FD and FDP bZIP transcription factors and FT florigen regulate floral development and control homeotic gene expression in Arabidopsis floral meristems Open Access

Plant shoots are formed from meristems that appear at different stages of plant development. The shoot apical meristem (SAM) is formed during embryogenesis and contains a population of undifferentiated stem cells. These cells are located at the tip of the shoot, and their descendants differentiate on the periphery of the meristem to give rise to organ primordia (Barton, 2010; Fuchs and Lohmann, 2020). During vegetative development of Arabidopsis, the primordia on the flanks of the meristem give rise to leaves, and axillary meristems in the axils of these leaves form lateral shoots. Later during shoot growth, and dependent on the environmental conditions, the identity of the SAM changes to an inflorescence meristem, growth of leaf primordia is suppressed, and the identity of axillary meristems change to that of floral primordia that contain floral meristems. The apical inflorescence meristem and floral meristems share aspects of their genetic programs, so that some regulatory genes that confer inflorescence identity on the SAM are also expressed in flowers and contribute to floral development (Gregis et al., 2009; Liu et al., 2009; Xi and Yu, 2009).

In the SAM, FD is expressed during vegetative development and persists throughout the transition to flowering (Abe et al., 2005; Romera-Branchat et al., 2020; Wigge et al., 2005). Moreover, when the plant is exposed to long photoperiods (LDs), mutations in FD delay the transition of the SAM from a vegetative to an inflorescence meristem, causing late flowering (Abe et al., 2005; Wigge et al., 2005). The closely related FD PARALOGUE (FDP) protein is expressed below the SAM in a complementary expression pattern to that of FD, and fdp-CRP mutations slightly accelerate flowering (Romera-Branchat et al., 2020). FD and FDP physically interact with the florigen proteins FLOWERING LOCUS T (FT) and TWIN SISTER OF FT (TSF) (Abe et al., 2005, 2019; Wigge et al., 2005). The FT and/or TSF genes are transcribed in the vasculature of leaves in response to LDs and their proteins are transported through the phloem to the SAM where they interact with FD to promote flowering (Adrian et al., 2010; Chen et al., 2018; Corbesier et al., 2007; Jaeger and Wigge, 2007; Mathieu et al., 2007). The interaction between FD and FT is mediated by 14-3-3 proteins to form the florigen activation complex (FAC) (Taoka et al., 2011), which allows transcriptional activation of primary FD targets such as SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1) and FRUITFULL (FUL), floral meristem identity genes such as APETALA1 (AP1; AGL7) and LEAFY (LFY), which activate each other (Bowman et al., 1993; Wagner et al., 1999), and floral organ identity genes such as those of the SEPALLATA family (SEP) (Collani et al., 2019; Romera-Branchat et al., 2020; Wigge et al., 2005; Zhu et al., 2020). Notably, FD has an additional role in maintaining indeterminacy of the inflorescence meristem by preventing transition of the SAM into a floral meristem. In this role, FD interacts with the FT-related protein TERMINAL FLOWER 1 (TFL1) to repress the expression in the SAM of genes involved in floral development and thereby maintain SAM indeterminacy (Cerise et al., 2023; Goretti et al., 2020; Hanano and Goto, 2011; Zhu et al., 2020).

Arabidopsis floral primordia transition through distinct stages from emergence at stage 1 to the mature flower opening at stage 12 (Smyth et al., 1990). During this process, floral organs form in four concentric whorls: sepals initiate at stage 3, petals and stamens at stage 5, and carpel primordia at stage 6. By the end of stage 2, the floral meristem forms centrally, characterized by regulatory genes like WUSCHEL (WUS) and CLAVATA3 (CLV3) (Fletcher et al., 1999; Long and Barton, 2000; Mayer et al., 1998). The mature flower contains four sepals, four petals, six stamens and two fused carpels. At stage 6, the floral meristem converts into carpels, involving the repression of WUS (Lenhard et al., 2001; Lohmann et al., 2001). The identity of each whorl is controlled by homeotic genes within the ABC model, where combinations of A, B and C functions specify sepal, petal, stamen and carpel identities, mainly through MADS-domain transcription factors: AP1 (A function), PISTILLATA (PI) and APETALA3 (AP3) (B function), and AGAMOUS (AG) (C function), which is also involved in termination of growth of the floral meristem by transcriptional repression of WUS (Lenhard et al., 2001; Lohmann et al., 2001). Additionally, SEPALLATA MADS-box genes (SEP1-SEP4) provide an E function, which is essential for floral organ identity, as inactivation of all four SEP genes converts floral organs into leaves (Ditta et al., 2004).

Here, we have studied the roles of FD, FDP, FT and TSF in Arabidopsis floral development. FT is expressed in the vasculature of the pedicel and silique of mature flowers, and has been described to maintain inflorescence and floral identity, so that inflorescences of ft mutants contain inflorescence nodes that form vegetative structures rather than flowers (Liu et al., 2014; Muller-Xing et al., 2014). We found that FD, FDP, FT and TSF contribute to the regulation of floral meristem size and floral organ development by directly activating SEP gene transcription. Moreover, FD, FDP and FT are present in the floral meristem from floral stage 2-3, where they increase expression of the SEP2, SEP3 and AG genes, which is associated with reduced WUS expression and floral meristem size regulation. We discuss the significance of the different functions of FT, TSF, FD and FDP at distinct stages of reproductive development to control the identity of the SAM and the growth of floral meristems.

FD and FDP are expressed in floral primordia (Abe et al., 2005; Romera-Branchat et al., 2020; Wigge et al., 2005), but their patterns of expression have not been described in detail. To analyze their spatio-temporal expression patterns in flowers, in situ hybridizations were performed on plants grown under short days (SDs) and transferred to long days (LDs) for 9 days (Fig. 1Ai). FD mRNA was present at the base of stage 1 and stage 2 floral primordia, and appeared in the floral meristem at stage 3 (Fig. 1Ai, Fig. S1Ai, ii). FDP mRNA was detected at stage 3 in the lower floral meristem and adaxial sepals (Fig. 1Aii, Fig. S1Aiii), soon after the formation of the floral meristem at stage 2-3 (Long and Barton, 2000). At stage 3, the spatial patterns of expression of FD and FDP in the floral meristem are similar to their distribution in the inflorescence meristem (Romera-Branchat et al., 2020).

