Inflorescence Commitment and Architecture in Arabidopsis

See allHide authors and affiliations

Science  03 Jan 1997:
Vol. 275, Issue 5296, pp. 80-83
DOI: 10.1126/science.275.5296.80


Flowering plants exhibit one of two types of inflorescence architecture: indeterminate, in which the inflorescence grows indefinitely, or determinate, in which a terminal flower is produced. The indeterminate condition is thought to have evolved from the determinate many times, independently. In two mutants in distantly related species, terminal flower 1 in Arabidopsis and centroradialis in Antirrhinum, inflorescences that are normally indeterminate are converted to a determinate architecture. The Antirrhinum gene CENTRORADIALIS (CEN) and the Arabidopsis gene TERMINAL FLOWER 1 (TFL1) were shown to be homologous, which suggests that a common mechanism underlies indeterminacy in these plants. However, unlike CEN, TFL1 is also expressed during the vegetative phase, where it delays the commitment to inflorescence development and thus affects the timing of the formation of the inflorescence meristem as well as its identity.

The architecture of inflorescences depends on when and where flowers are generated (1, 2, 3). Most species have a vegetative phase of growth whereby the apical meristem generates leaf primordia on its periphery. Secondary meristems arise in the axils of leaf primordia and may lie dormant or grow out to form side shoots. Upon receiving the appropriate environmental and developmental signals, plants switch to the reproductive phase, giving rise to an inflorescence bearing flowers in a set pattern. Two basic types of inflorescence are found among flowering plants: indeterminate and determinate (1, 4). In determinate species, the inflorescence meristem is eventually converted to a floral identity, resulting in the production of a terminal flower. Indeterminate species produce an inflorescence meristem that only generates floral meristems from its periphery.

Comparisons of inflorescence architectures from a large range of species have suggested that the indeterminate pattern was derived from the determinate (5), and therefore a mechanism arose in determinate species to inhibit the production of the terminal flower. Moreover, the wide taxonomic distribution of species with indeterminate inflorescences suggests that this condition arose several times, independently. This raises the question of whether the mechanism for generating an indeterminate inflorescence is the same or different between distantly related species. We addressed this question by exploring the genetic control of inflorescence architecture in Arabidopsis and Antirrhinum.

Recessive mutations in the CEN gene of Antirrhinum and the TFL1 gene of Arabidopsis result in the conversion of the normally indeterminate inflorescence to a determinate condition (Fig. 1A) (6, 7, 8, 9). Here, we show that CEN and TFL1 are homologs and are expressed in a similar pattern in the inflorescence apex. This finding suggests that a common mechanism for preventing terminal flower formation arose very early in evolution and may have been lost or modified in some species with determinate inflorescences; alternatively, Arabidopsis and Antirrhinum may have independently recruited the same mechanism to promote indeterminate growth. However, the time to flowering is not affected in centroradialis (cen) mutants of Antirrhinum but is significantly reduced in terminal flower 1 (tfl1) mutants of Arabidopsis (10). We show that this additional TFL1 function correlates with its expression during the vegetative phase, during which it delays the commitment of plants to form an inflorescence.

Fig. 1.

The tfl1 mutant of Arabidopsis. (A) Photograph of the tfl1-1 mutant, grown under LD. The apical meristem of tfl1 mutants first produces a basal rosette of leaves before bolting and forming the inflorescence. Bolting occurs earlier (after fewer leaves) in tfl1 plants than in the wild type, and the inflorescence meristem generates only a few flowers before it is converted to a floral meristem. (B) Cartoons of tfl1 and wild-type plants grown under LD. In the wild type, the inflorescence grows indefinitely, and flowers (circles) are generated from the periphery of indeterminate inflorescence meristems (arrowheads). Secondary inflorescences (coflorescences) arise in the axils of stem leaves. In tfl1 plants, these secondary inflorescences are often replaced by a single, terminal flower.

During the vegetative growth phase of wild-type Arabidopsis, primordia arise in a spiral and give rise to leaves separated by short internodes, forming a compact rosette. The induction of flowering by appropriate environmental signals, such as long days (LD), results in the apical meristem acquiring an inflorescence identity and generating floral meristems from its periphery. In addition, the shoot elongates (bolts), bearing two or three leaves with secondary inflorescences (coflorescences) in their axils, above which flowers occur (Fig. 1B).

The tfl1 mutant of Arabidopsis has two key features that distinguish it from the wild type: (i) it bolts early (after producing fewer rosette leaves); and (ii) the inflorescence meristem eventually acquires floral identity, leading to the production of a terminal flower (Fig. 1) (8, 9, 10, 11). Up to five floral meristems arise from the periphery of the inflorescence meristem before it acquires floral identity (11, 12). The structure of the terminal flower is often abnormal, displaying altered numbers, arrangements, and identities of organs relative to the wild type (12, 13). All of the above phenotypic effects, except for a marked change in flowering time, are also seen in cen mutants of Antirrhinum (7).

