Arabidopsis thaliana SHOOT MERISTEMLESS Substitutes for Medicago truncatula SINGLE LEAFLET1 to Form Complex Leaves and Petals

LEAFY plant-specific transcription factors, which are key regulators of flower meristem identity and floral patterning, also contribute to meristem activity. Notably, in some legumes, LFY orthologs such as Medicago truncatula SINGLE LEAFLET (SGL1) are essential in maintaining an undifferentiated and proliferating fate required for leaflet formation. This function contrasts with most other species, in which leaf dissection depends on the reactivation of KNOTTED-like class I homeobox genes (KNOXI). KNOXI and SGL1 genes appear to induce leaf complexity through conserved downstream genes such as the meristematic and boundary CUP-SHAPED COTYLEDON genes. Here, we compare in M. truncatula the function of SGL1 with that of the Arabidopsis thaliana KNOXI gene, SHOOT MERISTEMLESS (AtSTM). Our data show that AtSTM can substitute for SGL1 to form complex leaves when ectopically expressed in M. truncatula. The shared function between AtSTM and SGL1 extended to the major contribution of SGL1 during floral development as ectopic AtSTM expression could promote floral organ identity gene expression in sgl1 flowers and restore sepal shape and petal formation. Together, our work reveals a function for AtSTM in floral organ identity and a higher level of interchangeability between meristematic and floral identity functions for the AtSTM and SGL1 transcription factors than previously thought.


Introduction
Meristems are essential for plant development, as they are required for the continuous growth and development that are distinguishing features of plants. Amongst all the different types of meristems, the shoot apical meristem (SAM) and the floral meristem (FM) share many features and have been well characterized. The class I KNOTTED-like homeobox (KNOX1) SHOOT MERISTEMLESS (STM) and CUP-SHAPED COTYLEDON (CUC1), CUC2 and CUC3 genes are essential regulators of meristem and boundary activities in Arabidopsis thaliana (A. thaliana) [1,2]. Boundaries are domains of restricted growth located between the meristem and initiating organ primordia or between two organs. These domains control organ separation, inflorescence architecture, organ abscission, fruit opening and leaf shape. Boundaries share overlapping features with meristems, and the regulation of both involves common factors [3]. CUC genes are required for SAM initiation and establish boundaries together with STM, which is in turn required for SAM maintenance [4][5][6][7][8][9][10][11] The three A. thaliana CUC genes, CUC1, CUC2 and CUC3, share partially rise to a lateral secondary inflorescence meristem (I2), which produces a bract, one to three flowers and a spike [59][60][61]. In contrast to other flowering species which show a sequential floral ontogeny with successive formation of sepals, petals, stamens and carpels, each floral organ derives from a specific primordium; petals and sepals differentiate from common primordia in M. truncatula [59]. Thus, each floral meristem gives rise sequentially to five sepals and four common primordia, which further differentiate into five petals and ten stamens, and one carpel. The Arabidopsis floral organ identity genes are conserved in legumes [62]. Loss of function of SGL1 leads to the reversion of common primordia into incomplete floral meristems, giving rise to sepals and carpels without petals and stamens [42]. This phenotype is related to a B function loss. Similar to LFY in Arabidopsis [63], SGL1 acts synergistically with MtPROLIFERATING INFLORESCENCE MERISTEM (MtPIM), the A. thaliana ortholog of APETALA1 (AP1) in M. truncatula to determine floral meristem identity [61,64]. MtNMH7 and MtTM6 are the A. thaliana AP3-like paralogs. MtNMH7 determines petal identity whereas MtTL6 controls stamen identity [65]. MtPISTILLATA (MtPI) and MtNGL9 are the two A. thaliana PI-like paralogs, with MtPI functioning as the master regulator of B function [66,67]. The M. truncatula genome harbors two redundant MtAG members, MtAGa and MtAGb, which specify stamen and carpel identity and floral meristem determinacy [68,69]. Recently, a novel regulator of inflorescence development and floral organ identity was identified in M. truncatula: the AGAMOUS-like FLOWERS (AGLF) gene, which encodes a MYB domain protein that promotes the C floral identity function besides repressing A and B functions [69,70].
