ABSTRACT
We isolated and analyzed mutants of Arabidopsis thaliana, acaulis, with flower stalks that are almost absent or are much reduced in length. The mutations are divided between two loci, acaulisl (acll) and acaulis2 (acl2). The acll-1 mutation has been assigned to linkage group 4 in the vicinity of locus ap2. The acl1-1 mutant showed premature arrest of the inflorescence meristem after the onset of reproductive development, followed by consequent reduction in the number of flower-bearing phytomers and therefore flowers. The apical meristem of the inflorescences was morphologically normal but its radius was about half that of the wild type. The acl1 mutants are also defective in the development of foliage leaves. Both defects could be rescued by growth at a specific temperature (28°C). The length of the cells in acl1-3 mutant was less than that in the wild type but the numbers of cells in leaves and internodes of acl1 mutants were calculated to be the same as those of the wild type. Thus, the defects in inflorescences and leaves were attributed to defects in the process of elongation (maturation) of these cells. Temperature-shift experiments showed that the Acl1+ product was necessary at all developmental stages. A critical stage was shown to exist for recovery from the cessation of development of inflorescence meristems that was caused by the acl1-1 mutation. Grafting experiments showed that the acl1-1 mutation does not affect diffusible substances. An analysis of double mutants carrying both acl1-1 and one of developmental mutations, ap1, clv1, Ify, or tfl1, showed that ACL1 is a new class of gene.
INTRODUCTION
A current focus of molecular biology is the elucidation of molecular mechanisms of developmental processes in multicellular eucaryotes. In plants, one of the most characteristic features of development is axial development. There is a gradient in terms of age along the axial structure, as there are active meristematic tissues at the apices of the shoot and root. These apical meristems divide continuously to establish the basic patterns of new organs and tissues. There are also lateral primordia, which start to develop when environmental conditions become favorable for the growth of lateral tissues. The main axial structure of a plant is its shoot system, which is a stack of unit segments, each composed of an internode with a leaf and a bud. The unit segment is called a phytomer (Evans, 1940). The ontogeny and morphology of shoots represent the most fundamental aspects of axial development. Shoot systems are divided developmentally into two stages, vegetative and reproductive. The vegetative shoot system encompasses leaves, stems, and buds. The reproductive shoot system is called the inflorescence and is involved in the formation of leaves, flower stalks and flowers. On the basis of their ontogeny, inflorescences of plants are categorized as determinate and indeterminate (reviewed by Weberling, 1989). The arrangement of components can be expected to be under genetic control and to be susceptible to environmental conditions (Hilu, 1983; Marx, 1983).
We are interested in identifying genetic tools with which to analyze the developmental control of reproductive shoots in Arabidopsis thaliana, which has indeterminate inflorescences. During the process of flowering in A. thaliana, the development of floral organs and the bolting (elongation of internodes of inflorescences) of flower stalks seem to be tightly coupled. At the vegetative stage, a shoot of A. thaliana is in a compressed form (rosette) and its internodes do not elongate. The first few flower buds differentiate at the center of the rosette, and then flower stalks begin to elongate, with the developing flower buds remaining at the tips of the stalks (Smyth et al., 1990). Under normal conditions, wild-type plants do not extend the internodes of their flower stalks before the completion of the development of the first few flower buds. Long flower stalks are always observed when these flower buds start to open. Some species (e.g., Antirrhinum majus L. and Petunia ×hybrida Hort. Vilm.) have elongated internodes during both vegetative and reproductive stages, and other species exhibit the rosette arrangement even at the reproductive stage (for example, most species of the genus Viola). The wild-type strain of Belgian endive (Cichorium intybusL., Compositae) initiates flowers on flower stalks after bolting. However, transgenic plants that had been transformed with the aid of Agrobacterium rhizogenes strain A4 produced erect stems (bolting) but did not produce flowers on the elongated stems (Sun et al., 1991). According to the review of Wareing and Phillips (1981), results of many physiological experiments with long-day plants suggested that bolting and differentiation of flower buds are apparently independent processes. Hyoscya-mus niger L. begins the elongation of internodes (bolting) before the differentiation of flower buds when the plantlets are supplied with gibberellic acid (GA3) under short-day conditions. Even when AMO-1618, an inhibitor of GA synthesis, which inhibits the elongation of internodes (bolting) completely, is added to cultures of Silene armeria L., the differentiation of flower buds progresses normally (Wareing and Phillips 1981). In certain species of Gramineae, including teosinte (Euchlaena mexicana), internode elongation is known to begin independently of panicle formation, when a certain number of internodes have differentiated (Takeda, 1977). Taken together, as Wareing and Philips (1981) discussed earlier, these various observations suggest that the processes of development of flower buds and bolting are independent of one another. The apparent coupling of bolting and floral development in Ara - bidopsis may only mean that these two processes are temporally coincidental events. In order to dissect the processes involved in the development of flower stalks, we attempted to isolate flower-stalkless mutants of A. thaliana. This paper describes a new mutation of this type, designated acaulis.
