Inflorescence Meristem Fate Is Dependent on Seed Development and FRUITFULL in Arabidopsis thaliana

After a vegetative phase, plants initiate the floral transition in response to both environmental and endogenous cues to optimize reproductive success. During this process, the vegetative shoot apical meristem (SAM), which was producing leaves and branches, becomes an inflorescence SAM and starts producing flowers. Inflorescences can be classified in two main categories, depending on the fate of the inflorescence meristem: determinate or indeterminate. In determinate inflorescences, the SAM differentiates directly, or after the production of a certain number of flowers, into a flower, while in indeterminate inflorescences the SAM remains indeterminate and produces continuously new flowers. Even though indeterminate inflorescences have an undifferentiated SAM, the number of flowers produced by a plant is not indefinite and is characteristic of each species, indicating that it is under genetic control. In Arabidopsis thaliana and other species with indeterminate inflorescences, the end of flower production occurs by a regulated proliferative arrest of inflorescence meristems on all reproductive branches that is reminiscent of a state of induced dormancy and does not involve the determination of the SAM. This process is controlled genetically by the FRUITFULL-APETALA2 (FUL-AP2) pathway and by a correlative control exerted by the seeds through a mechanism not well understood yet. In the absence of seeds, meristem proliferative arrest does not occur, and the SAM remains actively producing flowers until it becomes determinate, differentiating into a terminal floral structure. Here we show that the indeterminate growth habit of Arabidopsis inflorescences is a facultative condition imposed by the meristematic arrest directed by FUL and the correlative signal of seeds. The terminal differentiation of the SAM when seed production is absent correlates with the induction of AGAMOUS expression in the SAM. Moreover, terminal flower formation is strictly dependent on the activity of FUL, as it was never observed in ful mutants, regardless of the fertility of the plant or the presence/absence of the AG repression exerted by APETALA2 related factors.

After a vegetative phase, plants initiate the floral transition in response to both environmental and endogenous cues to optimize reproductive success. During this process, the vegetative shoot apical meristem (SAM), which was producing leaves and branches, becomes an inflorescence SAM and starts producing flowers. Inflorescences can be classified in two main categories, depending on the fate of the inflorescence meristem: determinate or indeterminate. In determinate inflorescences, the SAM differentiates directly, or after the production of a certain number of flowers, into a flower, while in indeterminate inflorescences the SAM remains indeterminate and produces continuously new flowers. Even though indeterminate inflorescences have an undifferentiated SAM, the number of flowers produced by a plant is not indefinite and is characteristic of each species, indicating that it is under genetic control. In Arabidopsis thaliana and other species with indeterminate inflorescences, the end of flower production occurs by a regulated proliferative arrest of inflorescence meristems on all reproductive branches that is reminiscent of a state of induced dormancy and does not involve the determination of the SAM. This process is controlled genetically by the FRUITFULL-APETALA2 (FUL-AP2) pathway and by a correlative control exerted by the seeds through a mechanism not well understood yet. In the absence of seeds, meristem proliferative arrest does not occur, and the SAM remains actively producing flowers until it becomes determinate, differentiating into a terminal floral structure. Here we show that the indeterminate growth habit of Arabidopsis inflorescences is a facultative condition imposed by the meristematic arrest directed by FUL and the correlative signal of seeds. The terminal differentiation of the SAM when seed production is absent correlates with the induction of AGAMOUS expression in the SAM. Moreover, terminal flower formation is strictly dependent on the activity of FUL, as it was never observed in ful mutants, regardless of the fertility of the plant or the presence/absence of the AG repression exerted by APETALA2 related factors.

InTRODUcTIOn
For most plants, reproductive success depends on the ability to produce seeds that ensure the perpetuation of the species. Seed production is related to the number of flowers produced by the plant during the reproductive phase, and therefore, dependent on the activity of the inflorescence meristems that produce the flowers. Inflorescences have been classified in two major categories based on the fate of the inflorescence meristem: determinate or indeterminate. Determinate inflorescences are those where the shoot apical meristem (SAM) differentiates directly, or after the production of a certain number of flowers, into a flower. Thus, the end of the reproductive phase in determinate inflorescences is established by the final differentiation of the inflorescence SAM, determining the final number of flowers and seeds produced per shoot. On the other hand, indeterminate inflorescences are those where the SAM remains undifferentiated and produces continuously new flowers until the end of the reproductive phase. Interestingly, the length of the reproductive phase and the number of flowers produced by indeterminate inflorescences is finite, despite the undifferentiated nature of the SAM, which does not produce any terminal structure. Moreover, in indeterminate inflorescences, the number of flowers produced before SAM arrest is usually distinctive for each species and/or ecotype, suggesting a genetic control of the length of the reproductive phase.