Here, we further compared the meristematic activity of SGL1 (LFY) and AtSTM (KNOX1) using M. truncatula compound leaf as a model system. We first showed that AtSTM could substitute for SGL1 to form complex leaves. We next tested whether AtSTM could also substitute for SGL1's role during floral development. Indeed, AtSTM expression could restore petal formation in sgl1 flowers, revealing that AtSTM could substitute for SGL1 function to specify petal identity and promote floral organ identity gene expression. Therefore, our data reveal a high level of interchangeability between SGL1 and KNOX1 activities in M. truncatula that extends beyond the generally accepted meristematic function to the determination of the identity and growth of the flower perianth.

AtSTM Substitutes for SGL1 in M. truncatula to Form Compound Leaves
The M. truncatula genome harbors two MtSTM-like genes, MtKNOX1 and MtKNOX6, and a previous report describes in vitro plantlets overexpressing the MtSTM-like genes, MtKNOX1 and MtKNOX6, in M. truncatula [46]. However, only the vegetative phenotype was described, as the phenotype of MtKNOX1 and MtKNOX6 overexpressors was extremely severe. Therefore, to overcome such strong phenotypes, we thought to use a KNOX gene from a heterologous system. AtSTM shares 62.8% amino acid identity with MtKNOX1 and 64.86% with MtKNOX6, and in addition to modifying leaf shape when ectopically expressed, AtSTM also has an established role in Arabidopsis floral identity [5,6,32,52,71,72]. Thus, we selected AtSTM to be expressed in M. truncatula and to explore its potential more widely; we expressed it under two different promoters by generating the p35S:AtSTM and pSGL1:AtSTM constructs that we first introduced in wild-type plants (see Section 4 and Supplemental Figure S1).
Transgenic lines expressing high levels of AtSTM presented a severe phenotype and were not viable in the greenhouse, similar to in vitro plantlets overexpressing Mt-KNOX1like genes [46] ( Figure S2). Only transgenic plants with low levels of AtSTM expression could be investigated (Figures 1 and S3). The overall development of these lines was quite normal, although their fertility was reduced. In wild-type M. truncatula, the juvenile first leaf is simple, while adult later leaves are trifoliate and composed of a terminal leaflet with two lateral leaflets ( Figure 1A-D). Ectopic AtSTM under the p35S or the pSGL1 promoters occasionally led to the formation of an additional leaflet fused to the terminal leaflet of adult leaves ( Figures 1F,H and S3). Quantitative analyses were performed using the p35S:AtSTM line ( Figure 1Q). The wild-type first leaves (rank 1) were simple, while the majority of adult leaves (ranks 2 to 5) were trifoliate (only 4 out of 72 leaves had more than three leaflets). The p35S:AtSTM sequences seldom led to complex leaves, as only 8 out of 72 adult leaves (ranks 2 to 5) were more complex ( Figure 1Q). leaf is simple, while adult later leaves are trifoliate and composed of a terminal leaflet with two lateral leaflets ( Figure 1A-D). Ectopic AtSTM under the p35S or the pSGL1 promoters occasionally led to the formation of an additional leaflet fused to the terminal leaflet of adult leaves ( Figures 1F,H and S3). Quantitative analyses were performed using the p35S:AtSTM line ( Figure 1Q). The wild-type first leaves (rank 1) were simple, while the majority of adult leaves (ranks 2 to 5) were trifoliate (only 4 out of 72 leaves had more than three leaflets). The p35S:AtSTM sequences seldom led to complex leaves, as only 8 out of 72 adult leaves (ranks 2 to 5) were more complex ( Figure 1Q).  We then tested whether AtSTM expression is sufficient to rescue the sgl1 leaf phenotype (see Section 4). In the sgl1 mutant, the majority of leaves are simple ( Figure 1I-L). All rank 5 leaves were simple, but 12 out 54 leaves (ranks 2 to 4) were bi-or trifoliate in the sgl1 mutant ( Figure 1Q). In contrast, in p35S:AtSTM sgl1 plants, the majority of adult leaves were trifoliate as in wild-type ( Figure 1B-D,O,P). The p35S:AtSTM construct restored almost systematically the capacity to form trifoliate leaves, with 52 out of 54 leaves (ranks 3 to 5) producing at least three leaflets ( Figure 1Q). Therefore, we concluded that AtSTM can replace SGL1 to promote leaflet formation.
