flower Development in Arabidopsis 41
Table 1. Summa~ofPhenotypes
Fi~tWhod SecondW h o r l ThirdWhod Fou~hWhod
Landsbergerecta
Wild-Type(wt) Sepals Petals Stamens Carpels
agamous 16-25°C wt wt Petals Flower development
repeats
apetala2 16°C Leaves wt or slightly wt wt
phylloid petals
25°C Stigmoid Staminoid petals wt wt
leaves
29°C Carpelloid Absent wt wt
leaves
show characteristics of both petal and leaf cells (Figure 5). First, simple temperature shifts were done. Plants were
Even the organs most resembling petals are seen to have taken from an incubator at 16°C and transferred to one
stomata, which are not ordinarily found on petals. This maintained at 29°C, or vice versa. Since each plant had
outward transformation decreases in later flowers, with many flowers at many different stages of development,
those flowers after the first 10 showing a slight inward the effect of a shift on the organs of the second whorl was
transformation, as at higher temperatures. recorded at all stages of floral development. As seen in
At 29°C, the outer whorl consists of leaves with a the data in Figure 6, A and B, temperature shifts in either
greater transformation toward gynoecial tissue than at direction indicate that the function of the AP2 product is
lower temperatures. Stigmatic tissue occurs at the tip of no later than the developmental period from stage 2 to
almost every organ and may extend down the lateral stage 4. Later than this, temperature shifts have no effect.
margins. Naked ovules develop on one (13 out of 136 This developmental time extends from just prior to the
organs counted) or both (2 out of 136 organs) margins; appearance of the outer whorl primordia to the time just
this occurs primarily on the medial organs. The most before the appearance of the second whorl pdmordia.
carpelloid organs resemble solitary unfused carpels, but The second type of temperature shift experiment done
with the stellate trichomes characteristic of leaves on their was a temperature pulse, in which plants at 16°C were
abaxial surface. The second whorl organs of flowers grown shifted to 29°C for 48 hr, and then shifted back to the
at 29°C either fail to develop at all or are transformed lower temperature, or, conversely, plants at 29°C were
more toward stamens than at 25°C. As at 25°C, the extent
of staminody of these organs increases with increasing
inflorescence age. The effects of temperature on the de-
velopment of the organs of the second whod are detailed Table 2. Phenotypes of SecondWhorl Organsinap2 Flowersa
in Table 2.
16oc 25oc 29°C
Observations of developing flowers indicate that the
failure of an organ to appear in the second whorl is a result % % %
of a failure of the organ primordium to initiate development. Absent <1 29 73
At no temperature does there appear to be a correlation Stamen 0 0 3
between the phenotype of a second whorl organ and its Deformed 0 2 5
position within the whorl. Nearly wild-type and almost Stamen
Filament 0 0 3
completely transformed organs may develop in adjacent <1 24 7
Petaloid Stamen
positions. The organs of the second whorl, regardless of Staminoid Petal 6 37 10
their eventual developmental fate, develop on a time Petal 82 9 0
course characteristic of wild-type petals: they develop after Phylloid Petal 6 0 0
the organs of the other whorls of the flower. Petaloid Leaf 3 0 0
The fact that the ap2-1 allele is temperature sensitive Cauline Leaflike 2 0 0
allows temperature shift experiments to reveal the time at The second whorl organs of the first 10 to 14 flowers produced
which the wild-type AP2 gene product is active in flower on at least four plants were scored and classified according to
development. Two types of experiment were performed. the phenotypes described in Figure 4.
Flower Development in Arabidopsis 43
shifted to 16°C for 90 hr (a developmental time at 16°C range from apparently normal stamens to carpelloid sta-
equivalent to 48 hr at 29°C), and then returned to 29°C. mens to normal-appearing, but unfused, carpels, as shown
Flowers developed to show the phenotype corresponding in Figure 7. The degree of carpellody increases with in-
to the temperature they experienced at stage 2 to 4, creasing temperature and increasing age of the inflores-
indicating that the AP2 product acts no earlier and no later cence, and the organs replacing the two lateral stamens
than this period of development. That ap2 flowers held at are less carpelloid than those replacing the four medial
16°C for a brief period in their development can form stamens (Table 3). These organs, when fully carpelloid,
petals shows that the AP2 product need only be active for have up to five well-developed ovules along each margin
this brief period to specify the initiation and differentiation and are capped with stigmatic papillae. The appearance
of the organs of the second whorl. The temperature- of their epidermal cells in the scanning electron microscope
sensitive period, from stages 2 to 4, lasts approximately is identical with the appearance of the epidermal cells of
75 to 90 hr at 16°C, and only 30 to 50 hr at 29°C. genuine ovaries. In some flowers two or three of the six
carpelloid organs may fuse to form a hemispherical struc-
ture. Intermediate organs are mosaics, consisting of a
apefa/a3 (ap3) patchwork of sectors with epidermal features of either
stamens or carpels. A single organ may possess both
The ap3-1 allele of the AP3 gene (Bowman et al., 1988), ovules and pollen.