MATERIALS AND METHODS
Isolation and characterization of mutant lines
Seeds were sown on rockwool moistened with MGRL medium and plants were grown at 22°C under continuous white light as described elsewhere (Tsukaya et al., 1991). For temperature-shift experiments, growth temperatures were 16°C, 22°C and 28°C, as described in the text.
The screening for mutants was performed using M2 populations derived from ethyl methanesulfonate(EMS)-mutagenized seeds of Arabidopsis thaliana ecotype Columbia in Tokyo laboratory. The marker lines used for genetic mapping were GPR1 (an, gl1l, tt4, th1) derived from ecotype Columbia, and MSU22 (er, hy2, gl1l, tt5), MSU15 (er, cer2, bp, ap2) and W100 (an, ap1, er, py, hy2, gl1, bp, cer2, ms1, tt3) derived from ecotype Landsberg. We used the following genetic nomenclature, which is based on the proceedings of the Third International Arabidopsis Meeting (East Lansing, MI, USA, 1987) and has been widely used by molecular geneticists who study Arabidopsis: wild-type alleles are given in block capitals and italicized; mutant alleles are given in lower-case, italicized letters; and protein products are given in block capitals, without italics.
Morphological characterization
All measurements were made after growth of mutant lines and wild-type plants (Columbia wild) as described above. The measurements were made on several independent plants.
Organs used for histological analysis were obtained from plants grown under the same conditions. For anatomical studies, all organs were fixed overnight in FAA, which contained 5% (v/v) acetic acid, 45% (v/v) ethanol, and 5% (v/v) formaldehyde and they were then dehydrated in a graded ethanol series at room temperature. Completely dehydrated tissues were then preincubated in a solution of 50% (v/v) Technovit 7100 (Kulzer & Co. GmbH, Wehrheim, FRG) and 5% (v/v) glycerol in ethanol for about 1 hour and then washed in 100% Technovit resin. Next, the samples were dipped in 100% resin and incubated for 6 hours in a vacuum to ensure penetration by the resin. The samples were incubated overnight in air. Hardening of resin and embedding in 100% Technovit 7100 resin were achieved by the method described in the manual from the supplier of the Technovit 7100 resin. Sections of 10 pm thickness were cut with Histoknives (Kulzer & Co. GmbH) on an LR-85 microtome (Yamato-Koki, Asaka, Japan), affixed to glass slides, and stained with 0.1% (w/v) toluidine blue in 0.1 M phosphate buffer (pH 7.0) at 50°C for 30 seconds. Specimens were then photographed under bright-field illumination.
Organs used for SEM were obtained from plants grown under identical conditions to those described above. Samples were fixed in FAA solution overnight and dehydrated in a graded ethanol series at room temperature. Isoamyl acetate was then gradually substituted for the ethanol. Tissue samples were critical-point dried in liquid carbon dioxide. After mounting of individual samples on stubs for SEM, the tissue was sputter-coated with gold and palladium. A Hitachi HCP-2 type SEM was used for examination of specimens, and samples were photographed on Kodak TMAX 100 film.
Grafting experiments
For grafting experiments, a cleft-graft technique was used. The basal part of a scion (i.e., donor) was cut into a wedge with a stainless-steel razor blade. The prepared scion was inserted into the slit of a stock (i.e., acceptor) that had been slit with a razor blade, and cut surfaces were carefully realigned. The stock and scion were then bound with a strip of several-mm-wide ‘Band-Aid’ tape (Johnson and Johnson, USA). Culture conditions were the same as in the other experiments.