Despite of the developmental differences between them, determinate and indeterminate inflorescences can be found in multiple plant families, as well as in plants with different growth habits, like in monocarpic (reproducing only once during their life cycle) or polycarpic plants (alternating reproductive and vegetative phases). In monocarpic plants with indeterminate inflorescences, the end of the reproductive phase has been associated with a process named Global Proliferative Arrest (GPA). GPA has been well characterized for the model species Arabidopsis thaliana (Hensel et al., 1994). During GPA, after the production of a determined number of flowers, the SAM arrests its growth, and all floral buds, up to the last non-pollinated flowers, do not develop further. In a short period of time, all active meristems in the plant undergo the same process. At this point, fruit filling and seed maturation is completed in all fertilized flowers and then, the plant senesces and dies. Although the end of the flowering phase might be assumed as a default process, linked to meristem exhaustion and plant senescence, classical studies indicate that it is a regulated process, preceding senescence of reproductive branches in polycarpic species or of the whole plant in annual species (Murneek 1926;Leopold et al., 1959;Lindoo and Nooden, 1977;Hensel et al., 1994;Wilson, 1997;Noodén et al., 2004). It has been proposed that proliferative arrest could be related with the proper allocation of nutrients to the developing seeds, and, thus, the establishment of strong source-sink relationships between the seeds and the inflorescence meristem could restrict plant growth and trigger the end of flowering (Sinclair and de Wit, 1975;Kelly and Davies, 1988). In agreement with this, the major factor controlling the end of flowering is seed production, as proven by the extended flowering period of plants with strongly reduced fertility (Murneek, 1926;Leopold et al., 1959;Lindoo and Nooden, 1977;Hensel et al., 1994;Wilson 1997;Noodén et al., 2004). The mechanism of this correlative control exerted by the seeds is still unknown (Walker and Bennett, 2018), but it has been shown that it modifies the SAM activity, inducing a state reminiscent of meristem dormancy, with low mitotic activity, a reduction of reactive oxygen species, and accumulation of abscisic acid response genes (Wuest et al., 2016).
In addition to the correlative control of seeds, the end of the reproductive phase in Arabidopsis indeterminate inflorescences is also controlled genetically by a recently described pathway likely dependent on the age of the inflorescence (Balanza et al., 2018). Briefly, APETALA2 (AP2) and other related factors of the same family sustain the expression of WUSCHEL (WUS), a key gene involved in stem cell maintenance, in the inflorescence meristem (Laux et al., 1996;Mayer et al., 1998). FRUITFULL (FUL), a MADSbox transcription factor involved in multiple developmental processes and strongly expressed in the inflorescence meristem, directly represses the expression of AP2 and AP2-like genes in this domain (Gu et al., 1998;Ferrandiz et al., 2000a;Ferrandiz et al., 2000b;Shikata et al., 2009;Wang et al., 2009;Yamaguchi et al., 2009;Balanza et al., 2014;Bemer et al., 2017). The genes in the AP2 clade are also negatively regulated by the action of the miR172 in an age dependent way (Aukerman and Sakai, 2003;Chen, 2004;Wang et al., 2009;Wu et al., 2009). Based on the phenotypes of the different mutants, we previously proposed that the combined action of miR172 and FUL, increasingly accumulated through inflorescence development, would lead to decreasing levels of AP2 and AP2-like factors in the SAM, eventually unable to maintain WUS activity. Accordingly, ful mutants and AP2 alleles resistant to the action of miR172 delay the end of the flowering phase, resulting in an increased flower production (Balanza et al., 2018).