Higher spatial resolution of FD and FDP expression patterns was obtained by imaging flowers of FD::VENUS:FD and FDP::VENUS:FDP transgenic lines (Romera-Branchat et al., 2020) by confocal microscopy. VENUS:FD was present at the base and in the abaxial region of stage 1 and stage 2 floral primordia (Fig. 1Bi). At stage 3-4, VENUS:FD was detected broadly at the floral meristem apex and at the base of sepals (Fig. 1Bi,Ci). At stage 6, VENUS:FD was broadly detected in the first, third and fourth whorls and in the pedicel (Fig. 1Cii), and was expressed throughout developing petal primordia at stage 9 of floral development (Fig. 1Ciii). VENUS:FDP was detected only from stage 2 and 3 at the base of the floral primordia (Fig. 1Bii). From stage 4, it was more broadly detected in the basal part of the floral meristem, in the pedicel, epidermis and adaxial side of the sepals (Fig. 1Civ, v). This pattern was also maintained at stage 6 (Fig. 1Cv). In older flowers (stage 9), VENUS:FDP was also detected at the base of the petals (Fig. 1Cvi). Co-localization of these bZIP transcription factors (TFs) was tested using plants containing both FDP::VENUS:FDP and FD::mCHERRY:FD transgenes (Martignago et al., 2023). VENUS:FDP and mCHERRY:FD were co-expressed in a few cells at the center of the floral meristem at stage 3 and in the epidermis, and this overlap was maintained in older floral buds, including the pedicel cells (Fig. 1Di-iv; Fig. S1Bi-vi).

The floral phenotypes of fd-3, fdp-2 and fdp-CRP2 single mutants, and double mutants fd-3 fdp-CRP2 and fd-3 fdp-2 (Romera-Branchat et al., 2020) were analyzed. The fdp-2 and fdp-CRP2 mutations are a dominant missense mutation in the DNA-binding domain and a null frameshift mutation, respectively, and the respective mutant flowers resembled wild-type (Col-0) flowers, with no significant changes in floral organ number (Fig. 2B; Table 1, Table S1). Moreover, fd-3 mutant flowers showed minor differences in organ number (4% with three sepals, 15% with more than four petals) (Fig. 2B; Table 1, Table S1). However, the double mutants showed a synergistic effect on floral organ number: over 36% and 50% of fd-3 fdp-CRP2 flowers had altered numbers of sepals or petals, respectively, and over 50% and 65% of fd-3 fdp-2 flowers had these alterations (Fig. 2A,B, Table S1). Introduction of FD::VENUS:FD into fd-3 fdp-2 largely complemented these defects to Col-0 levels (Fig. 2B, Table 1, Table S1), supporting a partially redundant role for these bZIP TFs in floral development.

Young floral buds of fd-3 fdp-2 and fd-3 fdp-CRP2 double mutants and of single mutants at stages 4 and 5 were analyzed by scanning electron microscopy (SEM). Notably, only the double mutants exhibited defects in organ development, and formed more than four sepal primordia, unlike Col-0, which always had four (Fig. 2C). In these mutants, adaxial and lateral sepals were smaller, abnormally shaped and more numerous, whereas abaxial sepals appeared similar to Col-0 (Fig. 2C). Older flowers at stages 11, 12 and 13 were also examined (Fig. S2). At stage 13, fd-3 fdp-CRP2 double mutants displayed a wider pedicel compared with Col-0 (Fig. S2E-G). To quantify this effect, pedicel width at anthesis (stage 13) was then measured for Col-0, fd-3 and fdp-CRP2 and for the fd-3 fdp-CRP2 double mutant. Both fd-3 single and the double mutant showed wider pedicels than the Col-0 (Fig. S2H, Table S2). Additionally, double mutants showed altered organ number and identity, with abnormal sepal shapes and bifurcate trichomes (Fig. S2A-G). Sepaloid cells were also observed on petals (Fig. S2G). These findings suggest that FD and FDP play partially redundant functions in regulating floral organ number and identity.

The promotion of floral transition by FD depends on FT and TSF in the FAC (Taoka et al., 2011). To investigate the involvement of FT and TSF in floral development, we examined the floral phenotypes of ft-10, tsf-1 and ft-10 tsf-1 mutants, and compared them to the fd-3 fdp-CRP2 double mutant (Fig. 3A-D; Fig. S3). The tsf-1 mutant exhibited a phenotype similar to Col-0, whereas the ft-10 mutant showed variability in sepal and petal number, with only 70% of flowers developing four sepals and 60% developing four petals (Fig. 3B). However, ft-10 tsf-1 mutants displayed more severe defects across all four floral whorls, including variable sepal numbers, reduced petal numbers and abnormally shaped carpels (Fig. 3A,B, Table 1, Table S1). While ft-10 tsf-1 flowers contained an average of 14 floral organs, the fd-3 fdp-CRP2 double mutant showed a mean of 16 organs, resembling Col-0 plants (Table 1). In the first whorl, ft-10 tsf-1 mutants exhibited a high degree of variability in sepal number. Only 20% of flowers had four sepals, whereas 40% had fewer than four sepals and the remaining 40% had more than four sepals. In contrast, the fd-3 fdp-CRP2 double mutant had a higher proportion of flowers with the normal pattern of four sepals, with ∼60% of flowers developing four sepals, 20% having fewer than four, and 20% having more than four sepals (Fig. S3). The second whorl was more severely affected in ft-10 tsf-1 mutants, with only 15% of flowers developing four petals, while ∼50% had either no petals or just a single petal. In contrast, fd-3 fdp-CRP2 did not show such a drastic reduction in petal number (Fig. S3). Approximately 50% of fd-3 fdp-CRP2 flowers had the normal pattern of four petals, as in Col-0, while the rest predominantly developed five petals. This suggests that, while fd-3 fdp-CRP2 does not fully maintain normal petal numbers, its petal phenotype differs from ft-10 tsf-1, where petal loss is a major characteristic.