Detachment of flower buds
For simplicity, the first flower bud that differentiated was named the first flower bud, the second one was named the second flower bud, etc. The relationship between the developmental stage of the first or second flower bud (Smyth et al., 1990) and the length of the inflorescence was examined. Decapitation was carried out by cutting off the apical region that included the third flower bud of the inflorescence at various stages. Detachment of the first flower bud from the inflorescence was carried out by cutting off the peduncle. 2 days after the decapitation or detachment of a flower bud, the developmental stage of oldest flower bud and the length of the inflorescence were recorded.
Analysis of double mutant
The following developmental mutant lines were used for construction of double mutants with acl1-1: flo5 (clvl-2) (supplied by Dr G. Haughn of the University of Saskatchewan, Saskatoon, Canada), ap1-1 (supplied by Dr M. Koornneef of the Agricultural University, Wageningen, the Netherlands), terminal flower (isolated from the M2 generation as acaulis mutants and then used for this study), and leafy-6 (supplied by Dr D. Weigle of the California Institute of Technology, Pasadena, CA, USA).
erecta (er) and hy2 mutants were also used. An er mutation was obtained after the crossing of co er strain (Redéi’s) with Columbia wild type and elimination of the co mutation. This allele was tentatively named er-101. An hy2 mutation was obtained by S. Naito and named hy2-314 allele.
RESULTS
Mutant isolation and genetic analysis of mutants
In an attempt to isolate mutations that affect the elongation of the flower stalk we screened 5000 EMS-treated M2 families with a Columbia wild-type background (in Tokyo). The common dwarf mutants were not included in this study and are distinguished by short internodes and various reproductive anomalies (Fig. 1). Many of the dwarf mutants respond favorably to treatments with gibberellic acid (see, for review, Rédei and Koncz, 1992).
Three mutants with stalks of greatly reduced length were isolated (Fig. 2). A similar mutation had been obtained earlier (in Columbia) after X-ray irradiation (Rédei, 1990; Koncz and Rédei, unpublished data). These mutants, unlike the dwarf mutants, have fully fertile flowers, fruits of normal size and completely normal seed sets. This type of mutation was named acaulis (acl in the three-letter nomenclature), for stalkless (Fig. 1). The wild-type gene is, thus, named ACAULIS (ACL). These mutants were selfed twice and then backcrossed to a Columbia wild type. All the mutations except for ATYK2033 were found to be recessive. ATYK2033 was found to be semi-dominant. The resulting acaulis plants were detected in F2 generations and were selfed more than four times before further study. Reciprocal crossing of four mutants was performed in order to determine the genetic basis of the Acaulis− phenotypes. The results defined two complementation groups that generate the Acaulis− phenotype. The first locus, represented by more than a single mutant allele, was named acll and the alleles were designated acll-1 (in strain 294-321 isolated by one of us, GPR), acll-2 (in strain ATYK2032), and acll-3 (in strain ATYK2034). The other locus was named acl2 and the allele was designated acl2-1 (in strain ATYK2033). Fig. 2 shows the Acaulis− phenotype of each mutant in a comparison with Columbia wild type. For detailed studies we chose to examine the acl1 locus with three alleles that are associated with different degrees of expression (Fig. 2).
The chromosomal location of the acl1 locus was determined by use of testers that represented the five linkage groups in Arabidopsis. In the 124 progeny of F2 plants obtained from crossing the acl1-1 mutant and the MSU15 tester, four acl1 bp double homozygotes and no acl1 ap2 homozygote were obtained. From this result the acl1 locus was calculated to lie 0±9.0 cM from the ap2 locus and 45.2 cM from the bp locus in linkage group 4 (at map position 4-63; Koornneef and Stam, 1992). From 133 progeny of F2 plants obtained from the acl1-1 ×ag cross, no acl1 ag homozygote was obtained. The other tested markers localized on the other chromosomes, such as er, flo5, gi-2, gl1, hy2, as, an, tt4, and th1, segregated independently of the acl1 locus.