Interestingly, in Arabidopsis sterile mutants, or in wild type plants where flowers are removed, the end of the reproductive phase differs from that observed in fertile plants. As mentioned above, sterile mutants produce more flowers than fertile plants, and instead of ending flower production with meristem arrest, the inflorescence meristem of sterile mutants become determinate producing a terminal flower of carpelar nature (Chaudhury 1993;Schultz and Haughn, 1993;Hensel et al., 1994;Modrusan et al., 1994). The nature of this shift in the inflorescence SAM fate is still unknown, although it could be interpreted as a differentiation of the SAM into a floral meristem, similar to that observed in determinate inflorescences. The major factors that control which meristems become flowers and which remain as undifferentiated shoots have been studied in model species but also in many other plants. In the Arabidopsis indeterminate inflorescence, once floral transition has taken place, the lateral meristems produced by the SAM acquire floral identity. The main factors that control the specification of floral fate in these meristems are LEAFY (LFY) and APETALA1 (AP1), which confer floral meristem identity upstream of floral organ identity factors (Mandel et al., 1992;Weigel et al., 1992;Weigel and Meyerowitz, 1993). The SAM remains indeterminate by restricting the expression of these floral promoting factors, which are excluded by the activity in this domain of TERMINAL FLOWER1 (Shannon and Meeks-Wagner, 1991;Liljegren et al., 1999;Ratcliffe et al., 1999), homologous to phosphatidylethanolamine binding proteins and a member of a small gene family, which includes FLOWERING LOCUS T (FT), a key regulator of flowering time regarded as the florigen (Wigge et al., 2005). The antagonistic interaction between TFL1 and AP1/LFY is crucial to maintain the indeterminacy of the Arabidopsis inflorescence. In Arabidopsis tfl1 loss of function mutants, the inflorescence meristem ectopically expresses AP1 and LFY differentiating prematurely into a flower, and thus, the inflorescence becomes determinate (Shannon and Meeks-Wagner, 1991;Liljegren et al., 1999;Ratcliffe et al., 1999). Likewise, plants overexpressing AP1 or LFY, and even some of their downstream floral organ identity factors, like AGAMOUS (AG), also cause the conversion of the SAM into a flower (Mizukami and Ma, 1992;Mandel and Yanofsky, 1995b;Blazquez et al., 1997). The basic TFL/AP1/LFY module has been characterized in many other species and shown to be largely conserved in angiosperms, and variations in its configuration appear to correlate well with determinate/indeterminate growth habits (Pnueli et al., 1998;Liljegren et al., 1999;Ratcliffe et al., 1999;Benlloch et al., 2007;Mimida et al., 2011;Mohamed et al., 2010;Imamura et al., 2011;Iwata et al., 2012).
Here, we show that the FUL-AP2 pathway that regulates the length of the flowering phase in fertile plants is also controlling the fate of the inflorescence meristem in sterile plants. We show that the formation of the terminal flower in the absence of seeds associates with the ectopic expression in the SAM of the AG gene. Loss of AP2 function leads to sterile plants that produce an early differentiation of the inflorescence meristem into a terminal flower, suggesting that AP2 participates in the repression of AG in the SAM. Finally, we also show that the ectopic activation of AG in the SAM is strictly FUL-dependent, as ful mutants suppress the terminal flower formation in the absence of seeds, even in 35S:miR172 background where AP2-like factors are not active.

Meristem Fate Is Dependent on the Presence of Seeds and FUL activity
It has been described that seed development directly impacts the end of the flowering phase in Arabidopsis thaliana inducing the SAM arrest. In the absence of seeds, the SAM remains active for longer and the number of flowers produced is increased. In these conditions, the SAM will produce new flowers until it differentiates into a terminal floral structure, resembling the behavior of a determinate inflorescence, and also of mutants where inflorescence meristem identity is compromised (Hensel et al., 1994;Ohshima et al., 1997;Bradley et al., 1997). As the FUL-AP2 pathway genetically controls the length of the flowering phase, we decided to assess if the FUL-AP2 module had also a role in the determination of the fate of the inflorescence meristem and the formation of the terminal structure.