Further analysis by SEM revealed that ft-10 tsf-1 floral buds showed more growth defects and disorganized whorls compared with ft-10 (Fig. 3Ci-iv). At stages 4-5, Col-0 and single mutants had larger abaxial and adaxial sepals than lateral sepals, but in ft-10 tsf-1, all sepals were smaller, similar in size and present in higher numbers (Fig. 3Cv-viii). By stage 5-6, sepals fully covered the floral meristem in Col-0 and single mutants, but in ft-10 tsf-1, inner whorls remained visible, with sepals that were smaller, variably shaped and present in higher numbers (Fig. 3Cix-xii). Older ft-10 tsf-1 flowers exhibited leaf-like features, including bifurcate-stellate trichomes on sepals and stipules on the pedicel, both of which were absent in Col-0 and tsf-1 (compare Fig. 3Di-iv with vii, viii). Measurement of pedicel width showed that ft-10, but not ft-10 tsf-1, mutants had wider pedicels than Col-0 (Fig. 3ii, iv, vi and viii, Fig. S2H). Conversely, 35S::FT plants overexpressing FT exhibited narrower pedicels (Fig. S2I, Table S2). Bifurcate trichomes were also observed on ft-10 sepals but less frequently (Fig. 3Dv, vi). Overall, SEM analysis again indicated that the flowers of ft-10 tsf-1 were more severely affected than those of fd-3 fdp-2.

FT transcription in leaves is induced by LDs, but its expression in inflorescences under both LDs and SDs prevents reversion to a vegetative state (Hiraoka et al., 2013; Liu et al., 2014; Muller-Xing et al., 2014). To determine whether ft-10 tsf-1 plants exhibit floral defects under SDs as well as under LDs, we examined their inflorescence and flower phenotypes under SDs. Flowering time for ft-10 tsf-1 and Col-0 was similar under SDs when measured by rosette leaf number (Fig. 4A). However, ft-10 tsf-1 plants produced approximately ten more cauline leaves than Col-0, indicating delayed transition from the I1 phase of inflorescence development to the formation of flowers during the I2 phase (Fig. 4A, Table S3), consistent with previous findings for ft mutants (Muller-Xing et al., 2014).

We evaluated flower morphology of the mutants and double mutants under SDs at two stages: when the first flower opened and 1 week later (Fig. 4B). In Col-0 and tsf-1 mutants, sepals enclosed flower buds, forming a regular spiral pattern in the inflorescence (Fig. 4B, Fig. S4A). In contrast, in ft-10 tsf-1 double mutants, and to a lesser extent ft-10 mutants, the flowers were not fully enclosed by sepals, exposing inner whorls and disrupting the regular spiral pattern (Fig. 4Bi-iv, Fig. S4A).

Floral organs in mature flowers were also counted under SDs (Fig. 4C). The flowers of ft-10 tsf-1 double mutants showed greater variability in organ number compared to Col-0 (Fig. 4Bv-viii). Approximately 60% of ft-10 tsf-1 mutant flowers had five sepals, and 100% had fewer than four petals, ranging from 0 to 2, whereas 45% of Col-0 flowers had five sepals and 45% had four petals (Fig. 4C, Tables 2, 3, Table S4). One week after the first flower opened on the inflorescence (see Materials and Methods), all Col-0 flowers had four sepals and four petals, but ft-10 and ft-10 tsf-1 mutants continued to show altered organ numbers (Fig. 4C, Table 3). Over 40% of ft-10 tsf-1 flowers had fewer than four sepals and petals (Fig. 4C). Moreover, older flowers of SD-grown ft-10 and ft-10 tsf-1 mutants exhibited leaf-like characteristics, such as bifurcated-stellate trichomes on sepals and stipules at the distal part of the pedicel (Fig. 4D). These results demonstrate that FT TSF are also required for floral development under SDs, where they do not control the timing of floral transition.

Both FD and FDP are expressed in floral meristems. Therefore, we measured the size and shape of floral meristems using SEM and confocal microscopy. SEM images suggested that floral meristems were larger in fd-3, fd-3 fdp-CRP2 and ft-10 tsf-1 double mutants compared to Col-0 (Figs 2C and 3C). To quantify these differences, we measured floral meristem dimensions using confocal microscopy. At stage 4, Col-0 floral meristems displayed an elliptical shape, with the lateral axis being wider and taller than the medial axis (Fig. 5A,B, Fig. S4B). However, in the double mutants, the medial axis was significantly wider than in Col-0 (80 µm versus 70 µm), but remained proportional to the lateral axis, effectively eliminating axis differences and resulting in a rounder shape (Fig. 5A,B, Fig. S4B). Additionally, the height of the floral meristems was also more similar between axes in the double mutants, though the lateral axis was reduced compared to Col-0, giving a flatter appearance (Fig. 5C,D, Fig. S4B). These results indicate that FD, FDP, FT and TSF play comparable roles in regulating floral meristem size and shape in stage 3-4 flowers, where FD and FDP are expressed.

