The ERECTA, CLAVATA and class III HD-ZIP Pathways Display Synergistic Interactions in Regulating Floral Meristem Activities

In angiosperms, the production of flowers marks the beginning of the reproductive phase. At the emergence of flower primordia on the flanks of the inflorescence meristem, the WUSCHEL (WUS) gene, which encodes a homeodomain transcription factor starts to be expressed and establishes de novo stem cell population, founder of the floral meristem (FM). Similarly to the shoot apical meristem a precise spatial and temporal expression pattern of WUS is required and maintained through strict regulation by multiple regulatory inputs to maintain stem cell homeostasis. However, following the formation of a genetically determined fixed number of floral organs, this homeostasis is shifted towards organogenesis and the FM is terminated. In here we performed a genetic study to test how a reduction in ERECTA, CLAVATA and class III HD-ZIP pathways affects floral meristem activity and flower development. We revealed strong synergistic phenotypes of extra flower number, supernumerary whorls, total loss of determinacy and extreme enlargement of the meristem as compared to any double mutant combination indicating that the three pathways, CLV3, ER and HD-ZIPIII distinctively regulate meristem activity and that they act in parallel. Our findings yield several new insights into stem cell-driven development. We demonstrate the crucial requirement for coupling floral meristem termination with carpel formation to ensure successful reproduction in plants. We also show how regulation of meristem size and alternation in spatial structure of the meristem serve as a mechanism to determine flower organogenesis. We propose that the loss of FM determinacy due to the reduction in CLV3, ER and HD-ZIPIII activity is genetically separable from the AGAMOUS core mechanism of meristem termination.


Introduction
In angiosperms, the production of flowers marks the beginning of the reproductive phase. At the emergence of flower primordia on the flanks of the inflorescence meristem, the WUSCHEL (WUS) gene, which encodes a homeodomain transcription factor, starts to be expressed in a few cells known as the organizing center (OC). WUS then specifies stem cell identity in the overlying cells to establish de novo stem cell population, founder of the floral meristem (FM) [1][2][3]. The stem cells divide and their daughter cells can either remain stem cells or proliferate before being incorporated into floral organ primordia. To maintain the organization of the FM, an homeostasis which is the balance between stem cell renewal, cell proliferation and cell differentiation, must be kept [4]. Similar to the shoot apical meristem (SAM), stem cell homeostasis within the FM is mediated by the CLAVATA-WUS feedback loop [5][6][7]. However, following the formation of a genetically determined fixed number of floral organs, this homeostasis is shifted towards organogenesis and the FM activity terminates. Genetic studies have identified numerous mutants in which the homeostasis between stem cell population size and cells that are recruited for floral organ primordia formation is disrupted, leading to a decrease or increase in floral organ number. For example mutations in WUS, TOUSLED and AUXIN RESISTANT 6 genes lead to reduced meristem size and organ number [1,8,9], whereas loss-of-function alleles of CLAVATA3 (CLV3), ULTRAPETALA1 (ULT1) and PLURIPETATA lead to an increase in FM size and floral organ number [5,6,10,11].
In Arabidopsis thaliana the flowers consist of four sepals, four petals, six stamens and a bicarpellate gynoecium arising in four concentric whorls in a centripetal chronological manner (i.e. with the outer whorls developing before the inner ones).
During early flower development, the flower primordium becomes separated from the inflorescence meristem and starts to grow larger very quickly in all directions to increase meristem size and to provide sufficient cells for the formation of floral organ primordia [12]. The sepal primordia appear first concurrently with further increase in the diameter of the developing flower primordium [13]. Next the petal and stamen primordia become visible. At floral stage 6 (stages according to Smyth et al. [13]), simultaneously with FM termination a rim at the central dome of the flower primordium grows upward to produce a hollow tube that will become the gynoecium consisting of two fused ovule-bearing carpels [14,15]. Early in flower development, at floral stage 3, WUS activates the expression of the AGAMOUS (AG) MADS-box gene, a central factor in specifying stamen and carpel identities [16]. AG in turn terminates floral stem cell maintenance by repressing WUS expression thus promoting meristem determinacy [2,[17][18][19]. Mutations in AG cause the formation of indeterminate meristem which produces flowers-within-flower phenotype with no stamens and carpels [20][21][22]. Accordingly, late stage ag mutant flower still shows strong WUS expression in the center of a regular size floral meristem [17]. Therefore, to allow proper flower development, a precise spatial and temporal expression pattern of WUS is required and maintained through strict regulation by multiple regulatory inputs.
Our recent study demonstrated that all three act in separate pathways at the SAM and have similar functions in the FM [33]. Here, we further address how loss-of-function in those pathways affects FM activity and flower development, and demonstrates the importance of cell proliferation regulation for correct flower organogenesis to ensure successful reproduction.