For this purpose, ful mutants were grown in the absence of seed production, by continuously removing all the flowers produced by the SAM at anthesis. As described previously, wild type plants and ful mutants responded to the pruning by increasing the number of flowers produced by the SAM (Balanza et al., 2018) ( Figure 1A). Surprisingly, while the wild type SAM eventually differentiated into a terminal flower (Figures 1B, C), the SAM of ful mutants remained undifferentiated, generating flowers until the death of the plant (Figures 1D, E), without producing the terminal structure observed in the wild type.
To confirm that the observed phenotype in manually pruned ful plants was independent of the treatment, we decided to characterize inflorescence meristem fate in mutant combinations of ful with unrelated mutations previously described that caused sterility or a severe reduction of fertility, such as crabs claw (crc) (Bowman and Smyth, 1999), spatula (spt) (Alvarez and Smyth, 1999), the quadruple ngatha (nga) mutant (Trigueros et al., 2009), or pistillata (pi) (Hill and Lord, 1989). While all the sterile mutants tested ended flowering with the formation of the typical terminal flower (Figures 2B-E), in combination with ful they remained active for longer without SAM differentiation, similarly to what was observed in the ful pruning experiment, and ending flowering as the ful single mutant (Figures 2A,  F-J). These phenotypes suggested a role for FUL in controlling inflorescence meristem fate, where it could act by inducing the formation of the terminal structure observed in plants that did not produce seeds.

AG Is expressed ectopically in the Inflorescence Shoot apical Meristem Under Sterile conditions
AG is a homeotic MADS-domain transcription factor that confers carpel identity during flower development (Yanofsky et al., 1990;Bowman et al., 1991). It has been described that the constitutive expression of AG induces the differentiation of the inflorescence meristem into a terminal structure after the production of a reduced number of flowering nodes (Mizukami and Ma, 1992). As the terminal structure observed in sterile plants was mainly composed by carpel-like structures, we decided to check AG expression throughout inflorescence development in the presence (untreated wild type plants) or absence of seed development (pruned plants) by monitoring the activity of an AG::GUS reporter previously generated and characterized (Sieburth and Meyerowitz, 1997). As expected, in untreated wild type plants no β-glucuronidase (GUS) signal was observed in the SAM at any inflorescence developmental stage where the meristem was proliferative and producing new flowers (Supplementary Figure 1), while the signal was clearly detected in the center of the floral meristems. The lack of AG::GUS expression in the SAM was also evident in the arrested inflorescence meristems at the end of the flowering period 3-4 weeks after bolting, when no further initiation or development of floral buds was taking place (Figure 3A). In pruned plants, where seed production was avoided, the GUS signal pattern was identical to that of the fertile control plants during the proliferative phase of inflorescence development (Supplementary Figure 1), being only present in the center of floral buds but absent in the SAM. However, the AG::GUS reporter activity was clearly detected in the SAM of pruned plants at late stages of inflorescence development, preceding the formation of the terminal flower. In pruned plants, AG::GUS FIgURe 2 | Terminal flower formation is never observed in ful mutants. ful mutants show an extended flowering phase with a morphologically distinct inflorescence meristem always present that remains active (a, F). Mutants with strongly reduced fertility show inflorescences that terminate with the formation of a terminal flower as shown for the quadruple nga (B), spt-2 (c), crc-1 (D), or pi (e). When these sterile mutants are combined with ful, the formation of the terminal structure is suppressed and meristems remain active until the death of the plant (g, h, I, J).
December 2019 | Volume 10 | Article 1622 signal started to be detected in the periphery of the SAM at 6-7 weeks after bolting (Figures 3D, E) to later extend to the entire SAM (Figures 3F, G) until its differentiation in the terminal structure ( Figure 3H). Our results indicate that the terminal carpelar structure observed at the end of the flowering phase in the absence of seed production is associated with the ectopic expression of AG in the SAM. As the ful mutant does not undergo SAM differentiation into a terminal structure in the absence of seed production, we decided to analyze the activity of the AG::GUS reporter in ful plants where flowers were continuously removed. In agreement with the phenotypes observed when this treatment was applied in ful mutants, the AG::GUS signal was only detected in the floral meristems in both ful AG::GUS untreated (fertile) (Figures 3B, C) and pruned (sterile) plants (Figures 3I, J), but never in the SAM, even 2 or 3 weeks after the formation of the terminal flowers in the pruned wild type control plants (Figures 3I, J) (8-9 weeks after bolting). These results strongly suggested that the activation of AG in the SAM in the absence of seed development was FUL-dependent.