FT is transcribed in the mature vasculature of leaves, stems, sepals, pedicels of flowers and siliques (Hiraoka et al., 2013; Liu et al., 2014; Muller-Xing et al., 2014), while FT protein is detected in the vascular tissue of leaves and in the SAM (Abe et al., 2019; Corbesier et al., 2007; Tamaki et al., 2007). To determine if FT is present in developing flowers at stages 3 to 5, as previously shown for FD and FDP, we analyzed FT::FT:GFP ft-7 and SUC2::FT:GFP ft-7 reporter lines (Corbesier et al., 2007). The FT::FT:GFP transgene contains the endogenous FT regulatory sequences, whereas SUC2::FT:GFP is specifically expressed in the phloem companion cells. Both transgenes complemented the ft-7 flowering-time delay (Corbesier et al., 2007) and were expressed in the vascular tissue of developing and mature pedicels (Fig. 6Ai-iii, Fig. S5i-iii). In addition, FT transcription was also detected in the pedicel using an FT::GUS fusion (Fig. S5iv). FT:GFP expressed from FT::FT:GFP was detected in young sepals (stage 4-5) and in the center of the floral meristem at stages 3-5 (Fig. 6Aiii-v). However, when FT was expressed from SUC2::FT:GFP, it was not detected in these regions (Fig. 6Ai,ii, Fig. S5i), suggesting that expression in the phloem companion cells does not account for FT protein accumulation in the floral meristem and developing floral organs. Additionally, the ft-7 mutant in the Ler background displayed a stronger floral phenotype than ft-10 in the Col-0 background, with only 30% and 40% of flowers producing four sepals and petals, respectively (Fig. 6B, Table 4, Table S5). This phenotype was fully restored in two independent FT::FT:GFP ft-7 lines, which showed a floral phenotype similar to wild-type Ler (Fig. 6B, Table S5), but was not corrected in SUC2::FT:GFP ft-7 plants. These results suggest that FT is expressed in the floral meristem and developing floral organs where the SUC2 promoter is not active, and that FT protein distribution in flowers overlaps that of FD and FDP.

The fd-3 fdp-CRP2 and ft-10 tsf-1 mutants have larger floral meristems than Col-0, and FD, FDP and FT proteins are present in floral meristems; therefore, the expression of genes that regulate floral meristem size was examined in these mutants. Floral meristem size is controlled by the CLV/WUS feedback loop, and determinacy of the floral meristem is conferred by repression of WUS transcription by AG (Lenhard et al., 2001; Lohmann et al., 2001).

To understand the contributions of FD, FDP and FT to the larger floral meristems observed in the mutants, we analyzed AG and WUS expression by in situ hybridization in fd-3, fd-3 fdp-CRP2, ft-10 and ft-10 tsf-1 mutants (Fig. 7A,B, Fig. S6). In Col-0, AG was expressed throughout the floral meristem from stage 3-4, but in fd fdp-CRP2 and ft-10 tsf-1 mutants, AG mRNA was present at the boundaries of the floral meristem but weak or absent in the center (Fig. 7Ai-iii, upper panels, Fig. S6A). By stage 6, AG mRNA was detected in the developing carpels and stamens of Col-0 flowers, and was reduced in flowers of ft-10, ft-10 tsf-1 and fd-3 fdp-CRP2 at the same stage, but remained similar to Col at this stage in the fd-3 single mutant (Fig. 7iv-viii). Therefore, FD, FDP and FT are required for AG expression in floral organs at this stage.

Next, we examined WUS mRNA by in situ hybridization at stages 6-8. In flowers of Col-0, WUS expression was reduced to a small region at stage 6 and was not detected in the center of stage 7 or 8 flowers (Fig. 7Bi,iv,vii). In stage 6 ft-10 tsf-1 or fd-3 fdp-CRP2 flowers, WUS expression was detected more strongly than in Col-0 (Fig. 7Bi-iii). Moreover, in flowers of fd-3, ft-10, fd-3 fdp-CRP2 and ft-10 tsf-1, WUS persisted for longer into stages 7 and 8 (Fig. 7Bi-ix, lower panels; Fig. S6B). This increased and extended pattern of WUS expression is consistent with the reduced AG expression and the enlarged floral meristems observed in these mutants.

SEP MADS-domain transcription factors activate AG transcription (Castillejo et al., 2005; Liu et al., 2009), and sep mutants are defective in floral organ number (Biewers, 2014) and produce flowers with leaf-like features (Ditta et al., 2004). Given the similarities between these phenotypes and the floral defects observed in ft-10 tsf-1 and fd-3 fdp-2 or fd-3 fdp-CRP2 mutants, we analyzed the expression levels of all four SEP genes (SEP1, SEP2, SEP3 and SEP4) by RT-qPCR using RNA extracted from inflorescences of ft-10, tsf-1, ft-10 tsf-1, fd-3, fdp-CRP2 and fd-3 fdp-CRP2 mutants (Fig. S7). In ft-10, tsf-1 and ft-10 tsf-1 mutants, the mRNA levels of SEP1, SEP2, SEP3 and SEP4 were reduced, with the strongest reductions observed for SEP2 and SEP3 mRNAs (fourfold and fourfold lower, respectively, in ft-10 tsf-1) (Fig. S7). Similarly, in fd-3 fdp-CRP2 double mutants, all four SEP gene mRNAs were also reduced, though less than in ft-10 tsf-1 (Fig. S7). To determine whether the defects observed in all four floral whorls could also be attributed to reduce expression of ABC genes within the floral organ identity model, we examined the expression of AP1, PI, AP3 and LFY (as a C class gene) by qRT-PCR across all mutant combinations. However, no significant differences in their expression were detected in ft-10 tsf-1 and fd-3 fdp-CRP2 double mutants (Fig. S7). These data indicate that, among the floral development genes tested, FT TSF and FD FDP predominantly regulate the expression of the four SEP genes, which likely contributes to the observed floral defects in the respective mutants.

