Evolution of novel traits, such as floral monosymmetry, is driven by genetic changes. Comprehensive knowledge of the molecular mechanisms controlling formation of a monosymmetric corolla has been gained by analysis of loss-of-function mutants, particularly from the model species Antirrhinum majus. This process is controlled by two central regulatory genes with partially redundant functions, the TCP transcription factors CYCLOIDEA (CYC) and DICHOTOMA (DICH) (10, 11). In cyc dich double mutants, monosymmetry is lost and instead, flowers with a polysymmetric corolla are formed that are composed solely of abaxial petals (10). In single cyc or dich mutants, flowers are only weakly abaxialized, indicating that a partial subfunctionalization of the two recently evolved paralogs has occurred after duplication of the ancestral gene (11–13). A duplication event has also been reported for species from the Fabaceae (14, 15), but not for the Brassicaceae. The model species Arabidopsis thaliana harbors only one copy of the CYC ortholog, namely TCP1. In A. majus, CYC and DICH are expressed in the adaxial domain of the young flower meristem, and asymmetric CYC expression is maintained until late floral stages. Late DICH expression is restricted to the inner domain of adaxial petal lobes, reflecting its function in regulating internal petal symmetry (11). The MYB-like transcription factors RADIALIS (RAD; ref. 16) and DIVARICATA (DIV; refs. 17 and 18) participate in symmetry regulation. RAD is a target gene of CYC (19), and rad mutants, similar to cyc mutants, form partially abaxialized flowers (10). RAD acts antagonistically to DIV (16), a factor promoting abaxial petal identity.

TCP transcription factors form a diverse family. All members possess a conserved TCP domain that is unique to plants and adopts a helix–loop–helix structure (20). The closely related TCP genes CYC, TCP1, and TEOSINTE BRANCHED 1 (TB1) from Zea mays (21) cluster together in the so-called ECE clade (22) and share an arginine-rich R domain (20). The function of TCP1 from Arabidopsis, a species with a polysymmetric corolla, remains to be shown, because mutants do not exhibit obvious phenotypes (23). TCP1 is only transiently expressed in the adaxial domain of the young floral meristem, which might not be sufficient to regulate monosymmetry in the second whorl (24). This observation supports the assumption that the ancestral progenitor of the CYC/TCP1 gene was asymmetrically expressed even though the presumed ancestral corolla state was polysymmetric (19, 24). So far, analysis of CYC homologs focused on species from taxa forming monosymmetric corollae, such as Linaria, Mohavea, and Lotus. The CYC/TCP1 homologs of these species all exhibit an asymmetric expression during early and later floral development, similar to A. majus (12, 14, 25). For those species, where loss-of-function or transgenic knockdown mutants were analyzed, data support a role for the investigated TCP genes in floral symmetry formation (10, 14, 25).

Here, we report an approach to elucidate the molecular mechanism that led to the establishment of monosymmetry in the Brassicaceae. The Brassicaceae comprise ≈350 genera (26) whose members mainly form polysymmetric corollae. However, flowers of three genera, Iberis, Teesdalia, and Calepina, develop a monosymmetric corolla with differentially sized petal pairs. Iberis species exhibit the strongest size differences between the small adaxial and large abaxial petals. We isolated the CYC/TCP1 ortholog of Iberis amara, IaTCP1, and investigated its function during flower development. Studying a nonmodel species from the Brassicaceae allowed us to conduct functional analyses in Arabidopsis. Together with expression studies and analysis of a rare natural peloric Iberis flower variant, our data indicate that changes in the timing of the expression of the orthologous TCP1 gene IaTCP1 control the formation of a monosymmetric corolla in the Brassicaceae.