Gross morphology of acl1 mutants
Fig. 3 shows the morphological characteristics of the acl1l-1l mutant and the wild type. We should note that the control of the elongation of flower stalks by the acl1 mutation was not absolute and the length of flower stalks varied among plants grown side by side, for unidentified reasons. The variations among mutant plants are shown in Fig. 3B,C, as two groups: the most severely affected plants (indicated by open squares on the curve marked ‘s’) and the least severely affected plants (open squares). Fig. 3A illustrates the changes in length of flower stalks during the life cycle of normal and mutant plants. The mutant (acl1-1) had a defect in the elongation of flower stalks that resulted in an approximately 30-fold reduction in length at maturity. The number of flowers also decreased (Fig. 3B). However, the number of leaves (including cauline leaves and rosette leaves) did not change very much (Fig. 3C). Thus, the defect in the elongation of flower stalks due to the acl1-1 mutation was not caused by a general reduction in plant growth. The length of a flower stalk varied from almost 0 mm to several cm. The stalks were thin and not straight, and they had small and twisted cauline leaves (schematic illustration in Figs 1, 2).
We employed metamer terminology (Schultz and Haughn, 1991) in order to describe our acl1 mutants. As shown in Table 1 and Fig. 2, wild-type plants had about three phytomers in the type 2 metamer, which was characterized by the presence of a cauline leaf and an elongated internode. Acl1-1 and acl1-3 plants also had about 3 phytomers on the type 2 metamer. In type 2 metamers, the length of each internode of acl1-3 mutants was 22%, 18%, and 14% (from bottom to top) of that of wild-type plants (mean values from 16 acl1-3 plants and 20 wild-type plants). Thus, the number of components of type 2 metamer in acl1 mutants is normal and only elongation is defective.
In type 3 metamers, a lack of components was found. As shown in Table 1, the number of phytomers of type 3 metamers, was 2.4 in the acl1-1 mutant, 7.1 in the acll-3 mutant, and 20.7 in the wild type. A defect in the elongation of internodes was also recognized in type 3 metamers (the length of the botom internode in aca1-3 mutants was 15% that of the wild type). We examined secondary lateral shoots from both acl1 and wild-type inflorescences to determine whether the development of acl1 secondary lateral shoots was similar to that of shoots on main stems. Secondary lateral shoots of acl1 mutants had the same gross morphology as those on main stems.
We noted an additional defect in the morphology of leaves of acl1 mutants (Fig. 2). Rosette leaves of mature acl1 plants were small, irregularly buckled and twisted and they tended to curl downwards. In all mutant lines, this altered leaf morphology cosegregated with the abovementioned Acaulis− phenotype. Initially, seedlings with the first pair of foliage leaves were similar to wild-type seedlings. The altered leaf morphology usually appeared after the development of third foliage leaves or sometimes after that of fourth foliage leaves, depending on growth conditions. This observation agreed with that of Medford et al. (1992). However, we also observed that first foliage leaves were irregular in shape in the severest cases. The seventh and eighth (i.e., mature) leaves were about 3.5 times shorter than wild-type leaves and the same was true of their width (Table 2). Fig. 4 shows measurements of the radii of rosettes. The mutant rosettes had a smaller radius after development of foliage leaves, and radii of rosettes were 3-to 10-fold smaller than those of the wild type at maturity, reflecting the differences in the dimensions of leaves. The defect in leaves was further studied by light-microscopic examinations of sectioned leaves. The measurements of leaf cells obtained in this way are also summarized in Table 2. Comparing the sizes of parenchymatous cells and epidermal cells between leaves, we found that all the cells in a given leaf were smaller in the presence of the acll-1 mutation. The intercellular spaces were also smaller in the mutants. The number of cell layers was increased about 3-fold to compensate for the decrease in cell volume caused by the mutation (Table 2). The stomatal indices of wild-type plants and acl1-1 were the same.