Results and Discussion
The ERECTA, CLAVATA and class III HD-ZIP pathways synergistically regulate floral meristem activity To investigate the interaction of the ER, CLV and HD-ZIPIII pathways in regulating floral meristem activity, we analyzed flowers of single, double and triple mutant combinations of clv3-2, er and jabba-1D (jba-1D). In the jba-1D mutant overexpression of miR166g leads to decrease in several HD-ZIPIII transcripts level leading to up-regulation of WUS expression and enlargement of the SAM [25]. No detailed analysis of FM size or floral organ numbers has been reported for jba-1D. Loss of function of the three ER family genes leads to flattened and broadened SAM and aberrant flower development [34].Yet the single mutant does not exhibit increase in FM size or in floral organ number. However, the clv3-2 mutant exhibits enlarged floral meristem and increased number of floral organs [5,6].
Analyzing the flowers of the single double and triple mutant combinations we observed two major categories of phenotypes. The first is increase in floral organ numbers, [Fig 1], extra whorls S1 Fig and fasciated FM, all which can be attributed to an enlargement of the FM. To date, in all identified mutants that display enlarged FM but no reduction in floral organ numbers, the increase in meristem size was the results of WUS up-regulation [10,[30][31][32]35]. Therefore the first category of phenotype is likely to be a result of WUS up-regulation or an expansion of the WUS expression domain.
The second category is supernumerary whorls [Figs 2 and 3] and a total loss of determinacy [ Figs 2E and 3D] that can be attributed to extended meristem activity. As was shown previously for mutants such as clv3, ag and ult1, this phenotype of extended meristem activity is most likely the outcome of prolonged WUS expression [10,17].
In the single or combined mutants studied, all floral organs were present in higher numbers [ Fig 1C], indicating that the increase in organ and whorls numbers result from either an increase in the stem cell population or from the temporal alteration in stem cell termination rather than an homeotic conversion of one organ type into another. An enlarged stem cell population provides more cells for organogenesis allowing more organ primordia to develop.
Among the single mutants, clv3-2 displayed the most dramatic increase in floral organ number, particularly for stamens and carpels with more than 80% of flowers producing at least one extra stamen and up to a total of 10 stamens, and all flowers producing at least one extra primary carpel [ Fig 1]. Furthermore, we observed two types of supernumerary carpels within the gynoecium of the clv3-2 flowers: one that developed from extended meristem activity, seen as ectopic carpel inside the primary carpel [ Fig 2A and 2F], and the other arising from the septum [ Fig 2C]. We also observed abnormally wide repla S1 Fig, which might be due to enhanced carpel margin meristem activity [36]. Clark et al [5] showed that the clv3-2 FMs are significantly taller and only slightly wider than wild-type FMs. This may explain the much greater increase in numbers of stamens and carpels arising from the inner pair of whorls as compared to numbers of sepals and petals arising from the perianth whorl, the outermost two whorls. In the jba-1D/+ single mutant low percentage of flowers exhibits increased floral organ number although these flowers occasionally displayed petals arising on consecutive whorls, seen as a petal on top of another petal S1 Fig, suggesting that in jba-1D/+ the FM expands laterally thus providing more cells for the development of extra perianth whorl. Both the Ler background and jba-1D/+ enhanced the clv3-2 flower phenotypes of floral organ numbers [Fig 1] and extra whorl numbers [ Fig 2D], displaying up to 3 extra whorls of carpels formed within carpels. However, the enhancement of clv3-2 by jba-1D/+ was stronger than the enhancement by Ler. Indeed, clv3-2 jba-1D/+ double mutants showed increased numbers of floral organ in all whorls [ Fig 1C], with an average of 4.5 sepals, 4.7 petals, 8.5 stamen and 5 carpels [S1 Table]. This increase suggests that the FMs of the clv3-2 jba-1D/+ double mutant expanded both laterally and horizontally. Among clv3-2 jba-1D/+ flowers, 46% and 61% developed 5 sepals and 5 petals, respectively, indicating that the FM is already enlarged at early stages of flower development, since sepal primordia arise at flower stage 3, shortly after the emergence of the flower primordium [13]. Altogether, these results suggest that defects in the CLV3 and HD-ZIPIII pathways result in strong up-regulation of WUS, starting at the onset of its expression and leading to a born enlarged FM.
Of the three double-mutant combinations studied, jba-1D/+ er-20 showed the lowest percentage of flowers with supernumerary organs [Fig 1]. Moreover, this double mutant did not produce extra carpel whorls [ Fig 2B]. Our analysis showed that in the combinations carrying mutation in ER, the carpels and siliques were shorter and wider S2 Fig, as has been previously reported for er mutants [37,38]. In wild type flower the gynoecium develops from two congenitally fused carpels that arise from the center of the meristem as a tube [14,15]. Therefore the wide carpels in er backgrounds S2 Fig might be due to an increase in the width of the FM central zone resulting in a wider gynoecial tube. Uchida et al. [28] showed that loss of function of all three ER family genes leads to lateral expansion of the WUS expression domain, resulting in flattened and broadened SAMs. A lateral expansion of the WUS domain was also shown in jba-1D/+ seedling SAM [25].  We propose that reduction in both ER and HD-ZIPIII activity in the FM leads to lateral expansion of the WUS domain, resulting in spatially alteration of the meristem so that the stem cell population expands horizontally at the expense of a reduced peripheral zone. As a consequence more cells remain as stem cells with strong pluripotent signals that are not competent to differentiate and to be recruited for primordia initiation, leaving fewer cells for the periphery. This can provide an explanation for the low percentage of flowers with supernumerary floral organs that were observed in jba-1D/+ er-20 plants. The evidence that introducing er-20 to clv3-2 jba-1D/+ suppresses the increased number of sepals, petals and stamens phenotype observed in clv3-2 jba-1D/+ but not the supernumerary carpel phenotype [S1 Table], further support this idea that er-20 changes the meristem structure, leading to a reduced peripheral zone from where sepals petals and stamens are formed. In the clv3-2 jba-1D/+ er-20 the most dramatic increase was detected in the number of primary carpels in all the flowers [Fig 1]. We found flowers with up to 14 carpels, all of which developed from the primary gynoecial tube [data not shown]. Our findings suggest that in the triple mutant compared with WT all the FM domains are expanded, providing more cells for all whorls. In addition, the triple mutant flowers exhibited reiteration of multi-fused carpel within carpels that encircle a gradually growing everlasting meristem [ Fig 3C]   background resulted in a wus-1 phenotype, indicating that wus is fully epistatic to the three pathways [data not shown], thus in agreement with previously reported data [25,33].
These strong synergistic phenotypes observed in the triple mutant of extra flower number, supernumerary whorls, total loss of determinacy and extreme enlargement of the meristem as compared to any double mutant combination, indicate that the CLV3, ER and HD-ZIPIII pathways regulate FM activity in parallel.