AP2-Like genes Repress AG in the Shoot apical Meristem
In the floral meristem, AP2 is a classical repressor of AG acting in the two outer whorls of the flower Drews et al., 1991). In ap2 loss-of-function mutants, AG becomes ectopically expressed in the external whorls of the flower producing the conversion of sepals to carpels and petals to staminoid structures. As FUL is a known repressor of AP2 and other AP2-like genes in the SAM (Balanza et al., 2018), it could be expected that, in the ful mutants, increased AP2 and AP2-like gene expression in the SAM could prevent AG upregulation in the inflorescence meristem at late stages in sterile plants. To test this hypothesis, we first characterized the end of the flowering phase of an ap2 mutant compared to wild type. Since ap2-12 is a sterile mutant, we compared it with a wild type control plant where seed production was avoided by pruning the developing flowers. ap2-12 mutants produced far fewer flowers than the pruned control (Balanza et al., 2018), but both ended flowering with the formation of a terminal flower (Figures 4A, B, G), as expected for sterile backgrounds. Then, we wondered if the early determination of the inflorescence meristem observed in ap2-12 could be related to an early activation of AG in the SAM. When we analyzed the AG::GUS reporter line in the ap2 background we observed that, as in wild type pruned plants, the GUS signal was ectopically expressed in the SAM (Figures 3D-H, Figures 4C,  D). Our results indicated that AP2 could mediate the repression of AG in this domain. To assess if the rest of the AP2-like genes also contributed to the repression of AG in the inflorescence meristem, we also characterized inflorescence development and inflorescence meristem fate in plants overexpressing miR172, where all genes in the AP2 family are simultaneously downregulated. All 35S::miR172 plants showed a characteristic early flowering phenotype, and most of them also showed the sepal-to-carpel homeotic conversions observed during flower development in the ap2 mutant (Yant et al., 2010). Interestingly, a small fraction of the plants (around 20%) did not exhibit the homeotic transformations associated to the ap2 floral phenotype, developing normal flowers that were fully fertile. As observed for the single ap2-12 mutant, both phenotypic categories of 35S::miR172 plants had inflorescences that produced a terminal flower (Figures 4E, F), although inflorescence meristem determination occurred earlier than in the single ap2-12 mutant, after the production of a further reduced number of flowers (Balanza et al., 2018) (Figure 4G). The earlier differentiation of the terminal flower in the 35S::miR172 plants, even in the presence of seeds ( Figure 4F) indicates that not only AP2, but also the rest of AP2-like genes contribute to prevent the differentiation of the SAM into a terminal structure, possibly by jointly repressing AG in the SAM. Then, the increased levels of AP2-like gene expression in ful mutants (Balanza et al., 2018) could explain the suppression of the terminal flower differentiation in plants where seed production was prevented.

FUL Is Required for the ectopic expression of AG in the Shoot apical Meristem
If the suppression of the terminal flower observed in ful mutants in the absence of seed production was mediated by the upregulation of AP2 and AP2-like genes in the SAM, in a double ful ap2 mutant we could expect the determination of the inflorescence meristem into a terminal structure. Surprisingly, the double ful ap2 mutant, despite being sterile, ended the flowering phase without the formation of the terminal flower ( Figures  5A, B). As in this double mutant the expression of other AP2like genes still was elevated (Balanza et al, 2018), it was possible that this increased expression could be sufficient to suppress the formation of the terminal structure. To test this hypothesis, we checked inflorescence meristem fate in a ful 35S::miR172 line. In this genetic combination, the levels of all the AP2-like proteins are downregulated by the overexpression of miR172 even in the absence of FUL activity. As described previously, the ful 35S::miR172 plants produced a similar number of flowers to 35S::miR172 plants, suppressing the increased flowering period and delayed proliferative arrest observed in the single ful mutant, and therefore indicating that the FUL effect on the duration of the flowering phase is mediated by AP2-like factors (Balanza et al., 2018). Surprisingly, the differentiation of the inflorescence meristem into a terminal flower that occurred in 35S::miR172 plants was never observed in ful 35S::miR172 lines ( Figures  5C, D). These results indicate that to promote the formation of the terminal floral structure and probably the ectopic AG expression in the SAM, it is necessary the downregulation of AP2-like genes and the presence of FUL in the inflorescence meristem at later stages of inflorescence development.