The effect of FT TSF and FD FDP on the pattern of expression of SEP2 and SEP3 in the floral meristem was then examined. Notably, RNAscope analysis of mRNA colocalization confirmed that the spatial expression pattern of SEP3 mRNA broadly overlapped with that of FD throughout the floral meristem from stages 3 to 5 (Fig. 7C and Fig. S8A). SEP3 mRNA expression was then compared in Col-0, ft-10 tsf-1 and fd-3 fdp-CRP2 by in situ hybridization. As previously described (Mandel and Yanofsky, 1998), SEP3 mRNA was first detected in stage 2 floral buds of Col-0, and this early expression pattern was similar in Col-0, ft-10 tsf-1 and fd-3 fdp-CRP2 (Fig. 7B). Later, at stage 3, SEP3 mRNA was present throughout the floral meristem of Col-0, but in ft-10 tsf-1 was present only on the flanks of the floral meristem, similar to the pattern described earlier for AG, and in fd-3 fdp-CRP2 SEP3 mRNA was detected in a more restricted pattern than in Col-0 towards the apex of the floral meristem (Figs 1, 3 and 7D, Fig. S8B, upper panels). In situ hybridization also showed that SEP2 mRNA, which appears in Col-0 floral primordia at stage 2 (Savidge et al., 1995), was either absent or more narrowly expressed in ft-10 tsf-1 and fd-3 fdp-CRP2 mutants at the same stage (Fig. 7D; Fig. S8B, upper panels). At stage 3, SEP2 mRNA was detected in the entire floral meristem dome and sepal primordia of Col-0, but was unevenly distributed and restricted to the first three layers in the mutants (Fig. 7D, Fig. S8B, upper panels). We also analyzed the expression pattern of SEP3 in stage 4 and stage 6 flowers of Col-0 and different mutants by in situ hybridization (Fig. S8B). Compared with Col-0, SEP3 mRNA was reduced in the ft-10 single mutant to a similar extent as in ft-10 tsf-1 double mutants. By contrast, SEP3 expression in fd-3 mutants resembled that of Col-0, suggesting that both FD and FDP activate SEP3 (Fig. S8B). These experiments demonstrate that FT and FD FDP are required for the spatial expression pattern of SEP2 and SEP3 in the Col-0 floral meristem. In agreement with the RT-qPCR results, LFY mRNA showed a similar expression pattern by in situ hybridization in Col-0, ft-10 tsf-1 and fd-3 fdp-CRP2 (Fig. S8B).

The genomic binding sites of FD and FDP were identified by ChIP-seq (Collani et al., 2019; Romera-Branchat et al., 2020; Zhu et al., 2020). Analysis of the ChIP-seq data showed that SEP2 and SEP3 were bound by FD, suggesting that their activation by FD, as detected by in situ hybridization and RT-qPCR may be a direct effect. Similarly, SEP1 was also bound by FD (Fig. 8A).

The target genes of several MADS-domain TFs contributing to floral development have also been identified by ChIP-seq (Kaufmann et al., 2009, 2010; Wuest et al., 2012). Comparing FD targets with those of AP1, PI, AP3 and SEP3 revealed 74 genes bound by these four MADS-domain TFs and FD (Fig. 8B, Table S6). GO-term analysis of this subset showed enrichment in processes related to organ development, floral development, growth rate and floral organ number specification (Fig. 8C). Among the identified targets, SEP2, SEP3, LFY, AP1 and SOC1, which all encode TFs that contribute to floral development, were bound by FD and all of the MADS-domain proteins (Table S6). RT-qPCR analyses showed that, among these, only SEP2 and SEP3 were reduced in expression in both ft tsf and fd fdp genotypes (Fig. S7), although SOC1 was also reduced in ft tsf (Fig. 8E). Similarly, ASYMMETRIC LEAVES 1 (AS1) and AUXIN RESPONSE FACTOR 6 (ARF6), which act in parallel to negatively regulate floral organ growth (Tabata et al., 2010), were bound by FD (Fig. 8A) and all of the MADS-domain proteins. AS1 mRNA was reduced in ft mutants and ARF6 expression was reduced in ft tsf (Fig. 8E). These DNA-binding and RT-qPCR data suggest a direct role for FD in regulating organ growth.

This study demonstrates that the FAC components FD, FDP, FT and TSF contribute to floral development independently of their roles in promoting floral transition. These functions begin at stage 3 of floral development, and influence floral meristem shape and size, organ number and identity, and organ growth. FD, FDP, FT and TSF activity in the center of the floral meristem increases transcript levels of regulatory MADS-domain TFs, including all four SEP proteins and AG, and reduces WUS expression. This highlights that FT proteins and the FAC complex formed with FD are crucial not only for initiating floral transition and floral primordium identity but also for later stages of floral development.

Prior to floral development, FD and FDP are expressed in the vegetative and inflorescence SAM; we find that in the stage 1 flower they are expressed at the base of the primordium and at stage 1-2 on the abaxial side of the primordium (Romera-Branchat et al., 2020). Later during flower development, FD appears in the meristematic region of stage 2-3 flowers and FDP is expressed from stage 3 onwards in a domain mainly below that of FD, so that their patterns of expression in the floral meristem resemble those in the inflorescence meristem (Romera-Branchat et al., 2020). From stage 3 of floral development, their patterns of expression in the floral meristem partially overlap, but fd single mutants already show persistent WUS expression, suggesting that FD has a major effect in the floral meristem. Later, as floral organs develop, the domains of FD and FDP expression overlap in sepals, petals and pedicel tissues, consistent with their partially redundant effects on floral organ number. FT is transcribed in the mature phloem of several shoot tissues, including leaves, inflorescence stems, pedicels and sepals (Adrian et al., 2010; Kardailsky et al., 1999; Kobayashi et al., 1999; Liu et al., 2014; Muller-Xing et al., 2014). Moreover, to initiate floral transition, FT protein moves from the phloem tissue to the shoot apex, where it interacts with FD to promote flowering gene transcription (Abe et al., 2019; Corbesier et al., 2007; Jaeger and Wigge, 2007; Mathieu et al., 2007). Using an FT::FT:GFP transgenic plant (Corbesier et al., 2007), we detected FT protein in the meristematic region of stage 3-4 flowers. This result indicates that FT, FD and FDP are present in the same regions of the floral meristem, and therefore, in addition to its role in the SAM, the FAC is likely formed in the floral meristem to regulate meristem size and determinacy (Fig. 8E). However, this role of FT does not depend on expression of FT in the phloem, and therefore most likely not on the movement of FT protein from the vasculature, because a SUC2::FT:GFP transgene that strongly drives FT protein expression in the phloem and complements the late-flowering phenotype of ft mutants (Corbesier et al., 2007) does not complement the floral phenotype of ft. We propose, therefore, that FT is transcribed directly in developing flowers to produce FT protein that can interact with FD and FDP, and form the FAC to regulate floral development. FT mRNA has not so far been detected in flowers at this stage, but might be present at extremely low levels and/or transiently, and may not necessarily be transcribed in the cells in which FT protein is present, because it may move locally within the flower. FT expression in flowers might be photoperiod-independent, since similar floral defects were observed under both LDs and SDs. Photoperiod-independent expression of FT in the inflorescence has been detected previously (Liu et al., 2014; Muller-Xing et al., 2014), although its spatio-temporal pattern of expression in developing floral buds was not determined. Therefore, FT expression in flowers may be controlled by a different set of upstream regulators than has been defined in the phloem.