like ap2-7, is a recessive temperature-sensitive mutation. The development of ap3 flowers at 25°C is morpholog-
This third chromosome mutation, like ap2-7, also causes ically identical to wild-type until stages 7 to 8, when the
transformations in two adjacent whorls, but they are the filament and anther would normally become distinct. At
second and third, rather than the first and the second, this time it becomes clear that the organs of the third whorl
ones (Table 1). At 25°C or 29°C, ap3-7 homozygotes are not producing these structures, but are elongating
develop flowers in which the organs of the second whorl vertically and forming the epidermal cell files seen on
are sepals (Figure 1D), indistinguishable from wild-type normal gynoecia (Figure 2D). The stigmas of the isolated
sepals except by their slightly smaller size. Despite their carpels appear slightly earlier than those of the the central
transformation, these organs develop in the positions, and gynoecium, with ovule formation beginning before this
on a time course, characteristic of petals. The organs of stage.
the third whorl of ap3 flowers grown at or above 25°C At 16°C, plants homozygous for ap3-7 have a second
44 The Plant Cell
(A) apetala2 16 C to 29 C
absent 65 13 46 1
stamen 1
deformed stamen 1 1 6 2
petaloid stamen 1 6 3
staminoid petal 1 1 10 2
petal 1 22 30 39
phylloid petal 1 1 18
petaloid leaf 4 3 4
(B) apetala2 29 C to 16 C
absent I 9 11 9 8
stamen
deformed stamen 1 6
petaloid stamen 1 2 5
staminoid petal 2 1 4 3 4 4
whorl that is nearly wild-type, with organs that are partly Since each inflorescence consists of flowers at many stages of
or completely white and with the smooth margins of petals, development, shifting a small number of plants provides data on
all stages of flower development. Plants were either germinated
rather than the uneven edges of a sepal. The organs are
and grown at the permissive (16°C) temperature and shifted to
not completely normal petals: they fail to reach normal the restrictive (29°C), or the converse shift was performed. The
petal size, and their epidermis is like that of sepals, con- developmental stage of all flowers at the time of the shift was
sisting of stomata and irregularly shaped cells without the noted. Because all stages defined using SEM are not distinguish-
radial cuticular thickenings of petal epidermal cells. The able in the dissecting microscope, the following stages were used:
organs in the third whorl of ap3 plants grown at 16°C all those flowers that were not initiated at the time of the shift were
develop as stamens; at this temperature the flowers can placed in the meristem (m) stage. Stages 1 and 2 were defined
self-fertilize and produce homozygous seed. as buttress (b) and primordia (p), respectively. Stages 3, 4, and 5
Figure 8, A and B, summarizes temperature shift exper- were combined into two stages: sepals (se) and sepals-plus (se+).
The sepals stage has only sepals initiated, whereas the sepals-
iments with ap3-1 homozygotes, performed to determine
plus stage has third whorl primordia initiated but the sepals have
the temperature-sensitive period of the third whorl in these
not yet enclosed the bud. The remaining flowers (stage 6 on) were
flowers. This period is later than with ap2-7. A shift up as simply classified as buds. When the flowers had matured, the
late as stage 5 can cause complete conversion of third second whorl organs of each were scored as outlined in Figure 4.
whorl organs to carpels, and there is a partial conversion The numbers are the number of organs of each type.
after a temperature shift up as late as stage 7 and perhaps (A) Fourap2-7 plants shifted from 16°C to 29°C.
even 8. A shift down, from 29°C to 16°C, has a clear (B) Four ap2-7 plants shitted from 29°C to 16°C. The TSP of the
effect until after stage 6. Thus, the AP3 gene product second whorl appears to encompass stages 2, 3, and part of 4.