The requirement of FUL to induce the differentiation of the inflorescence meristem into a terminal flower suggested that FUL could work as an AG activator. To confirm the ability of FUL to activate AG, we analyzed the expression of AG in a FUL::FUL:VP16 line, where the strong transcriptional activation domain of the herpes virus protein VP16 was fused to the FUL protein. The FUL:VP16 chimeric protein should cause the upregulation of FUL direct targets, overcoming other regulatory effects, as for example the possible repression exerted by other factors or the effect of other FUL interacting proteins in its transcriptional output (Bemer et al., 2017;Balanza et al., 2018). Interestingly, the FUL::FUL:VP16 line ends flowering very early, with the formation of a terminal floral structure (Balanza et al., 2018). In situ RNA hybridization indicates that AG was expressed ectopically in the periphery of the SAM of the FUL::FUL:VP16 line, just before the formation of the terminal flower (Figure 6). Thus, FUL:VP16 was able to bypass the negative regulation exerted by AP2-like genes at early inflorescence development, suggesting that FUL could activate directly AG expression. In agreement with this observation, when we searched the results of a FUL chromatin immunoprecipitation sequencing data set available in a public repository (NCBI-GEO-DataSet GSE108455, provided by van Mourik, Muiño, Smaczniak, Bemer, Chen, Angenent, and Kaufmann) we found that FUL was able to bind to the AG genomic region on two different regions, one centered at approximately 1.5 kb upstream the START codon of the AG coding sequence, and another in the distal portion of the  mutations (a, B). In the 35S::miR172 plants, where the levels all the AP2-like genes are reduced and the terminal flower is formed very early in inflorescence development, the ful mutation also suppresses its formation (c, D). second regulatory intron (Supplementary Figure 2), which has been described to contain key regulatory elements controlling AG (Sieburth and Meyerowitz 1997;Deyholos and Sieburth, 2000;Hong et al., 2003).

DIScUSSIOn
The regulated arrest of the flowering phase is still a not wellunderstood developmental process. In plants with indeterminate inflorescences, seed development is a major factor directing the end of flowering, exerting a negative control over SAM activity. Thus, in the absence of seeds, the length of the inflorescence phase is increased until, at least in some species like Arabidopsis, the inflorescence meristem fate changes from indeterminate to determinate (Hensel et al., 1994). In Arabidopsis, a genetic mechanism has been also described to control SAM activity and the length of the flowering period in parallel with the correlative control mediated by seeds: the FUL-AP2 pathway. Our results indicate that FUL and AP2-like genes are also key regulators of the inflorescence meristem fate: FUL appears to induce AG activation in the SAM both directly and indirectly, through negative regulation of the AG repressors AP2 and AP2-like genes in the SAM. Our results can be easily integrated into the previously proposed model controlling the proliferative arrest of the inflorescence meristems and the end of the flowering phase in Arabidopsis (Balanza et al., 2018) (Figure 7). When the arrestinductive seed effect is absent (as in sterile mutants or by pruning of flowers as they are produced), the inflorescence meristem activity is extended in time, forcing the SAM to produce extra flowers. In this scenario, increasing activity of FUL in the inflorescence meristem should simultaneously cause a direct activation of AG and a further reduction of the AP2 repressive activity on AG in the SAM, which eventually would allow AG expression in the inflorescence meristem and the formation of the terminal flower. This model is in agreement with the observed phenotypes of related mutant backgrounds. Thus, the downregulation of AP2 and AP2-like genes in 35S::miR172 lines would cause the early differentiation of the terminal structure, even in the presence of seeds, likely caused by the derepression of AG in the SAM. Likewise, in the FUL::FUL:VP16 line, despite the high levels of AP2 (Balanza et al., 2018), the strong upregulation of AG in the SAM caused by FUL:VP16 activity would explain the determination of the meristem and the production of a terminal structure consisting of stamens and carpels.