Development of the Monosymmetric I. amara Corolla. To unravel the molecular mechanism controlling corolla monosymmetry formation in I. amara, we first characterized the dynamics of unequal ad- and abaxial petal morphogenesis. Flower development starts with initiation of sepal primordia, with the abaxial sepal arising first and finally being slightly larger than the other sepals, similar to what has been described for A. thaliana (27). Initially, the four I. amara petal primordia are formed simultaneously as equally sized protrusions in ad- and abaxial positions on the floral meristem (Fig. 1A). Deviations in petal growth start to become apparent at the onset of stamen differentiation (Fig. 1B). From then on, the size difference between the two petal pairs increases continuously throughout flower development (Fig. 1C), generating abaxial petals that are 1.6 times larger than adaxial petals at anthesis, when the flower opens (stage A1; Fig. 1D and Table 1). Unequal petal growth is strongly enhanced after anthesis until flowers reach maturity (stage A2; compare Fig. 1 D and E). Thereby, the petal size ratio is increased from 1.6- up to 3.7-fold (Table 1).

Fig. 1. Development of the monosymmetric corolla in I. amara. Different stages of Iberis flower development are shown. (A–C) Developmental stages before anthesis were monitored under the SEM; sepals were removed to reveal inner organs. (A) After simultaneous (more ...)

Table 1. Quantification of I. amara petal surface area and petal cell size

To determine whether this unequal petal growth is due to an altered rate of cell proliferation and/or cell expansion, sizes of adaxial epidermal petal cells were quantified and compared. Cell sizes of ad- and abaxial petals are similar in flowers of stages A1 and A2, ≈240 and 510 μm², respectively (Table 1). Given that observation, it is likely that unequal cell proliferation rather than differential cell expansion accounts for the strong petal size differences and thus for the formation of the monosymmetric corolla in I. amara.

Iberis corolla monosymmetry allows flowers to sit closely together without canopying neighboring flowers with protruding adaxial petals. Along with retarded early internode elongation and pedicel growth, this contributes to shaping the peculiar architecture of young I. amara inflorescences. Inflorescences adopt a corymboid structure with a flattened top and superficially resemble one single flower (Fig. 1F), an effect typically generated by the umbel of Apiaceae.

Isolation and Expression Analysis of IaTCP1. To test whether a CYC/TCP1 homolog is involved in controlling the formation of unequally sized Iberis petal pairs, we isolated IaTCP1 from I. amara. The 1.480-nt-long IaTCP1 transcript contains an ORF of 981 nt coding for a predicted IaTCP1 protein of 327 aa. As characteristic for CYC/TCP1 homologs, IaTCP1 contains a TCP domain of 59 aa that adopts a basic helix–loop–helix structure known to be involved in DNA binding and protein dimerization (28). Also, an R domain is present, likely forming a coiled-coil structure that might function in protein–protein interactions (20, 29). IaTCP1 and TCP1, the likely CYC ortholog from Arabidopsis (23), share 64% identity at the nucleotide and 59% identity at the protein level. Considering only the TCP domain, amino acid identity increases up to 95%, which together with phylogenetic tree analysis [supporting information (SI) Fig. 6] supports the assumption that IaTCP1 represents the orthologous TCP1 gene.

IaTCP1 expression levels were quantified by RT-PCR experiments in vegetative and floral Iberis organs (Fig. 2). IaTCP1 expression is low in leaves and shoots, as well as in inflorescences, flowers before anthesis, fully mature flowers, and gynoecia. However, when ad- and abaxial petal pairs were separately analyzed, a strong differential IaTCP1 expression was detected. Expression is high in the two smaller adaxial petals and weaker in the large abaxial petals. This differential IaTCP1 expression is dynamic throughout corolla development. Before anthesis, IaTCP1 expression is ≈10-fold (10.96 ± 1.31) higher in adaxial as compared with abaxial petals. The difference increases up to a peak of >90-fold (93.18 ± 2.21) just after anthesis when buds open and decreases to a level of almost 20-fold (19.17 ± 4.56), once flowers reach maturity.

Fig. 2. RT-PCR analysis of IaTCP1 expression in vegetative and floral organs. (Upper) The graph depicts average IaTCP1 transcript values from three independent reactions, normalized to the expression strength of IaRan3. Error bars indicate standard deviations. (more ...)