To explore the developmental origin of the extra or rescued leaflets in the different backgrounds, we imaged by SEM young developing leaf primordia ( Figure 2). As observed in wild-type apices, a pair of lateral leaflets and a terminal leaflet initiated in AtSTM transgenic lines during early leaf primordium development (Figure 2A,C). At stage S8, additional leaflets could form at the base of the terminal leaflet in AtSTM (arrows Figure 2D), which were not observed in the wild-type ( Figure 2B) and therefore resulted from secondary morphogenesis. This indicates that the morphogenetic window during which leaflets can be initiated is extended following AtSTM expression. In p35S:AtSTM sgl1 plants, the terminal primordium was surrounded by two lateral primordia ( Figure 2G,H), already visible at early stages (S4), as seen in the wild-type (Figure 2A,G). Thus, leaflet restoration in p35S:AtSTM sgl1 does not appear to rely on a late production of leaflets but a rescue of the normal developmental process with a restoration of early lateral leaflet initiation, as occurs in the wild-type.
simple, while adult leaves are trifoliate and composed of a terminal leaflet plus two lateral leaflets. The petiole (p) and the rachis (r) are indicated. (E-H) p35S:AtSTM, a transgenic line expressing AtSTM under the p35S promoter producing a L2 and a L4 heart-shaped adult leaves with an ectopic leaflet fused to the terminal leaflet (arrowheads). This phenotype was occasionally observed. Leaflet margins are serrated. (I-L) sgl1 line, showing simple juvenile (L1) and adult leaves (L2-L4). (M-P) p35S:AtSTM sgl1 line, showing trifoliate L3 and L4 leaves similar to wild-type. (Q) Quantification of the leaflet number. Four-week-old plants were analyzed (n = 18 plants per genotype). Average ± SD are shown. Lowercase letters indicate significant differences between genotypes at each leaf rank (one-way ANOVA with Tukey's post hoc test; p ≤ 0.001). Bars = 5 mm.
We then tested whether AtSTM expression is sufficient to rescue the sgl1 leaf phenotype (see Section 4). In the sgl1 mutant, the majority of leaves are simple ( Figure 1I-L). All rank 5 leaves were simple, but 12 out 54 leaves (ranks 2 to 4) were bi-or trifoliate in the sgl1 mutant ( Figure 1Q). In contrast, in p35S:AtSTM sgl1 plants, the majority of adult leaves were trifoliate as in wild-type ( Figure 1B-D,O,P). The p35S:AtSTM construct restored almost systematically the capacity to form trifoliate leaves, with 52 out of 54 leaves (ranks 3 to 5) producing at least three leaflets ( Figure 1Q). Therefore, we concluded that AtSTM can replace SGL1 to promote leaflet formation.
To explore the developmental origin of the extra or rescued leaflets in the different backgrounds, we imaged by SEM young developing leaf primordia ( Figure 2). As observed in wild-type apices, a pair of lateral leaflets and a terminal leaflet initiated in AtSTM transgenic lines during early leaf primordium development (Figure 2A,C). At stage S8, additional leaflets could form at the base of the terminal leaflet in AtSTM (arrows Figure 2D), which were not observed in the wild-type ( Figure 2B) and therefore resulted from secondary morphogenesis. This indicates that the morphogenetic window during which leaflets can be initiated is extended following AtSTM expression. In p35S:AtSTM sgl1 plants, the terminal primordium was surrounded by two lateral primordia ( Figure  2G,H), already visible at early stages (S4), as seen in the wild-type (Figure 2A,G). Thus, leaflet restoration in p35S:AtSTM sgl1 does not appear to rely on a late production of leaflets but a rescue of the normal developmental process with a restoration of early lateral leaflet initiation, as occurs in the wild-type.   Figure 3B), a corolla containing three types of yellow petals, the standard or the vexillum at the adaxial position ( Figure 3C), the keel formed by two fused petals at the abaxial position surrounded by two lateral petals and the alae or wings ( Figure 3D-G). The third whorl consists of an independent stamen filament at the adaxial position, the vexillary stamen filament and nine stamen filaments fused into a staminal tube that surrounds a monocarpous gynoecium [59] ( Figure 3H,I). The sgl1 mutants produce inflorescences with cauliflower-like floral structures, containing incomplete floral meristems (FMs), elongated sepals and occasionally carpels [42] ( Figure 3U). These cauliflowers do not produce petals nor stamens, similar to lfy mutants in Arabidopsis [73].     