Remarkably, FUL appears to be essential to promote inflorescence meristem determination, since no terminal flowers were formed in ful 35S::miR172 mutants or when the ful mutation was introduced in lines of very reduced fertility. Together with the likely direct effect on AG activation reported in this study, FUL has also been shown to directly upregulate the floral meristem identity gene LFY (Balanza et al., 2014). Interestingly, the interplay between LFY, WUS, and AG largely control the determination of the floral meristems. The combined action of WUS and LFY in young floral meristem directly induce AG expression in the center of flower buds, where stamens and carpels would differentiate. Once activated, AG represses WUS to allow floral meristem termination and the development of the proper set of floral organs in the central whorls of the flower. Accordingly, loss of AG function leads to indeterminate flowers, due to the extended WUS activity in the floral meristem (Lenhard et al., 2001;Lohmann et al., 2001).
Because FUL is strongly upregulated in the inflorescence meristem at the time of floral transition (Mandel and Yanofsky, 1995a;Hempel et al., 1997;Torti et al., 2012), LFY and AG activation should be avoided in the SAM to allow the indeterminate growth of the inflorescence and the maintenance of WUS expression in the SAM. Our results indicate that terminal flower formation is prevented by AP2 and AP2-like genes, suggesting that these factors could block the FUL-mediated AG activation. Thus, the balance between FUL activity (promoting SAM differentiation) and AP2-like activity (preventing SAM differentiation) could be important to maintain the indeterminate fate of the inflorescence. The formation of the terminal flower is premature in the single ap2 mutant, but it appears even earlier when the expression of all AP2-like genes is reduced (35S::miR172 line) suggesting an additive role of the members of this gene family in AG repression. Interestingly, the 35S::miR172 plants did not produce a terminal flower immediately after floral transition, despite FUL strong upregulation at this time in the SAM (Hempel et al., 1997). This suggests that perhaps FUL levels should be high enough to overcome the AP2 repressive effect, or that additional factors are also acting to prevent inflorescence meristem determination or required to promote floral identity. A clear candidate for this putative function could be TFL1, a repressor of LFY in the inflorescence meristem and whose mutation causes the early differentiation of the inflorescence meristem into a flower (Bradley et al., 1997;Ohshima et al., 1997;Ratcliffe et al., 1998). While the functional relationship of FUL and TFL1 has not been precisely elucidated, it has been described that FUL acts by balancing the activities of LFY and TFL1 in the inflorescence (Ferrandiz et al., 2000a). The possible interaction of TFL1 and the FUL-AP2 genetic module in the control of the inflorescence meristem fate, and also on the regulation of the inflorescence proliferative arrest and the length of the flowering phase is of high interest, but remains to be addressed in further studies.
Both FUL and AP2-like genes have been described to be ageregulated by the balance between the miR156 and miR172, with important roles directing developmental phase transitions in plants Wu et al., 2009). Phase transitions occur in a gradual and progressive way during early stages of plant development Wu et al., 2009). Interestingly, a similar model was also proposed to control inflorescence progression in the past (Schultz and Haughn, 1993;Bradley et al., 1997;Ratcliffe et al., 1998). The moment of the floral transition marks the transition from the vegetative phase to the reproductive phase, and the beginning of inflorescence development. Schultz and Haughn (1993) proposed that during inflorescence development there is a progression of different developmental phases. They also proposed that these inflorescence developmental phases should be controlled by unknown factors that gradually would control the levels of the floral identity genes (Bowman et al., 1993;Schultz and Haughn, 1993). Following this model, the authors distinguished three phases during inflorescence development in Arabidopsis: an initial phase where the SAM produces leaves that sustain new branches; a second phase where the SAM produces directly flowers, and a last phase or terminal phase where the SAM differentiates into a flower. It was also discussed that the last phase, the formation of the terminal flower, usually was not observed due to the meristem arrest associated with the end of flowering (Schultz and Haughn, 1993;Bradley et al., 1997;Ratcliffe et al., 1998). Our proposed model is in agreement with the original model proposed by Schultz and Haughn, where the FUL-AP2 pathway could act as the unknown factors that gradually regulate the activation of the floral identity genes in the apical part of the inflorescence. In agreement with the model, ful mutant shows an extension on the length of all these phases: it shows a delay in floral transition, a longer first phase of inflorescence development producing extra caulinar leaves (Balanza et al., 2014), an extended second phase with the production of more flowers before meristem arrest (Balanza et al., 2018) and, moreover, it does not enter the last phase (the inflorescence determination) even in the absence of seed production. On the other hand, the ap2-like sextuple mutant is early flowering, shortens both phases of inflorescence development (branches and flowers) (Yant et al., 2010;Balanza et al., 2018), and finally enters the last phase developing a terminal flower.