The fd and fdp mutations have a synergistic effect on floral organ development that, together with their overlapping expression patterns, argues for a partially redundant effect of the genes at this stage. The double-mutant phenotypes include variable floral organ number, floral homeotic changes and wider pedicels. Some features, such as bifurcated trichomes on sepals and the presence of stipules on the pedicel are characteristic of a reversion to a leaf-like vegetative phenotype (Ditta et al., 2004). The effects of ft tsf on floral organ number were stronger than those of fd fdp, as previously described for flowering-time control (Martignago et al., 2023; Romera-Branchat et al., 2020). This effect was particularly evident in the strong reduction in petal number in ft-10 tsf-1 mutants, which was not found in fd-3 fdp-2 or fd-3 fdp-CRP2 mutants, despite the FD and FDP proteins being expressed in young petal primordia. This observation suggests that other Group A bZIP TFs might also act during flower development and contribute additional layers of genetic redundancy with FD and FDP, as found in the SAM in the promotion of flowering (Martignago et al., 2023).

The phenotypic effects of fd fdp and ft tsf in floral development correlate with reduced expression of floral homeotic genes that are direct targets of FD, namely the class E MADS-domain SEP genes. Analysis by RT-qPCR in developing floral buds showed that the reduction in SEP gene transcription was stronger in ft tsf than in fd fdp, which correlates with the stronger effects on floral organ development discussed above. Moreover, SEP3 mRNA was reduced to a lesser extent than SEP2 and this might reveal layers of redundancy with other floral regulators in the regulation of SEP3, e.g. transcript levels of LFY and AP1 were not significantly reduced in the inflorescences of fd fdp or ft tsf and they have been shown to affect SEP gene expression (Kaufmann et al., 2010; Winter et al., 2011). SEP proteins are present in many different tetrameric MADS-domain protein complexes (de Folter et al., 2005; Honma and Goto, 2001), and reduced SEP gene expression not only results in homeotic changes, as predicted by its function in floral organ identity, but also the single or double combination of sep mutants affects floral organ number (Biewers, 2014) in a similar way to that observed in fd fdp and ft tsf mutants. How SEP genes regulate floral organ number is not yet clear, but might involve many MADS domain TF complexes. Thus, our study reveals a new layer of regulation of floral organ number by FT TSF, and to a lesser extent FD FDP, that is required for the formation of protein complexes including SEP proteins. Floral organ number is also more severely affected in ft tsf than in fd fdp mutants. Sepals of ft tsf mutants are smaller and grow asynchronously, exposing the floral meristem and inner whorls, suggesting a role for FT and TSF in sepal size robustness (Burda et al., 2024). Moreover, ARF6 and AS1 were downregulated in ft tsf, and their mutants display smaller organs in the first three whorls (Tabata et al., 2010), similar to the effect of ft tsf on sepals.

The floral meristems of fd-3 fdp-CRP2 and ft-10 tsf-1 double mutants were larger, flatter and more symmetrical than those of Col-0 flowers. The almost identical defects in floral meristem size and shape in ft tsf and fd fdp supports the idea that both classes of protein act in the FAC to control common processes during floral development. In situ hybridization showed that SEP2 and SEP3 mRNAs are reduced in the floral meristem in fd fdp and ft tsf mutants, at stage 3-4 when FD, FDP and FT are present. Floral meristem size is likely regulated independently of organ number, because fd fdp and ft tsf had altered floral meristems of similar size and shape but different effects on floral organ numbers. Moreover, in these FAC component double mutants, AG expression was strongly reduced in the center of the floral meristem, whereas on the flanks of the floral meristem it can be expressed more independently of the FAC. This experiment reveals a spatial pattern of regulation of AG transcription during the early stages of carpel development. The role of FD, FDP, FT and TSF in activating AG is likely to be indirect, probably through the SEP proteins, because direct binding to several SEP genes was detected but no direct binding of FD and FDP to AG was found (Collani et al., 2019; Romera-Branchat et al., 2020; Zhu et al., 2020), and SEP proteins have previously been shown to activate AG transcription (Castillejo et al., 2005; Liu et al., 2009). We propose therefore that the primary function of the FAC in the floral meristem is the activation of SEP genes, as shown schematically in Fig. 8E. This model could now be tested by exploring the effects of expressing SEP genes from FD regulatory sequences in the fd-3 fdp-CRP2 and ft-10 tsf-1 double mutants.