Coordinated FM termination and carpel formation ensure successful reproduction
The flowers of the triple mutant continued to produce carpels while stem cells were still proliferating. This provides additional evidence that carpel identity specification and floral stem cell termination are two separable processes [39]. Yet scanning electron microscope analysis of the triple mutant flower at stage 7, which reveals a dome shaped meristem at the center of the gynoecial tube [Fig 3A], strongly emphasizes the fundamental requirement for coordination between those two independent processes to ensure successful reproduction of plants [2,[40][41][42].
In wild-type flowers, following the initiation and growth of the gynoecial tube, the medial ridges meet and fuse to form the septum, which divides the fruit, stretching from the inner side of one replum to the other replum while the ovule primordia arise from the placenta [ Fig 3I]. Later, cells in the middle of the septum will form the transmitting tract, a specialized tissue essential for pollen tube growth [15]. In Col, Ler, jba-1D/+ and jba-1D/+ er-20 double mutant, the gynoecium internal tissues developed properly, leading to successful fertilization and seed production [ Fig 3E and 3F]. In clv3-2 single and double mutants, the septum and the transmitting tract developed normally, and fertilization occurred prior to the development of the supernumerary carpel whorls that later disrupted the septum [Fig 2A and 2D]. In contrast, the triple mutant FM showed strong meristematic activity and loss of determinacy. Upon the rising of the gynoecial tube, the active meristem physically blocked the development of the septum and consequently the formation of the transmitting tract [ Fig 3A and 3B]. Dissecting mature fruits, we found dense meristematic tissue filling the inner spaces of the gynoecium [Fig 2A]. Though at later stages, when the meristem erupts, some ovules can be seen developing on the exposed placenta [ Fig 3H]; with no transmitting tract the pollen tube is incapable of reaching the ovules thus resulting in infertile silique.
We conclude that, although carpel specification and floral stem cell termination are separable processes, both processes must be strictly coordinated for successful reproduction.