In situ hybridization experiments were carried out to determine the tissue-specific IaTCP1 expression at earlier floral stages (Fig. 3). For the Arabidopsis TCP1 gene, early and transient asymmetric expression in the young floral meristem has been detected, which vanishes before floral organ primordia are initiated (24). No distinct asymmetric IaTCP1 expression could be detected in young floral Iberis meristems (Fig. 3A). After onset of stamen differentiation, low IaTCP1 expression was observed in young petals and stamens (Fig. 3B). However, once floral organ differentiation advanced further, a stronger IaTCP1 expression was detectable in adaxial compared with abaxial petals (Fig. 3C). Therefore, in contrast to the expression of the orthologous Arabidopsis TCP1 gene, asymmetric IaTCP1 expression is established during later flower development, after floral primordia are initiated and organs started to differentiate.

Fig. 3. In situ expression pattern of IaTCP1 and IaH4 in young Iberis flowers. (A–C) IaTCP1 antisense probe hybridized to a longitudinal section through the Iberis inflorescence meristem and young developing flowers. (A) No distinct asymmetric IaTCP1 (more ...)

Because the observed petal size difference is likely caused by a differential rate of cell proliferation, Histone 4, a cell cycle marker gene indicative of the S phase (30), was isolated from Iberis (IaH4) and its expression analyzed. Generally, IaH4 is strongly expressed in regions with high cell proliferation activity, e.g., in inflorescence and floral meristems (data not shown). Comparing serial sections probed with IaTCP1 and IaH4, we detected a complementary RNA expression strength in young petals. IaTCP1 is expressed more strongly in adaxial petals (Fig. 3C), whereas IaH4 transcript is present in a larger number of abaxial compared with adaxial petal cells (Fig. 3D). This implies that unequal petal sizes are established by different cell proliferation rates in the two petal pairs. The dynamic of IaTCP1 expression correlates with the dynamic of petal morphogenesis. This observation supports a function for IaTCP1 during late petal development, reaching its maximal activity probably around the time of anthesis.

Analysis of a Peloric Flower Variant of I. amara. During cultivation of a large number of I. amara plants, we could identify a few plants that randomly produced one to two flowers in an inflorescence where corolla monosymmetry was lost. These flower variants showed an abaxialized corolla, where the two small adaxial petals of wild-type Iberis flowers (Fig. 4A) were replaced by two large abaxialized petals (Fig. 4B). Sufficient petal material from these peloric flower variants could be harvested from stage A2 to compare the IaTCP1 expression levels in wild-type and peloric petals by RT-PCR. Wild-type petals showed a strong IaTCP1 expression (Fig. 4C). Given the above-reported wild-type RT-PCR data, this is likely due to high IaTCP1 expression in adaxial petals. However, in petals from the peloric variant, the amount of IaTCP1 transcript was strongly reduced in the abaxialized corolla compared with the wild-type corolla (Fig. 4C), further corroborating an important role for IaTCP1 in controlling petal growth.

Fig. 4. A peloric flower variant of I. amara develops an abaxialized corolla. (A) Wild-type flower at stage A2. (B) Naturally occurring peloric flower variant with only abaxialized petals. (C) Representative RT-PCR result showing reduced IaTCP1 expression in (more ...)