Ultrastructure of apices of inflorescences
In order to determine how the acl1-1 mutation interrupts the development of inflorescences, the apical regions of inflorescences were observed by scanning electron microscopy (SEM). In the case of wild-type plants, the apex of 42-day-old plants was similar to that of 29-day-old plants in terms of the retention of apical domes (Fig. 5A,B,D). The apical meristem continued to differentiate flower buds for more than two weeks and flower buds at various stages of development were apparent in inflorescence apices as a result. In contrast, the apex of inflorescences of acl1-1 plants differentiated only a few flower buds (Fig. 5E), and then ceased further development with more aborted flower buds (Fig. 5F,H). This pattern of development resulted in the Acaulis− phenotype. The cessation of production of flower buds in the acl1-1 mutant was not caused by the disappearance of the inflorescence meristem itself, because the apical meristem appeared to be morphologically normal and dome-shaped (Fig. 5F-H). The terminal morphology of the 29-day-old inflorescence of acl1-1 plants was similar to that of old (42-day-old) wild-type plants with respect to the presence of some aborted flower buds adjacent to the apical dome (Fig. 5D,F). The diameter of 29-day-old wild-type apical meristems was 45.6±6.5 μm (mean value ± s.e. for 5 independent apices). The diameter of 29-day-old apical meristem of acl1-1 mutant plants was 22.9±2.7 μm and that of old (with aborted flower buds, in the final stage of development) wild-type apical meristems was 24.6±2.1 μm. The small size of the apical dome was confirmed by observations of sections of shoot apices (Fig. 5C,G). The difference in size of the apical region seemed to occur only after floral induction and no abnormality was recognized in the vegetative stage from both SEM observations and sectional views (data not shown). The size of cells in apical domes was observed to be the same in the wild type and acl1-1 mutants (Fig. 5C,G).
Lateral shoots in the acl1-1 mutant had the same characteristics as shoots on main stems in terms of ultrastructural morphology as seen by SEM. The observations by SEM did not reveal any significant differences between the acl1-1 mutant and its parent in other organs, namely, flowers and seeds (data not shown).
Elongation of flower stalks in wild-type and acll plants
To characterize the defects associated with the acl1 mutations, we first examined wild-type plants of ecotype Columbia for the elongation of flower stalks. Consistent with previous descriptions by Smyth et al. (1990) of ecotype Landsberg erecta, the main inflorescence began to elongate rapidly around 18 days after sowing at the developmental stage 9 (after the definition of Smyth et al., 1990), as the first flower bud differentiated (Fig. 6A). This correlation was unchanged in the acl1-3 mutant (Fig. 7). In order to determine whether a flower bud or an apical meristem is responsible for the elongation of a flower stalk, we performed decapitation and flower bud-detachment experiments. When apical regions above the third flower bud were detached from plants at the bolting stage, the elongation pattern of the inflorescence was not affected (Fig. 6B). However, when young first flower buds were detached from inflorescences, elongation of such inflorescences was delayed (Fig. 6C). Thus, the bolting process in wild-type plants is apparently regulated not by the apical meristem but by the developmental stage of the first flower bud itself (Fig. 6B,C).
Anatomical study of internodes
To clarify whether the defect in elongation of internodes is caused by a decrease in cell number or inadequate cell size, an anatomical study was performed of sections of second internodes of type 2 metamers from the wild type and a leaky mutant, acl1-3. As shown in Fig. 8, the length of the internode was proportional to the length of the internodal cell in both the wild type and the acl1-3 mutant (Fig. 8). This result suggests that cell numbers were unchanged by the acl1 mutation.
Grafting experiments
In order to determine the effects of leaves on the development of inflorescences in mutants, grafting experiments were carried out. Distal parts of inflorescences of acl1-1 plants were grafted onto proximal parts of inflorescences of wild-type plants as stocks (i.e., acceptors). As a control experiment, grafting between wild-type stock and wild-type scions (i.e., donors) was performed. In 4 out of 7 grafting experiments, successful connection (knitting) and normal growth were observed (Fig. 9A). Successful connection between the wild-type stock and the acl1-1 scion was achieved in 3 of 9 graftings. As shown in Fig. 9B, the grafted acl1-1 inflorescences were not affected by growth on the wild-type plants. And, as shown in Fig. 9C, the grafted wild-type inflorescences were not affected by growth on the acl1-1 plants in 3 successfully connected plants of 5 grafting experiments.
Cold sensitivity of the phenotype caused by acl1 mutations
To find clues that might facilitate the analysis of the mutation, we examined the effects of various changes in temperature, illumination, and nutrients on the severity of the phenotype. Of the conditions examined, only temperature reproducibly affected the severity of the phenotype. We found that growth at 28°C resulted in the absence of the defective phenotype, and leaves and inflorescences of plants with any of the three acl1 mutations were normal (Fig. 10D). When plants were germinated and cultured at 22°C for 7-10 days and then transferred to 25°C, 28°C and 16°C, wild-type plants showed evidence only of retardation of growth after incubation at lower temperatures (Fig. 10A). By contrast, the acl1-1 plants grown at 28°C showed discrete differences in morphology to those grown at 16°C (Fig. 10A), 22°C (Fig. 10B) or 25°C (Fig. 10C). The same results were obtained with other acl1 mutations (acl1-2 and acl1-3).