Figure 3N,S), petaloid stamens ( Figure 3O,P) and petaloid carpels ( Figure 3Q). The fertility was severely reduced, with some plants infertile. The fruits were small, with fewer discs and unbent spines compared with wild-type fruits ( Figure 3T,J). These fruits contained a few seeds. The same phenotypes were occasionally observed in pSGL1:AtSTM flowers ( Figure S3F,G). We then tested the effects of p35S:AtSTM on sgl1 flower development. Surprisingly, the ectopic expression of AtSTM rescued sepal shape and petal formation in the sgl1 mutant ( Figure 3V-Y). Similar to wild-type flowers, p35S:AtSTM sgl1 flowers formed a calyx with five sepals fused at their base ( Figure 3B,W). Inside the calyx, the p35S:AtSTM sgl1 flowers showed a cauliflower phenotype with incomplete FMs, producing a few sepals and a majority of petals or petals with sepal sectors. Petals were partially restored as some of them had a vexillium-like, wing-like or keel-like shape ( Figure 3X,Y). Thus, when ectopically expressed, AtSTM restores petal formation in sgl1. These flowers did not form carpels, in contrast to sgl1 flowers, suggesting a deficiency in C function ( Figure 3). The majority of organs formed were petals, as one cauliflower flower from a 35S:AtSTM sgl1 line could produce up to 65 petals ( Figure S4). SEM analyses were performed to further characterize these flowers at early developmental stages. Figure 4A-D shows wild-type floral development. At stage 4, the wild-type floral meristem had formed five sepal primordia, four common primordia and a carpel primordium ( Figure 4B). At late stage 5, the wild-type floral meristem displayed the complete set of floral organ primordia, with petal and stamen primordia deriving from the differentiation of common primordia ( Figure 4D). Figure 4E-H shows the floral development of a p35S:AtSTM plant wild-type for SGL1. Figure 4F shows a late stage 5 p35S:AtSTM floral meristem. Based on sepal development, a delay in the formation of the inner floral organ primordia could be observed compared with the wild-type ( Figure 4F,D). In contrast, Figure 4G shows a stage 5 floral meristem containing differentiated petals and stamen primordia and two carpel primordia, indicating that the delay in internal organ primordia differentiation is variable between flowers. Figure 4H shows a p35S:AtSTM flower developing two carpels. Similar to previous data [42,61], sgl1 inflorescences showed multiple incomplete FMs, elongated sepals, defective common primordia and carpel primordia ( Figure 4I-L). Sepal primordia further develop into elongated sepals and carpel primordia into a carpel-like structure. The cauliflower phenotype is caused by the iterative conversion of common primordia into incomplete floral meristems ( Figure 4I,L). In sgl1 mutants overexpressing AtSTM ( Figure 4M-O), the sepal form was restored, suggesting that AtSTM could take over SGL1 function for the control of sepal shape. The late stage 5 floral meristems showed a delay in the differentiation of other floral organ primordia, as observed in p35S:AtSTM SGL1 plants ( Figure 4N,F). Later, petals and sepals differentiated from these primordia ( Figure 4O). Together, these observations show that expression of AtSTM partly restored normal early morphogenesis of sgl1 flowers.

AtSTM Substitutes for SGL1 to Promote Floral Organ Identity Gene Expression
To determine if AtSTM activates A and B functions to promote petal formation in sgl1 flowers, we used in situ hybridization to analyze the expression pattern of floral organ identity genes in p35S:AtSTM sgl1 flowers. We first investigated the expression of the A class gene MtPIM, the A. thaliana ortholog of AP1 in M. truncatula. MtAP1 has a conserved role with orthologous genes and is required to specify floral meristem and floral organ identity [61,64]. In wild-type inflorescences, MtAP1 transcripts localize to the floral meristem and bract ( Figure 5A,B). In a stage 4 flower meristem, MtAP1 expression was observed in sepal primordia and was restricted to the outer domain of the common primordia that further gives rise to sepals and petals and was absent from the inner part, which differentiates into stamens and carpel ( Figure 5C) [61,64]. At later stages, MtAP1 expression was maintained in sepals and petals ( Figure 5D). Similar to the pattern described in [61], in sgl1 flowers, MtAP1 was expressed in the floral meristem and in the bract ( Figure 5E). MtAP1 was expressed uniformly in defective common primordia and in reiterated floral meristems ( Figure 5F-H). At later stages, MtAP1 expression localized to the outer incomplete floral meristem and disappeared from the central domain that further differentiates into carpels ( Figure 5F,G). In p35S:AtSTM sgl1 flowers, MtAP1 was more widely expressed than in sgl1 flowers, with MtAP1 detected in reiterated floral meristems and in developing petals (I-K). Thus, in p35S:AtSTM sgl1 flowers, AtSTM acts as a positive regulator of A