Functional IaTCP1 Analysis by Constitutive Expression in A. thaliana. To further analyze the IaTCP1 function, the effect of constitutive IaTCP1 expression was studied in a related heterologous system by generating transgenic A. thaliana plants. Transgenic T1 plants expressing IaTCP1 under the control of the CaMV35S promoter exhibited deviations of different strength from vegetative and floral wild-type development and were accordingly grouped into three classes. Different T1 plants were selfed, and heritability of phenotypes was confirmed in segregating T2 populations. Individuals with a strong phenotype were dwarfish and often produced flowers that lacked mature floral organs and failed to open (Fig. 5C). Transgenic plants with an intermediate phenotype produced flowers with smaller and narrower petals (Fig. 5B). Strong and intermediate phenotype plants showed enhanced outgrowth of secondary shoots, and inflorescences continued to proliferate and produced flowers for an extended period compared with wild-type plants. In addition, we also observed plants with a wild-type-like appearance displaying a normal vegetative and floral development (data not shown). Intermediate phenotype T2 plants, derived from selfing the intermediate phenotype T1 plant 35S::IaTCP1/2, were further analyzed. The size of Arabidopsis wild-type petals was determined to be 3.22 mm² (±0.36). Petals of plants with an intermediate phenotype had a reduced petal surface area of 1.60 mm² (±0.33) and were thus ≈2-fold smaller than wild-type petals. The transcript abundance of IaTCP1 in inflorescences of T2 plants from the intermediate and the strong phenotype class was analyzed by RT-PCR. T2 plants with strong phenotypic changes had higher IaTCP1 expression levels than those belonging to the intermediate phenotype group (Fig. 5F), proving that expression strength of IaTCP1 transcripts in transgenic plants correlates with phenotype severity.

Fig. 5. Constitutive expression of IaTCP1, TCP1, and CYC in Arabidopsis affects petal growth. Front view of flower and habitus of wild type (A) and representative transgenic plants expressing IaTCP1 under the control of the CaMV35S promoter (B and C). (B) Transgenic (more ...)

To investigate whether a reduction of the petal area in the intermediate phenotype transgenic Arabidopsis plants is achieved via an effect on cell proliferation, petal cell sizes were determined and compared with wild type. Cell sizes of the smaller intermediate phenotype T2 petals were similar (206.89 ± 31.28 μm²) to those of the wild-type petals (198.22 ± 46.73 μm²). We therefore assume that IaTCP1 overexpression in the heterologous Arabidopsis system affects petal growth by a mechanism similar to the one governing differential petal growth in Iberis, namely, by unequal cell proliferation.

Comparison of IaTCP1/CYC/TCP1 Activities upon Constitutive Expression in A. thaliana. To compare the function of IaTCP1 with that of CYC/TCP1 genes from species with mono- and polysymmetric corollae, TCP1 from A. thaliana and CYC from A. majus were constitutively expressed in A. thaliana.

Transgenic T1 Arabidopsis plants overexpressing TCP1 showed heritable effects comparable with those ectopically expressing IaTCP1 and were accordingly grouped into three phenotypic classes. Intermediate phenotype flowers also formed smaller petals (Fig. 5D) and showed similar vegetative growth defects (data not shown). Petal measurement was conducted with representative intermediate phenotype T2 plants segregating in a population from the selfed T1 plant 35S::TCP1/6. As determined for petals overexpressing IaTCP1, overexpression of TCP1 causes an ≈2-fold reduction in petal size (1.73 ± 0.33 mm²) compared with wild-type petals (3.22 ± 0.36 mm²).

In contrast, overexpression of the Antirrhinum transcription factor CYC exerts an opposite effect on petal development and increases petal surfaces of transgenic plants compared with wild type (Fig. 5E). Similar results were recently described when CYC was transiently expressed in Arabidopsis petals by using an inducible system (19). The authors observed that after induction of CYC expression, flowers produce petals that are 1.5-fold larger compared with wild-type ones.

Thus, all the three orthologous TCP genes from Iberis, Arabidopsis, and Antirrhinum affect petal morphogenesis. However, whereas a conserved function was observed for the cruciferous TCP1 and IaTCP1 genes, CYC overexpression causes an opposite effect on petal size, implying that significant functional changes must have occurred since the separation of the two TCP lineages from their last common ancestor.

To investigate monosymmetry formation in the Brassicaceae, we isolated and characterized the likely CYC/TCP1 ortholog IaTCP1 from I. amara, a cruciferous species displaying a monosymmetric corolla. Our expression and functional data show that late and differential IaTCP1 expression contributes to the formation of the two differentially sized petal pairs.

IaTCP1 Expression Dynamic Correlates with Monosymmetry Formation. SEM analyses reveal that the four petal primordia of I. amara are initiated simultaneously. After onset of stamen differentiation, differences between ad- and abaxial petal sizes become detectable. However, strongest differential petal growth occurs late in flower development, after anthesis. Comparison of cell sizes from ad- and abaxial petals revealed similar values, indicating that the unequal petal size is due to changes in cell proliferation rather than cell expansion.