In order to examine the effects of temperature in greater detail, we performed a set of temperature-shift experiments. When acl1-1 mutant plants were cultured at a permissive temperature, 28°C, for about 3 weeks after sowing, the plants produced normal (wild-type) leaves in the form of a rosette. These plants were then transferred to a restrictive temperature, 16°C. They developed very short flower stalks on the top of the normal rosette after the transfer (Fig. 11A). At the permissive temperature (28°C), bolting began 22 days after sowing. When acl1-1 mutant plants with normal inflorescences, which had been cultivated at 28°C for 26 days, were transferred to 16°C, further elongation of the internodes stopped (Fig. 11B). In wild-type plants, the elongation of the internodes did not cease and the final lengths of inflorescences were not affected by a shift from 28°C to 16°C (Fig. 11). Whenever bolting acll-1 mutant plants grown at 28°C were transferred to 16°C, elongation continued for about another 10 mm and then ceased (Fig. 11C). When mutant plants grown at 28°C were transferred to 16°C before bolting, the resultant inflorescences varied in length from almost 0 mm to about 10 mm (Fig. 11C).
Shift experiments in which plants were transferred from a restrictive temperature (22°C) to a permissive one (28°C), showed a dramatic effect on flower stalk development (Fig. 11). When we shifted mutant plants with their first flower buds at stages less than stage 12, from 22°C to 28°C, the flower stalks started to elongate and normal sized inflorescences were formed. However, the flower stalk failed to elongate in the case of plants with first flower buds that were beyond stage 12 (Fig. 11D).
Analysis of double mutants
In order to determine the role of the ACL1 gene within the developmental network, some developmental mutations that affect the morphology of inflorescences and/or flowers were introduced into the acl1-1 mutant and double mutants were generated.
For the double mutant analysis we employed the following mutations: lfy and tfl1, which affect the initiation of floral primordia and the morphology of flower stalks; ap1, which affects the formation of flowers; and clv1 (flo5) which affects the development of shoot apices. Fig 12 shows some of the phenotypes of double mutants. The acl1-1 tfl double mutant made only a few flowers, as did acl1, but it differentiated a terminal flower (Fig. 12E). acl1-1l lfy-6 double mutants made only a few flower-like structures that were composed of carpelloid organs (Fig. 13), which is a characteristic of the lfy mutation (Schultz and Haughn, 1991). The increase in the number of cauline leaves, one of the phenotypes known to be caused by the lfy mutation, was also observed in the acll-l lfy double mutant (Fig. 12F). acl1-1 ap1-1 (Fig. 12G) and acl1-1 flo5 (clv1-2) (Fig. 12H) double mutants differentiated flowers with the phenotypes expected of flo5 or ap1 mutants on the short flower stalks caused by the acll-l mutation. Therefore, tfl, lfy, flo5 (clv1) and ap1 mutations are additive with respect to the acl1-1mutation.
In order to gain insight into the control of flower stalk growth, we also examined the effects of the mutations erecta (er, Rédei, 1962) and hy2, which appear to influence the length of internodes. The internodes of acl1-1 er-101 or acl1-1 hy2-314 double mutants were longer than those of acl1-1. Table 3 summarizes the measurement of the number of flowers on the main axis. The double mutants had twice the number of flowers as the acl1-1 single mutant. The morphology of leaves did not change after the introduction of the er-101 or hy2-314 mutation.
DISCUSSION
In this report we have described a set of acl1 mutants with flower stalks that are almost absent or are much reduced in length and with very small numbers of flowers. The acl1l mutations are associated with single recessive traits and the acl1 locus was mapped to linkage group 4. In the wild-type main inflorescence (i.e., flower stalk with flowers), an apex continues to grow and produce flower buds indefinitely, in principle. The apex of the inflorescence of the acl1-1 mutant cannot grow beyond a certain point. Apparently, cessation of the differentiation of each floral meristem occurs and then the flower buds near the apical region cease development or vice versa. Analysis of apical meristems using cross sections and SEM revealed no morphological abnormalities in apical regions of the acl1l-1 mutant. Although a decrease in diameter of the apical dome was observed at an earlier stage in the acl1-1 mutant than in wild-type plants, this decrease should be a reflection of cessation of the development of the apical dome at an earlier stage. We conclude that the acl1-1 mutant ceases to elongate its flower stalk, not because of loss of the inflorescence meristem but, apparently, because of the arrest of further development of meristems, for the following two reasons. The acl1-1 tfl1 double mutant still retains the capacity to differentiate terminal flowers and the acl1 mutants can recover a normal phenotype by the temperature shift after a certain period of time.