Flower Development in Arabidopsis 45
appears to be effective in specifying the fate of third whorl more advanced developmental stages at the time of the
organ primordia up to the time when they begin their shift than those in which third whorl effects are seen. Thus
differentiation. The results of temperature pulse experi- again, the AP3 product acts in flowers up to the time when
ments are in accord with those of the shift experiments. If differentiation of affected organs begins.
plants are grown at 16°C, shifted to 29°C for 54 hr, and
then returned to 16°C, flowers that were in stages 5 up
to and past bud closure (stage 6 and perhaps beyond) are pistillata (pi)
affected. The converse experiment, in which plants are
changed from 29°C to 16°C for 123 hr (a developmental PI is a gene on chromosome 5; the recessive mutant allele
time at 16°C roughly equivalent to 54 hr at 29°C), and pi-1 (Koornneef et al., 1983) affects the development of all
then returned to 29°C, flowers in stages 5 and 6 are floral organs except the sepals. The organs of the second
affected. TheX>P3 gene product thus acts much later than whorl develop as small sepals rather than petals, the
that of AP2. The AP3 temperature-sensitive period lasts organs of the third whorl do not develop at all, and the
approximately 40 to 50 hr at 29°C, and 80 to 100 hr at central gynoecium develops abnormally (Table 1). The
16°C. mature flowers thus consist of two outer, alternate whorls
The temperature-sensitive period of the second whorl in of sepals surrounding a large club-shaped gynoecium,
ap3 flowers is more difficult to specify since the organs of usually composed of more than two carpels (Figure 1 E).
this whorl are not completely converted to wild-type at The ovary may exhibit unfused carpel margins and vertical
16°C. Nonetheless, in a shift-down experiment, the 16°C filamentous appendages fused to its outer surface.
phenotype of the organs of the second whorl was ob- The development of pi flowers is indistinguishable from
served in flowers that were at the same or even slightly that of wild-type until the time of the appearance of the
46 The Plant Cell
Table 3. Phenotypes of Third Whorl Organs in ap3 Flowersa Two general classes of interaction were observed: com-
binations in which the effects of the two mutations ap-
25°C, peared purely additive (ag ap3; ag pi) or close to additive
Medial Only 29°C'
16°C, All (ag ap2), and combinations in which phenotypes were
Positions 1-7 b 8-15 b Lateral Medial observed that are not seen in plants homozygous for only
% one of the mutations (ap2 ap3; ap2 pi). ap3 pi mutant
Carpel 0 0 29 6 73 plants have not been studied: the crosses to produce them
Filament 0 0 0 8 8 gave no plants that looked different from single homozy-
Staminoid Carpel 0 0 40 9 4 gotes, indicating the possibility that the double mutant
Carpelloid Stamen 0 9 28 25 7 phenotype is identical to that of one of the single mutants.
Deformed Stamen 0 37 3 38 5 The additive combinations all included agamous, ag ap3
Stamen 100 53 0 0 0 flowers grown at 25°C have the multiplication of organs
priate place at the correct time, but the third whorl primor- stamen 36
dia are not seen (Figure 2E). The gynoecium forms from
the cells encircled by the second whorl primordia, so that stage of m/b/p se se+ ist 2nd, 3rd older
it appears that the cells that would ordinarily be in the third flower at bud buds buds
time of shift
whorl are instead incorporated into the developing ovary. (1,2) (3,4) (4,5) (6) (7,8?) (8+)
Gynoecium development proceeds with characteristically
rapid vertical growth of the periphery of the central dome
of the flower primordium, but the diameter of the cylinder
(B) apetala3 29°C to 16°C
that is formed is much greater than in wild-type. Growth
of the cylinder can soon be seen to be irregular, with extra
carpels often forming and regions sometimes lagging be- absent 2 2 1
hind in vertical growth. One to four filamentous appen- carpel 1 2 16
dages emerged from the surface of the gynoecium in 56% filament 1
of 75 flowers examined; these appear to arise at the margin staminoid carpel 2 15
between carpels and can be fused to the ovary at their carpelloid stamen 2 4 18
base or along their entire length. These develop late, from deformed stamen 3 1
the wall of a developmentally advanced gynoecium. The stamen 74 36 21 18 4
and indeterminate growth of ag flowers, but instead of of 109 counted) or be fused to the gynoecium (2 out of
whorls of sepals and petals, as in ag homozygotes, they 109).
have whorls of sepals only (Figure 9A). ag pi flowers (Figure The phenotype of ap2 ap3 double mutants (Figure 9E)
9B) also consist of many whorls of sepals. The outer two is also nonadditive and dependent on temperature. At
whorls are initiated correctly, but the third whorl primordia 25°C, the four organs of the outer whorl are similar to
fail to appear, as in pi flowers. The remaining tissue follows those in ap2 flowers, but with a greater degree of carpel-
the pattern of indeterminate growth characteristic of ag, Iody, in that nearly all have stigmatic papillae, and many of
with nested internal flowers. those in medial positions have ovules on their margins.