The similar effects of CEN and TFL1 on determinacy raised the possibility that they were homologs. We investigated this possibility by using a CEN cDNA at moderate stringency to probe a genomic library of Arabidopsis DNA, yielding one positive genomic clone, which was sequenced (14). In parallel, database searches with CEN revealed an Arabidopsis expressed sequence tag (EST), 129D7T7, that showed about 76% similarity over a 200-base pair (bp) region of CEN (15). This clone was fully sequenced and was shown to be identical to four regions (exons) of the genomic clone. The EST predicted a large open reading frame (ORF) that had the potential to encode a protein of 20.2 kD (Fig. 2).

Fig. 2.

Sequence comparison of the deduced amino acid sequences for TFL1 (Arabidopsis), CEN (Antirrhinum), two rice ESTs, and PBP from rats (7, 19, 20, 23). The TFL1 cDNA sequence was obtained from the Arabidopsis EST (15) and will be deposited in GenBank (accession number U77674). Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. The predicted longest ORF is shown, and point mutations detected in the tfl1 alleles indicated have the predicted changes: G → D in tfl1-1 (codon: ggc → gac); G → S in tfl1-11 (codon: ggc → agc); E → K in tfl1-13 (codon: gaa → aaa); and T → I in tfl1-14 (codon: act → att). Both rice clones appeared to be derived from unspliced transcripts or genomic DNA; only the putative exons with similarity are shown. The OSS1946A rice clone was fully sequenced but only gave the two predicted 3′ exons shown. Conserved intron positions are marked by filled triangles for TFL1, CEN, and the rice clones. The amino acids altered in the tfl1 alleles are conserved between all species. Identical and similar amino acid residues are indicated by black and gray backgrounds, respectively.

The Arabidopsis EST was mapped to the end of chromosome 5, in the region of the TFL1 locus (16). To determine whether this clone corresponded to TFL1, we sequenced the genomic region from the wild type and from four different tfl1 alleles that arose in the same Columbia background (17). Unique single point mutations were identified in each of the four tfl1 alleles: Gly → Asp in tfl1-1, Gly → Ser in tfl1-11, Glu → Lys in tfl1-13, and Thr → Ile in tfl1-14 (Fig. 2). The chance that four different mutations could have arisen in a locus other than TFL1, in each of the mutant plants, was negligible. These data, therefore, indicated that the CEN-like clone corresponded to the TFL1 gene.

The TFL1 and CEN genes are each composed of four exons that share high similarity throughout their length; the predicted proteins show ∼70% identity and 80% similarity (Fig. 2). Database searches revealed two additional plant ESTs, both from rice (OSR29181A and OSS1946A), that predicted peptides similar to the exons of TFL1 and CEN (Fig. 2) (18). The CEN and TFL1 proteins have similarity to animal phosphatidylethanolamine-binding proteins (PBPs), which can associate with membrane protein complexes, but the biological function of these proteins is unclear (Fig. 2) (7, 19). All tfl1 alleles were affected in residues that were conserved between TFL1 and PBPs, even though these residues represented only 25% of the full sequence.

RNA in situ hybridizations were used to determine the pattern of TFL1 expression. Young inflorescences of wild-type Arabidopsis showed strong TFL1 expression in a group of cells lying just below the apical dome of inflorescence and coflorescence meristems (Fig. 3A). To confirm the identity of the region in which TFL1 RNA accumulated, we compared the expression domain of TFL1 with that of LFY, a gene required for floral meristem identity (20). Double labeling showed that although LFY was expressed in floral meristems emerging on the flanks of the apex, TFL1 was confined to a distinct domain below the dome of each inflorescence (Fig. 3B). In addition to its subapical expression, TFL1 RNA was also observed throughout the stem of the inflorescence (21). The expression of TFL1 was similar to that of CEN in Antirrhinum, although CEN RNA appears to be weaker in the stem (7).

Fig. 3.

Expression of TFL1 and LFY in the inflorescence. Longitudinal sections of wild-type Arabidopsis were probed with digoxigenin-labeled antisense TFL1 or LFY RNA (24). Plants were harvested when the apical meristem had been converted to an inflorescence meristem. RNA signal was detected as a purple or red color on a white tissue background when viewed under a light field. (A) Expression of TFL1 (purple) just below the dome of the inflorescence (i). A secondary inflorescence or coflorescence (c), subtended by a leaf, arises from the inflorescence. (B) Double labeling reveals TFL1 (purple) in a subapical region and LFY (red) in young floral meristems arising from the flanks of the inflorescence. (C) Ectopic expression of LFY (purple) in the apical dome of a tfl1 mutant apex after 10 LD. Scale bar, 50 μm.