Foliage leaves of the acl1 mutants are very small and twisted. Both the small size of leaf cells and the smaller intercellular spaces imply that the process of elongation of the leaf cells ceases soon after their differentiation. As in other plant species, leaf cells derived from one leaf pri-mordium divide repeatedly and differentiate according to each cell’s fate (Pyke et al., 1991). After their differentiation, leaf cells of A. thaliana expand to the adult size by increasing the volume of each vacuole and the surface area of the cell wall. The change in leaf morphology in the mutant is caused by a defect in the last process.
acl1 mutants have a defect in elongation of inflorescences
Development of inflorescences in Arabidopsis thaliana involves two phases. First, the primary inflorescence meristem produces cauline leaves associated with secondary lateral inflorescences. Then, the inflorescence meristem switches to the formation of flowers. This situation is well documented by reference to the metamer concept of Schultz and Haughn (1991). Wild-type plants of A. thaliana produce three types of metamer: type 1, rosette; type 2, coflorescence-bearing with cauline leaves; and type 3, flowerbearing without bract (Fig. 1).
Using the metamer concept, we can describe the acll mutation with relative ease. Compared with wild-type plants, the mutants are defective in the elongation of type 2 and type 3 metamers but not in the start of elongation. The acl1 mutants have all of the three types of metamers, but they have very short type 2 metamers and precociously terminate type 3-like metamers after formation of a few flowers. This explanation of the defect is also applicable to the lateral shoots, i.e., the coflorescences (or branches).
The acl1 mutants that we have isolated to date ceased both the elongation of main stems and further development of floral meristems. In order to analyze this linkage between the length of internodes and the number of flowers, we introduced hy2 and er mutations into acl1-1 mutants. The hy2 mutant has long internodes (Koonneef et al., 1980) and the er mutant was believed to be short (Redei, 1962). We found that er-101 mutants are generally short but the first internode in the type 2-metamer elongates (our unpublished observations). When we could increase the length of the first internodes of type-2 metamer by a hy2-314 or er-101 mutation, the double mutant produced more flowers.
Thus, there is tight linkage between the length of flower stalks and floral development, although bolting is regulated independently from the development of the first few flowers (Smyth et al., 1990). In order to determine what organs are responsible for the synchronization of the two independent processes, we performed decapitation and flower bud-detachment experiments. Decapitation did not influence the tight linkage between the developmental stage of the first flower bud and bolting, but detachment of the first flower bud delayed bolting. Thus, the start of the bolting process is dependent on the start of a certain program(s) that is involved in the development of some floral organ(s) in the first flower bud and not on the shoot apex. Considering that lfy and pin-formed (Goto et al., 1991) mutants without normal flowers can exhibit bolting, this effect of the first flower bud would be secondary.
ACL1 belongs to a new category of inflorescence gene
Some genes have been identified within the genetic network that control the development of inflorescences and flowers. LFY and TFL1 are responsible for the initiation of development of floral primordia (Schultz et al., 1991, Shannon et al., 1991). Floral homeotic genes, such as AG, AP1, AP3, and PI (Bowman et al., 1991), are categorized as genes responsible for the formation of flowers after the onset of switches by LFY and TFL1. The CLV1 gene appears to control the development of shoot apices (Lyser and Furner, 1992). From the analyses of double mutants, the function of the ACL1 gene within the network of inflorescence-development is different from that of other known genes such as AP1, CLV1, LFY and TFL. The effect of another homeotic mutation agamous is very interesting because agamous mutants retain the capacity to develop further flowers at floral apices (Bowman et al., 1989). However, no ag acl1-1 double mutant appeared among about 300 F2 progeny examined because the two mutations are closely linked to one another.
Possible function of the ACL1 gene
The mutational phenomena in leaves and internodes of acl1-1 plants can be explained by the arrest of cell expansion. Such a change is categorized as neoteny (lack of completion of development), a kind of heterochronia (Guerrant, 1982), wherein the mature organ has juvenile features. Thus, the ACL1 gene must have an important role in cell maturation (elongation) in both leaf cells and internodal cells. Another important role is continuous production of flower-bearing phytomers in flower stalks.