Elevated transcript levels in adaxial petals compared with abaxial ones became recognizable by in situ hybridization experiments after stamens started to differentiate. RT-PCR revealed a dynamic differential IaTCP1 expression, with a maximal expression difference occurring after anthesis when the strongest gain in unequal petal growth is realized. Expression analysis using the cell cycle marker IaH4 further corroborates the assumption that differential petal growth is realized by unequal cell proliferation, because IaH4 and IaTCP1 expression strength differences in ad- and abaxial petal pairs are complementary. Late and enhanced IaTCP1 expression in adaxial petals might expose a negative effect on cell proliferation leading to the development of smaller adaxial petals compared with abaxial ones and thereby to a monosymmetric corolla. A negative effect on cell cycle progression has also been described for other TCP transcription factors, such as CYC and CINCINNATA from Antirrhinum and TCP2 and TCP4 from Arabidopsis (30–32).

The timing of IaTCP1 mRNA expression differs from that described for other CYC/TCP1 homologs from species with a monosymmetric corolla belonging to the Asterids (Antirrhinum and Linaria; refs. 10 and 25) or Rosid clades, other than the Brassicaceae (Lotus and Lupinus; refs. 14 and 33). There, respective CYC homologs are expressed in the adaxial domain of the developing flower, starting at early floral meristem stages and being maintained until late floral stages. In Arabidopsis, developing a polysymmetric corolla, TCP1 is also expressed early in the adaxial region of the floral meristem. However, this expression is transient and disappears before floral organ primordia are initiated, which might account for the lack of monosymmetry in the corolla (24). As an early asymmetry of CYC/TCP1 gene expression is common to both Asterids and Rosids, it is assumed that it might represent the ancient expression state in the common ancestor (19, 24). Accordingly, early asymmetric IaTCP1 expression might have been lost in Iberis. Alternatively, instead of a common ancestral expression pattern, a high flexibility in modulating CYC/TCP1 gene expression could have established an early and/or late asymmetric expression independently several times. Further expression studies with CYC/TCP1 homologs from evolutionary informative monosymmetric as well as polysymmetric species will help to elucidate this question. Development of a more complex monosymmetric corolla, such as that of A. majus, depends on petal form and folding in addition to petal size differences and might require an earlier onset and maintenance of CYC expression. Corolla monosymmetry in Iberis is revealed only by different petal pair sizes and, because it is a less complex feature, it bears the potential to unravel first, initial changes in regulatory networks that generated monosymmetry in the Brassicaceae. Establishment of the late IaTCP1 expression in Iberis, along with high expression differences in the two petal pairs, seems to govern the formation of unequal ad- and abaxial petals in Iberis.

Function of IaTCP1 and TCP1, but Not CYC, Is Conserved in the Brassicaceae. In rarely occurring peloric Iberis flowers, where all four petals adopt the size of abaxial petals, IaTCP1 expression was strongly reduced, further supporting a function of IaTCP1 in shaping a monosymmetric corolla. Peloric flowers with abaxialized petals have been described for other monosymmetric species. In Antirrhinum, peloric flower formation is caused by transposon insertions in the CYC and DICH loci, disrupting their function (10, 11). In peloric Linaria flowers, the respective CYC ortholog is silenced by hypermethylation (25). IaTCP1 expression, however, is not completely abolished in peloric Iberis petals, but the remaining transcript level is likely below the threshold required to exert a negative effect on petal growth.