ag ap2 double mutants grown at 25°C (Figure 9C) have Early flowers have second whorl organs resembling leaves,
an overall morphology characterized by indeterminate but most also have rudimentary Iocules, and are thus
growth and mosaic organs, as do ag single mutants. The staminoid. This is unlike the second whorl organs in the
identity of the organs is altered from that in ag, however. ap2 pi double mutants, which show no staminody. In later
at 16°C have leaves instead of sepals, but nearly normal Exceptions to this are those cells on the borders between
petals; ap2-1 plants that at 29°C have staminoid organs mosaic patches, which may be intermediate in morphol-
instead of petals have a normal third whorl of stamens; ogy, and organs in the second whorl of ap2-1 flowers.
ap3-1 plants at 25°C or 29°C have carpels in the third A notable feature of the development of most of the
whorl, but a normal gynoecium. That inner organs specify abnormal organs is that they develop on a time course
the adjacent outer whorl cannot be simply true, either. characteristic of their whorl and not of their organ identity.
Another class of model that is better supported by the With one exception, the various organs that develop in
evidence is that the flower primordium is divided into fields whorl 2 develop later than the adjacent whorl 3 organs;
or compartments, each consisting of adjacent whorls. The this is true even when all of the organs of both whorls are
early-acting gene AP2 may specify a developmental state of the same type (as, for example, in ap2-1 at 29°C). The
for the cells that will later give rise to whorls 1 and 2, identity of the organ to which a primordium develops, and
whereas the wild-type AG gene may specify a different
mutants were selected from the F2 plants. To establish strains thaliana (L.) Heynh. mit ver&ndertem Bletenbau und Bleten-
involving agarnous-l, which is sterile when homozygous, hetero- stand. Biol. Zentralbl. 90, 137-144.
zygotes were used as initial parents. Seeds were planted on a
peat moss/potting soil/sand (3:3:1, v:v:v) mixture in 55-mm pots. Dahlgren, K.V.O. (1919). Erblichkeitsversuche mit einer dekan-
The plants were grown in incubators under constant cool-white drischen Capsella bursa pastoris (L.). Svensk Bot. Tidskriff 13,
fluorescent light at 16°C, 25°C, or 29°C, and 70% relative 48-60.
humidity. Green, P.B. (1988). A theory for inflorescence development and
For scanning electron microscopy (SEM), young, primary inflo- flower formation based on morphological and biophysical analy-
rescences were fixed in 3% glutaraldehyde in 0.025 M sodium sis in Echeveria. Planta (Bed.) 175, 153-169.
phosphate (pH 7.0) at 4°C overnight, and then transferred to 1% Haughn, G.W., and Somerville, C.R. (1988). Genetic control of
osmium tetroxide in 0.05 M sodium cacodylate buffer (pH 7.0) at morphogenesis in Arabidopsis. Dev. Genet. 9, 73-89.
4°C for 12 to 24 hr. They were then rinsed in 0.025 M sodium Heslop-Harrison, J. (1963). Sex expression in flowering plants.
phosphate (pH 7.0) and dehydrated in a graded ethanol series at In Meristems and Differentiation, 16th Brookhaven Symposium
means of partitioning the developmental process. Arabidopsis Vaughan, J.G. (1955). The morphology and growth of the vege-
Inf. Serv. 2, 12-13. tative and reproductive apices of Arabidopsis tha/iana (L.)
Saunders, E.R. (1921). Note on the evolution of the double stock Heynh., Capse//a bursa-pastoris (L.) Medic., and Anagal/is ar-
(Matthiola incana). J. Genet. 11, 69-74. vensis L. J. Linn. Soc. Lond. Bot. 55, 279-301.
Sawhney, V.K., and Greyson, R.I. (1973). Morphogenesis of the
stamenless-2 mutant in tomato. II. Modifications of sex organs
in the mutant and normal flowers by plant hormones. Can. J. NOTE ADDEDIN PROOF
Bot. 51, 2473-2479.
Scott, M.P., and Carroll, S.B. (1987). The segmentation and Recent observations have shown that stigmatic tissue, which was
homeotic gene network in early Drosophila development. Cell not previously seen in ag homozygotes, may develop on the
51,689-698. leaflike organs of ap2 ag double mutant flowers grown at 29°C.
Shull, G.H. (1929). Species hybridization among old and new The isolation and characterization of two new ap2 alleles with