The larger floral meristem of fd fdp and ft tsf also correlated with higher levels of WUS expression persisting into later stages of floral development. This effect on WUS is probably an indirect result of reduced AG expression in the center of the floral meristem of ft tsf and fd fdp mutants, because repression of WUS transcription by AG has been shown to contribute to floral determinacy (Lenhard et al., 2001; Lohmann et al., 2001). Similarly, larger floral meristem size due to persistent WUS expression may contribute to the greater number of organs in inner whorls observed, e.g. in an ft soc1 agl24 svp background (Xi and Yu, 2009), although we did not observe more stamens in ft tsf double mutants, indicating that the increase in stamen number in ft soc1 agl24 svp is a synergistic effect of these mutations rather than being caused only by an increase in WUS. Nevertheless, ft tsf mutants show a larger floral meristem and a reduction in the total number of floral organs, suggesting that FT and TSF regulate floral organ number by mechanisms other than increasing floral meristem size.

The precise spatio-temporal expression pattern of FT in flowers and whether it is regulated by known floral regulators remain to be determined. This analysis is complicated by the extremely low level of FT mRNA and the possibility that it is only transiently expressed in small groups of cells. Moreover, the FT regulatory sequences are complex, and the observation that the floral phenotype is present in non-inductive SDs suggests that CONSTANS, which in leaves is activated only under LDs (Suarez-Lopez et al., 2001), may not be involved in the floral expression of FT, although this remains to be tested. Similarly, mutations in orthologs of FT have been described to cause floral phenotypes in other species (Molinero-Rosales et al., 2004), and, in these cases, whether the temporal and spatial patterns of expression we describe in Arabidopsis for FT, FD and FDP are conserved remains to be determined.

In the SAM, FT activity is antagonized by the closely related protein TFL1; in tfl1 mutants, FT and FD activity is increased at the SAM (Ahn et al., 2006; Cerise et al., 2023; Hanano and Goto, 2011; Zhu et al., 2020). TFL1 is not expressed in flowers (Bradley et al., 1997; Ratcliffe et al., 1999), suggesting that, in these tissues, FT can interact with FD and activate target genes independently of competition with TFL1. This effect would therefore be comparable to the determinate SAM phenotype of the tfl1 mutant because, in the mutant, TFL1 is absent at the tip of the SAM and therefore FT can carry out a comparable role to that in Col-0 flowers by activating genes required for floral organ development. Thus, the activity of FT and FD at the SAM in tfl1 mutants may be comparable to that we have described here for FT in the floral meristem.

In this study, Col-0 and Landsberg erecta (Ler) accessions were used as wild type. Seeds were stratified on soil for 2-4 days in the dark at 4°C and were then transferred to controlled environmental conditions at 20-21°C under long-day conditions (LDs) (16 h light/8 h dark) or short-day conditions (SDs) (8 h light/16 h dark) under a photosynthetic photon flux density of ∼180 μmol s⁻¹ m⁻².

All mutant and transgenic lines used in this study: fdp-2, fd-3, fdp-CRP2, fd-3 fdp-CRP2, fd-3 fdp-2, ft-10, tsf-1, ft-10 tsf-1, ft-10 tsf-1, fdp-2 fd-3, ft-7, pFD::VENUS:FD in fdp-2 fd-3, pFD::VENUS::FD in fd-3, pFDP::VENUS:FDP in fdp-2, pFT::GUS, pFD:mCHERRY:FD in fd-3, GAS1:FT, GAS1::FT in ft-10 tsf-1, SEP3::GFP, SUC2::FT:GFP in ft-7 and FT::FT:GFP in ft-7 have been previously described (Adrian et al., 2010; Andres et al., 2015; Corbesier et al., 2007; Jang et al., 2009; Martignago et al., 2023; Romera-Branchat et al., 2020; Torti et al., 2012; Urbanus et al., 2009).

To score floral organs under LDs, the floral organs of all open flowers at stage 12-13 (Smyth et al., 1990) on the primary inflorescence were counted on 3-4 consecutive days until between 25 and 150 flowers were scored in total from at least five plants per genotype. Each experiment was performed independently at least twice.

To score floral organs under SDs, measurements were performed at two time points to capture any effects associated with the age of the inflorescence. First, the floral organs were counted for the first flower to open on the primary inflorescence on a total of 19 plants in Col-0, 12 in tsf-1, 18 in ft-10 and 13 in ft-10 tsf-1. One week later, a second measurement was performed by counting the floral organs of all flowers at stage 12-13 using a total of 45 flowers for Col-0, 41 in tsf-1, 73 for ft-10 and 47 for ft-10 tsf-1.

For each genotype, flowers at anthesis (stage 13 according to Smyth et al., 1990) were harvested from the primary shoot from ∼15 plants and observed under a binocular microscope. Pedicel width was manually measured from flower images using ImageJ software (Schindelin et al., 2012). The width of each pedicel was measured twice from the same image and each experiment was independently performed twice.

Inflorescences were dissected manually after plants had bolted and fixed overnight in FAA [50% (v/v) ethanol, 5% (v/v) acetic acid, 1.85% (w/v) formaldehyde]. Samples were dehydrated through an ethanol series and were critical-point dried in liquid CO₂. Leaf primordia on the dried shoot apex samples were removed under a binocular microscope. The prepared samples were mounted on stubs, sputter-coated with gold and palladium and subjected to high-resolution scanning electron microscopy (Zeiss, Supra 55 VP FEG) with a Gatan Alto 2500 cryo system. More than five shoot apices for each genotype were analyzed.

Shoot meristems of pFD::VENUS::FD in fd-3 and pFDP::VENUS:FDP in fdp-2 were dissected manually and incubated overnight in 4% (w/v) paraformaldehyde (Sigma-Aldrich) before being treated with ClearSee for 2–4 days as previously described (Kurihara et al., 2015). Samples were stained with SCRI Renaissance 2200 dye (Musielak et al., 2015) for a minimum of 4 h to overnight, and were imaged using a Zeiss LSM780 confocal laser scanning microscope. Minor modifications were performed when FT:GFP protein was observed from ft-7 mutant lines carrying SUC2::FT:GFP or FT::FT:GFP; in particular, fixation was performed only for 1 h and samples were incubated in ClearSee solution overnight. The settings used to detect VENUS fluorescent protein have been described previously (Romera-Branchat et al., 2020). To detect mCHERRY, the excitation wavelength was 561 nm and detection wavelength was 570-650 nm.