The robust activity of the FM causes FM indeterminacy
Many regulators governing the temporal repression of WUS to terminate floral stem cell proliferation are part of the AG network, acting as AG activators or downstream to AG [2,19,[43][44][45][46][47]. AG has recently been shown to directly repress WUS expression by binding to the WUS locus and recruiting the Polycomb Group (PcG) factor CURLY LEAF for chromatin silencing via deposition of Histone3 Lysine27 repressive methylation marks at WUS [19].
Although a decrease in the activity of CLV3, ERECTA and HD-ZIPIII results in loss of determinacy, we propose that these pathways are genetically separable from the AG core mechanism of meristem termination: The AG gene is required for carpel development as demonstrated by the ag mutant having no carpels. Also, we have previously shown that AG is required to confer the jba-1D/+er-20 ectopic carpel phenotype [33]. Since the jba-1D/+ er-20 clv3-2 triple mutant flowers continually producing carpels it is likely that the AG gene is highly expressed and that AG did not lose its function, however fails to repress WUS.
We hypothesize that, due to a reduction in these three pathways that restricts WUS expression, the WUS locus is fully occupied by the transcriptional machinery complexes, keeping it in an active chromatin state and preventing the AG-dependent chromatin modifier mechanism to act. The result would be a failure in shutting down WUS and a total loss of FM determinacy. A future challenge will be to test the molecular basis of this hypothetical scenario.
In summary, the analysis of flowers from combined mutants demonstrates that CLV3, ER and HD-ZIPIII participate to regulate FM activity, presumably through distinct pathways, all controlling WUS expression. We show that meristem activity termination is not a prerequisite for carpel formation but that coupling FM termination with carpel formation is crucial for seed formation. We also demonstrate that regulation of meristem size is one mechanism that leads to determine flower organogenesis.

Microscopy
Plant images were captured using an Olympus SZX7 Stereomicroscope.
Scanning electron microscopy was performed by fixing tissue in methanol as described [48] and using a JEOL 5410 LV scanning electron microscope.

Floral organ counting and Statistical analysis
Floral organ counting was performed as described by Fiume et al. [49]. Statistical analysis was carried out using JMP 10 software (SAS Institute Inc., Cary, NC). For each line the significance of the categorical partition of organs number was calculated using Fisher`s exact test (p < 0.05) in comparison to it background (i.e Col or Ler). [H] clv3-2 jba 1D/+ er-20 exhibits short sphere-shaped siliques with numerous extra carpels. Note that siliques from mutants carrying mutation in ERECTA are much shorter and border compared to the background plants. Scale bar: 2mm (TIF) S1 Table. Average of floral organ numbers. Numbers of sepal petal stamen and carpel were counted according to Fiume [49] (number of flowers counted in parentheses) and the mean was calaculated [in bold]. The Std Err Mean appears on the right column. (DOCX) S2

Author Contributions
Conceived and designed the experiments: LEW UL. Performed the experiments: UL LA. Analyzed the data: LEW UL. Contributed reagents/materials/analysis tools: LEW. Wrote the paper: LEW.