Based in phylogenetic analysis, it has been suggested that determinateness would be the ancestral condition in plants and that indeterminateness could have evolved different times in many species (Stebbins, 1974). Other authors, however, have proposed that indeterminateness could have been acquired very early in evolution and that could have been lost later in some species (Coen and Nugent 1994;Bradley et al., 1997). Our analysis and previous work from other groups indicate that determinateness is facultative in Arabidopsis (Chaudhury, 1993;Schultz and Haughn, 1993;Modrusan et al., 1994;Hensel et al., 1994), but hidden by the SAM arrest induced by the correlative control exerted by seeds. From an adaptive point of view, increasing the number of flowers/seeds could provide a clear advantage and increase the reproductive success. On the other hand the production of too many flowers/seeds could also be deleterious, due to the investment of valuable resources to the production of unnecessary seeds, as well as to the increased competition of individuals following seed germination that could affect offspring fitness. According to our observations, Arabidopsis could have adopted the strategy to delay the SAM determinacy increasing seed production but at the same time developing a mechanism to limit their final number. Once an optimal number of seeds is produced, a signal could be sensed at the SAM to induce its arrest. This mechanism could also provide an extra benefit. The meristem arrest observed in Arabidopsis is reversible (Hensel et al., 1994;Wuest et al., 2016), and if some seed losses are produced, the SAM is able to reactivate and produce new flowers and seeds to ensure further seed production.
In many crops, the end of the flowering phase determines the maximum yield that the plant is able to produce. Thus, understanding the mechanisms that regulate this developmental process could allow the engineering of agronomically important crops in order to increase their yields. Our results complement the scarce information available about the control of the reproductive phase length, as well as provide evidence about how Arabidopsis could have evolved an indeterminate inflorescence.

Induced Sterility/Pruning
After bolting, all flowers produced by the SAM were removed manually every 2-3 days, leaving only the flowers before the anthesis stage. In addition, all the new branches developed by the plant were also removed as they appeared.

B-glucuronidase Staining
For GUS histochemical detection, samples were treated for 15 min in 90% ice-cold acetone and then washed for 5 min with washing buffer (25 mM sodium phosphate, 5 mM ferrocyanide, 5 mM ferricyanide, and 1% Triton X-100) and incubated from 4 to 16 h at 37°C with staining buffer (washing buffer+1 mM X-Gluc). Following staining, plant material was fixed, cleared in chloral hydrate, and mounted to be viewed under bright-field microscopy.

In Situ hybridization
In situ hybridizations were performed as described (Ferrandiz et al., 2000a). For hybridization in sections, tissue was fixed for 2 h in FAE solution, dehydrated, embedded, and sectioned to 8 µm. After dewaxing in histoclear and rehydrating, sections were treated for 20 min in 0.2 M HCl, neutralized for 10 min in 2× SSC and then incubated for 30 min with 1 µg/ml Proteinase K at 37°C. Proteinase action was blocked by treating with 2 mg/ml Gly for 5 min and postfixation in 4% formaldehyde for 10 min. Subsequently, sections were dehydrated through an ethanol series before applying the hybridization solution (100 µg/ml transfer RNA; 6× SSC; 3% SDS; 50% formamide, containing approx. 100 ng/µl of antisense digoxigenin-labeled RNA probe), and left overnight at 52°C. Then, sections were washed twice for 90 min in 2× SSC: formamide (50:50) at 52°C before performing the antibody incubation and color detection.

acKnOWLeDgMenTS
We thank Markus Schmid for gently sharing the 35S:miR172 line, and Concepción Gómez-Mena for the gift of the AG::GUS reporter line seeds, as well as to Paz Merelo (IBMCP) for comments on the manuscript. We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).