Although expression of TFL1 in the inflorescence apex might account for its effect on indeterminate growth, it is less clear how TFL1 affects flowering time. One possibility is that tfl1 mutants are committed to flower at the same time as the wild type, but the initiation of floral development and bolting are accelerated. Alternatively, commitment to flowering may occur earlier in tfl1 mutants. To distinguish between these possibilities, we compared the commitment of tfl1 mutant plants with that of the wild type by transferring plants from conditions that induce flowering [long days (LD)] to noninductive conditions [short days (SD)] at daily intervals so as to reveal the number of LD required for plants to be committed to flower. Under continuous LD, wild-type plants made about eight rosette leaves whereas tfl1 mutant plants made about six; both made about 25 leaves under SD (Fig. 4A). On average, wild-type plants were committed to flower at about 7 LD, after which transfer to SD had little effect. In contrast, tfl1 mutant plants were committed to flower at about 5 LD. This difference of 2 days can account for the difference of two or three leaves, which suggests that the early flowering in tfl1 mutants is the result of an earlier commitment to form floral meristems.

Fig. 4.

Time course of TFL1 expression and the commitment to flower. (A) Commitment to flower in wild-type and tfl1 mutant plants. Plants were grown in LD and transferred to SD at the time points shown, or grown in continuous (cont.) LD or SD as controls (25). Numbers of rosette leaves were counted for 10 to 20 plants for each time point. The commitment points (arrows) were defined as the days when ∼50% of plants still flowered after the same number of rosette leaves as plants grown in continuous LD. The error bars indicate standard error of the mean with 95% confidence limits. (B) Sections of wild-type plants were harvested after 4, 8, or 14 LD and probed for TFL1 expression (24, 25).

To determine the developmental stage of plants at the time of commitment, we analyzed wild-type and tfl1 mutant plants by scanning electron microscopy (SEM). The first floral meristems appeared on about day 8 for tfl1 mutants, but not until day 9 or 10 for the wild type. Therefore, in both the wild type and the tfl1 mutants, morphological evidence of flowering was not visible until 3 days after the commitment to flower. By day 10, the tfl1 mutants had produced about three floral meristems and expression of LFY was detected throughout the apical dome, consistent with the dome having a floral identity (Figs. 3C and 5). One day later, the apex of the tfl1 mutants was more rounded than that of the wild type, and it had sepal primordia on its periphery that were associated with its conversion to a terminal floral meristem (Fig. 5). No more lateral flowers were made once the terminal flower had initiated, and the developmental stage of the terminal flower was similar to that of the oldest lateral flower (Fig. 5). The apical meristem appeared to be recruited at about stage 2 of development, similar to cen mutants in Antirrhinum; this may account for the abnormal morphology of the terminal flower (7, 8, 9).

Fig. 5.

Inflorescence development in the tfl1 mutant. Wild-type and tfl1 mutant plants grown under LD were analyzed by SEM. Plants were harvested after 10, 11, or 12 LD, dissected, and prepared for SEM (26). After 11 or 12 LD, tfl1 mutants had a terminal flower (TF) bordered by sepals (s) and three lateral flowers. Stages of development are indicated for floral meristems. These first appear on the periphery of the inflorescence apex (stage 1) and become separated from the apex by a groove (stage 2); sepal primordia appear (stage 3); the sepal primordia develop (stage 4) and grow over the meristem (stage 5) before they cover the meristem, wherein petal and stamen primordia have initiated (stage 6) (26). Scale bar, 50 μm. All of the image below each inflorescence was blacked out using Adobe Photoshop, with the remainder unaltered.

The effect of TFL1 on commitment to flowering under LD suggested that it should be expressed during the vegetative phase, at or before day 5. To test this idea, we probed wild-type plants harvested at each LD time point with TFL1. Expression of TFL1 was detected from day 2 or 3, but it was weak up to the point of commitment in the wild type (day 7), after which the extent of TFL1 expression increased (Fig. 4B). Control sections also revealed that LFY expression was weak in leaf primordia from day 2 or 3 and appeared to increase after commitment, eventually becoming strong in floral meristems.

The roles for TFL1 in commitment and indeterminacy correlate with two patterns of expression: weak expression during early development delays commitment to flowering, whereas increased expression of TFL1 at later stages maintains inflorescence meristem identity. In Antirrhinum, CEN expression appears to be limited to the later inflorescence phase, consistent with CEN controlling only indeterminacy (7). It remains unclear which role of TFL1 is more ancestral; either TFL1 has gained a role during the evolution of Arabidopsis, or CEN has lost a role during the evolution of Antirrhinum. This question may be resolved by analyzing the roles of CEN and TFL1 homologs in other species. Phylogenetic studies have suggested that the determinate condition may have been ancestral and that the indeterminate condition arose several times in many species (5). It is possible that Arabidopsis and Antirrhinum have independently recruited the same genes, or that indeterminacy arose very early in flowering plants and has been lost in some determinate species (3).


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
View Abstract

Stay Connected to Science

Navigate This Article