Additional functional data were obtained by overexpression of IaTCP1 in the related cruciferous species A. thaliana. Transgenic plants with an intermediate phenotype produce flowers with an approximately two times reduced petal size compared with wild-type flowers. Similar to the results from cell size measurement in Iberis petals, reduction of the petal area in Arabidopsis plants overexpressing IaTCP1 is likely achieved via affecting cell proliferation rates rather than the extend of cell expansion. Overexpression of the orthologous Arabidoposis TCP1 gene causes the same phenotypes. Thus, both TCP proteins show similar effects upon ectopic expression, indicating that their cis- and transregulatory functions, involved in cell cycle regulation, are conserved. This observation leads to the assumption that establishment of the monosymmetric corolla in Iberis is largely accounted for by strong and asymmetric IaTCP1 expression during late petal development, whereas the early transient floral TCP1 expression in Arabidopsis might not be sufficient to affect morphogenesis. Contrarily, ectopic expression of CYC from Antirrhinum, belonging to the Asterid clade, leads to the formation of Arabidopsis flowers with enlarged petals. Similar results were recently obtained by Costa et al. (19), who used an inducible CYC expression system that caused enhanced growth of Arabidopsis petals. However, this size increase was mainly achieved by enhanced cell expansion. The authors demonstrated that CYC activity in transgenic Arabidopsis flowers is not able to activate endogenous Arabidopsis RAD genes, whereas the Antirrhinum RAD gene is a direct CYC target (19). Therefore, TCP genes from Rosids and Asterids seem to have diverged strongly at the level of protein function since they separated from their last common ancestor. This divergence, along with expression differences, might have led to a specific interplay between cis- and transregulatory mechanisms in the different taxa. Amino acid comparison of TCP domains shows that a neutral alanine found in TCP1 (20) and in IaTCP1 is replaced by a hydrophobic proline in CYC. Secondary structure predictions (ref. 20; data not shown) indicate this might have an impact on the shape of the basic helix–loop–helix structure which could account for differences in protein–protein interactions and/or DNA binding. Subfunctionalization of duplicated TCP genes likely also contributed to more complex flower morphologies as described for Antirrhinum and Mohavea (11, 12). However, in Iberis, no other close IaTCP1 homolog could be isolated, indicating that, as in Arabidopsis, no duplication event occurred.

What could be the adaptive advantage of corolla monosymmetry in a cruciferous species? Monosymmetry is considered a key innovation that contributed greatly to the successful radiation of angiosperm taxa like the Lamiales, Leguminosae, and Orchidaceae (5). The advantage generally accredited to monosymmetry is an enhanced precision of pollination via guidance of the approaching insects and a facilitated recognition by pollinators (6, 34). For the cruciferous species Erysimum mediohispanicum, it has been shown that the number of pollinator visits is increased for individuals forming monosymmetrical flowers (35). Our data show that altered timing of IaTCP1 expression governs unequal petal growth, allowing flowers to sit closely together within young inflorescences. This, along with retarded internode and pedicel elongation, shapes an untypical cruciferous inflorescence that adopts a corymboid architecture superficially resembling one large single flower. It will be intriguing to determine the effect of the combined morphological changes in Iberis on pollinator attraction.

To conclude, the altered temporal expression of a key regulator is a mechanism to generate morphological differences in closely related species from one family. This can serve as a basis to unravel how more complex morphological changes might have evolved between distantly related species by evolutionary tinkering.

Plant Material and Growth Conditions. I. amara seeds were obtained from Saatgut Kiepenkerl (Everswinkel, Germany). Plants were grown in the greenhouse or in a field at the Max Planck Institute for Plant Breeding Research (Cologne, Germany). Transgenic and wild-type A. thaliana (Col.) plants were grown at 20°C with 13-h light/11-h dark cycles.

Morphological Analyses. For scanning electronic microscopy, material was either prepared after critical-point drying (36) or frozen in liquid nitrogen. Material was sputter-coated with gold and examined with a digital scanning microscope (DSM940; Zeiss, Oberkochen, Germany). To determine the petal surface area, pictures were taken with a binocular (MZELIII; Leica, Wetzlar, Germany). Quantification of petal surface area and cell size was performed by using IMAGEJ 1.32J (http://rsb.info.nih.gov/nih-image). Averages of petal surface area were determined by measurement of at least 27 petals for each petal type/phenotypic category. For recording adaxial epidermal petal cell sizes, the upper third of at least seven petals was analyzed, and values were calculated based on a minimum of 200 measured cells. Young flowers just after flower opening (anthesis) were defined as stage A1 flowers (Fig. 1D). Fully mature flowers, releasing pollen, were determined as stage A2 flowers (Fig. 1E). For morphological and molecular analysis of I. amara stage A1 and A2 petals, flowers were isolated from inflorescences possessing up to 20 open flowers, such as shown in Fig. 1F.