The temperature-shift experiments revealed two aspects of the function of the ACL1 gene. The Acl1+ product is required constantly. If the Acl1+ product is missing or not functioning at a critical point or for a certain period of time, the acl1 mutants cannot recover normal development.
Since acl1 mutants expressed pleiotropic defects, the acl1 genetic defect may affect general and indirect function. The defects may be the consequence of some phytohormone imbalance. However, the addition of several growth regulators, including phytohormones and their analogs, failed to have any clear effect on the phenotype under our experimental conditions. The grafting experiments showed that the acl1 mutation does not affect a diffusible substance. An auxin-insensitive mutant, axr1, has a defect in internode elongation. This is caused by a decrease in cell number in the internodes but not by a decrease in cell length (Lincoln et al., 1990). This phenotype is different from the dwarfism of acl1 mutants which is caused by a decrease in cell length. Another auxin-insensitive mutant, axr2, has a defect in inflorescences that is caused by a decrease in cell length but not by a decrease in cell number (Timpte et al., 1992). This is the same as the acl1 mutants in this study, but axr2 mutants differ from acl1 mutants, which do not have a defect in the hypocotyl.
There is also another possible function of the ACL1 gene. It may be responsible for a specific cell-wall or cytoskeletal component that is required for both the expansion of cells in leaves/internodes and the division of meristem cells for the elongation of inflorescences. However, it is difficult to imagine a component that is necessary for the division of cells in inflorescence meristems but not for the division of cells in leaves and flower stalks.
A third possible explanation of ACL1 function is that the primary defect in the acl1 mutant is in the maturation (elongation) of cells, with the cessation of development of inflorescence meristems being a secondary effect. Since we suggested a linkage between the development of internodes and of flowers, negative feedback regulation of floral development from internodal development might be postulated. if there is such negative feedback, the cessation of development of further flowers in acl1 mutants would be the result of feedback regulation from internodes that cannot elongate. Results of double mutants of acl1-1 hy2 and acl1-1 er favor our negative-regulatory hypothesis. Although some compensatory mechanisms in reproductive organs must also be hypothesized, given that cells in the reproductive organs are not affected, as are vegetative cells, by acl1 mutations, the negative-feedback hypothesis appears to explain the phenotypes observed in acl1 mutants rather simply. Proof of the negative-feedback hypothesis awaits future study. Since there is a large gap in our understanding of the relationship between gross multicellular morphology and subcellular structures in plant development, many other possibilities can be postulated. Further physiological, morphological, and molecular investigations of the gross morphology of Arabidopsis must be performed.
A very similar mutation, known as the det (deteminate) mutation, was identified in Pisum sativum L. (Singer et al., 1990). The det mutant of pea also exhibits arrested development of inflorescence apices at early stages and produces a limited number of flowers, with the apical dome remaining intact. From these apparent similarities, it is possible that the acl1 mutation of Arabidopsis and the det mutation of pea may have a common mechanism.
Further clarification of the mechanism of action of the normal and mutant genes can be expected from molecular analysis. The acl1-1 mutation was obtained by X-ray irradiation, which is known to cause deletions and other chromosomal defects. Preliminary genetic analysis has shown a reduction of 30% or more in the transmission of the acl1-1 allele than would logically be expected, and such reductions are frequently an indicator of the involvement of a deletion at the locus of interest and/or in its vicinity. Therefore, an attempt will be made to isolate the ACL1 gene by the ‘genomic subtraction’ procedure of Sun et al. (1992).
ACKNOWLEDGEMENTS
We thank Dr R. Imaichi of Tamagawa University for technical advices and discussion about the histological analysis using Technovit resin, Dr T. Araki for naming the acaulis mutation, and Mrs K. Shinozaki, an operator of the SEM at the University Museum, the University of Tokyo, for her skilled assistance. We also thank Dr G. Haughn of the University of Saskatchewan, Dr J. Medford of Pennsylvania State University, and Dr Keiko U. Torii of University of Tokyo for critical readings of earlier versions of our manuscript and their kind encouragement.
H. T. was supported by a fellowship for Japanese junior scientists from the Japan Society for the Promotion of Science.