GUS staining was performed as described previously (Sessions et al., 1999) except that samples were not treated with acetone. After overnight incubation in GUS-staining solution, samples were dehydrated through an ethanol series and mounted on slides in Hoyer's solution (Hedhly et al., 2018).

To analyze the mRNA localization of FD and FDP genes, plants were grown for 3 weeks in SDs and were then transferred to LDs for 3, 5 or 7 days. To detect SEP2, SEP3, WUS, AG and LFY mRNAs, Col-0 plants were grown for 19LD or 23LD, fd-3 fdp-CRP2 mutants were grown for 27LD or 31LD, and ft-10 tsf-1 plants were grown for 37LD or 45LD. The samples were inflorescence apices harvested at ZT3. The in situ hybridizations were performed as described previously (Torti et al., 2012). The probes used to amplify SEP2, SEP3, WUS, AG and LFY genes are listed in Table S7. Probes for FD and FDP were described previously (Romera-Branchat et al., 2020).

The RNAscope assay was conducted following the RNAscope Multiplex Fluorescent Assay v2 protocol provided by ACDBio (materials and method are available at https://acdbio.com/rnascope-multiplex-fluorescent-v2-assay). Briefly, formalin-fixed, paraffin-embedded (FFPE) tissue samples were used for analyses. Probes for FD and SEP3 were custom designed by ACDBio and were assigned to channels C1 and C3, respectively, with the following catalog numbers: 1307011-C1 (FD) 15 pair specific probe, target region 230-1064 of NCBI Reference sequence NM_119756.5; and 1569931-C3 (SEP3)16 pair specific probe, target region 392-1268 of NM_180622.3. RNAscope 3-plex Negative Control Probe (catalog number 320871) was used as a negative control. The protocol was followed exactly except that probes were hybridized overnight at 40°C.

To visualize the targets, TSA Plus fluorophores (diluted in TSA buffer from ACDBio) were applied as follows: TSA Vivid 520 (323271, diluted 1:2500) for C3 and TSA Vivid 570 (323272, diluted 1:1500) for C1. Additionally, Renaissance (0.1% v/v in PBS, 5 min at room temperature) was used to stain the cell walls.

Confocal images were captured using a Zeiss LSM 880 confocal microscope. The Renaissance signal was detected at 410-503 nm with an excitation wavelength of 405 nm. The filter settings for FITC and Cy3 were used for the TSA Vivid Fluorophore 520 and 570 separately.

Total RNA was extracted with the RNeasy kit (QIAGEN) from the whole primary inflorescence (containing between 2 mm and 3 mm of stem) 2-4 days after the opening of the first flower of plants grown under LDs and was treated with DNA-free DNase (Ambion). For reverse transcription, 2 mg RNA was used following the manufacturer's instructions (Invitrogen). For each replicate, 4-6 inflorescences were harvested. To quantify the expression level of floral genes, cDNAs were synthesized with SuperScript IV (Invitrogen) and qRT-PCR was performed on the Bio-Rad CFX384 using iQ SYBR Green Supermix (Bio-Rad). Each sample was run in technical triplicate and three biological replicates were performed per genotype. For the quantification, the delta-delta CT method was used, the EIF4 gene was used as a reference gene to normalize expression. Primers used for qRT-PCR are listed in Table S7.

Shoot apices of Col-0, fd-3 fdp-CRP2 and ft-10 tsf-1 plants were harvested at 18LD, 27LD and 34LD, respectively, to sample the first ten flowers initiated by each genotype. Samples were prepared for confocal laser scanning microscopy, as described in the confocal analysis section. All images were analyzed using Fiji software (Schindelin et al., 2012). To obtain images, flowers at stage 4 were identified, reoriented to an upright position, and the height and width of the floral meristem was measured after identifying the medial and lateral axes of sepal formation, according to Smyth et al. (1990). Statistical analyses were performed using two-way ANOVA followed by Tukey's multiple comparison test (GraphPad, www.graphpad.com).

A list of MADS-domain-protein target genes wase obtained from previously published data: AP1 (Kaufmann et al., 2010), SEP3 (Kaufmann et al., 2009), and PI and AP3 (Wuest et al., 2012). A list of bound regions in the genome was obtained using the CSAR pipeline (van Mourik et al., 2015). To assign the corresponding target gene, a genomic interval from 3 kb upstream the start codon to 1 kb downstream the stop codon was considered. The list of FD targets was the same as previously identified (Romera-Branchat et al., 2020).

For GO analysis, the BiNGO tool (Maere et al., 2005) was used, and the results were visualized using the Cytoscape platform. The color of the circles represents the P-value and the size represents the over-representation or under-representation of GO categories according to a hypergeometric test [false discovery rate (FDR)<0.05]. For the comparison of genes bound by FD with those bound by MADS-domain transcription factors (AP1, PI, AP3 and SEP3), Fisher's one-sided exact test was used.

For graph representation, GraphPad PRISM software was used. When several data points were compared, an ANOVA and Kruskal–Wallis one-way ANOVA on Ranks was used. Different letters indicate statistically significant differences at P≤0.05. For simpler two-way comparisons, a non-parametric t-test was used. Asterisks indicate differences among two genotypes.

We thank Ton Timmers from the Central Microscopy service (MPIPZ) for help with microscopy and the members of the Coupland group, especially John Chandler, for reading and correcting the manuscript. We also thank Sara Cioffi for her assistance in scoring floral phenotypes, and Sandra Biewers and Brendan Davies (University of Leeds, UK) for sharing Sandra's PhD thesis.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.204241.reviewer-comments.pdf