Isolation of Nucleic Acids and Expression Analysis. Total RNA was isolated with the RNeasy Plant Mini kit (Qiagen, Valencia, CA). IaTCP1 was isolated from genomic Iberis DNA by using A. thaliana TCP1-specific primers (for primer sequences, see SI Data). Remaining sequence and transcript length information was obtained conducting 3′ RACE (FirstChoice RLM-Race; Ambion, Austin, TX) and 5′ RACE (5′/3′ RACE Kit; Roche, Indianapolis, IN) experiments. Sequences were cloned into the pCR2.1 (Invitrogen, Carlsbad, CA) and pGEM-T Easy (Promega, Madison, WI) vectors. RT-PCR analysis was performed as described (37) by using 1.8 μg of total RNA for cDNA synthesis. PCRs were carried out at an annealing temperature of 60°C, followed by 24 cycles for I. amara and 19 cycles for transgenic Arabidopsis plants. Averages of IaTCP1 expression in I. amara corollae are based on three PCR repetitions. Transcript abundance was quantified by using ImageQuant 5.2 (Molecular Dynamics, Sunnyvale, CA). For normalization, the I. amara homolog of the A. thaliana Ran3 GTPase (At5g55190) was isolated (IaRan3) and amplified with 23 cycles. Sequence comparison was carried out by using the program DNAMAN 4.0 (Lynnon BioSoft, Los Angeles, CA).

In Situ Hybridization. A 1,386-bp-long IaTCP1 fragment spanning the TCP and R domain was cloned into the vector pCR2.1 and used as a template for RNA probe preparation after vector linearization with HindIII. A 274-bp-long I. amara Histone 4 (IaH4) DNA fragment was isolated by using primers specific for H4 from A. thaliana (At5g59690) and cloned into the pGEM-T Easy vector. A reverse primer containing the T7-binding domain was used to prepare a PCR template for transcription. Digoxigenin-labeled riboprobes of IaTCP1 and IaH4 were made by using DIG RNA labeling mix and T7 Polymerase (Roche), and hybridizations were carried out as described by Zachgo (38).

Transgenic Plants. Complete coding sequences of IaTCP1, TCP1, and CYC were isolated from I. amara, A. thaliana, and A. majus, respectively. DNA fragments were cloned into pGEM-T Easy, sequenced, and subcloned into pBAR35S (AJ251014), carrying a CaMV35S promoter/terminator cassette. Constructs were introduced into A. thaliana plants by floral dip (39) by using the Agrobacterium tumefaciens strain GV3101. Transgenic plants were selected by spraying with 0.1% vol/vol BASTA (AgrEvo, Hoechst, Düsseldorf, Germany). Heritability of morphological changes observed in T1 populations was confirmed by analyzing T2 plants. To correlate phenotype strength of transgenic plants with IaTCP1 transcript abundance, RT-PCRs were conducted with inflorescences from 10 intermediate and 10 strong phenotype T2 plants, respectively. T2 plants were generated by selfing the intermediate T1 plant 35S::IaTCP1/2. For normalization, A. thaliana Ran3 GTPase (At5g55190) was amplified with 21 cycles.

We thank Zsuzsanna Schwarz-Sommer, Heinz Saedler, and Johanna Schmitt for comments on the manuscript. S.Z. is grateful to Heinz Saedler for ongoing support and stimulating discussions. A.B. received a scholarship from the German founding agency Deutsche Forschungsgemeinschaft (DFG; Graduierten Kolleg, “Molecular analysis of developmental processes”). This work was supported by DFG Grant Za 259/6-1 (to S.Z.)