function, contributing to enhanced petal identity. truncatula. MtAP1 has a conserved role with orthologous genes and is required to specify floral meristem and floral organ identity [61,64]. In wild-type inflorescences, MtAP1 transcripts localize to the floral meri stem and bract ( Figure 5A,B). In a stage 4 flower meristem, MtAP1 expression was ob served in sepal primordia and was restricted to the outer domain of the common primor dia that further gives rise to sepals and petals and was absent from the inner part, which differentiates into stamens and carpel ( Figure 5C) [61,64]. At later stages, MtAP1 expres sion was maintained in sepals and petals ( Figure 5D). Similar to the pattern described in [61], in sgl1 flowers, MtAP1 was expressed in the floral meristem and in the bract ( Figure  5E). MtAP1 was expressed uniformly in defective common primordia and in reiterated floral meristems ( Figure 5F-H). At later stages, MtAP1 expression localized to the oute incomplete floral meristem and disappeared from the central domain that further differ entiates into carpels ( Figure 5F,G). In p35S:AtSTM sgl1 flowers, MtAP1 was more widely expressed than in sgl1 flowers, with MtAP1 detected in reiterated floral meristems and in developing petals (I-K). Thus, in p35S:AtSTM sgl1 flowers, AtSTM acts as a positive reg ulator of A function, contributing to enhanced petal identity. We then investigated the expression of the B class gene MtPI. In wild-type, MtPI transcripts were localized to common primordia cells and later restricted to petal and stamens ( Figure 6A,B) and [66,67]. In the sgl1 mutant, no MtPI expression was detected in defective common primordia, consistent with the phenotype of sgl1 flowers, which lack petals and stamens ( Figure 6C). In sgl1 flowers overexpressing AtSTM, MtPI expression was detected in defective common primordia ( Figure 6D,E inset-a). At a later stage, MtPI localized to the outer domain of the defective common primordia that further gives rise to petal-like organs ( Figure 6E and inset-b). Later, MtPI is expressed in petal-like organs ( Figure 6D  We further determined the expression of the M. truncatula ortholog of the A. thaliana C-class gene AG. MtAGb was used as a probe as its signal is stronger and it is more restricted than that of MtAGa [68]. In wild-type flowers, MtAGb expression was first detected at stage 2 in the central part of the floral meristem where the carpel will develop ( Figure 7A). At stage 4, MtAGb expression was mainly localized to the inner domain of the common primordia that will further give rise to stamens and to carpel primordia (Figure 7B). At later stages, its expression was restricted to stamens, carpel and ovules ( Figure  7C,D). In sgl1 flowers, a weak signal was detected in floral meristems and defective common primordia and was absent in the L1 layer ( Figure 7E-G). Later, its expression was detected in carpel-like structures and ovules ( Figure 7H). In sgl1 plants overexpressing AtSTM, MtAGb expression was detectable in only a few flowers (3 of 13). In these flowers, the signal was weak and restricted to a few cells in FM beneath the two outer most layers ( Figure 7I). We further determined the expression of the M. truncatula ortholog of the A. thaliana C-class gene AG. MtAGb was used as a probe as its signal is stronger and it is more restricted than that of MtAGa [68]. In wild-type flowers, MtAGb expression was first detected at stage 2 in the central part of the floral meristem where the carpel will develop ( Figure 7A). At stage 4, MtAGb expression was mainly localized to the inner domain of the common primordia that will further give rise to stamens and to carpel primordia ( Figure 7B). At later stages, its expression was restricted to stamens, carpel and ovules ( Figure 7C,D). In sgl1 flowers, a weak signal was detected in floral meristems and defective common primordia and was absent in the L1 layer ( Figure 7E-G). Later, its expression was detected in carpel-like structures and ovules ( Figure 7H). In sgl1 plants overexpressing AtSTM, MtAGb expression was detectable in only a few flowers (3 of 13). In these flowers, the signal was weak and restricted to a few cells in FM beneath the two outer most layers ( Figure 7I).

Discussion
Here, we compared the activity of two transcription factors, AtSTM and SGL1, in M. truncatula. Our analysis is based on transgenic plants that were able to grow in a greenhouse and therefore expressed AtSTM at low levels. This allowed us to investigate the activity of AtSTM during flower development.
An increase in the leaflet number was only occasionally observed following AtSTM ectopic expression in wild-type M. truncatula. This limited effect of AtSTM could be linked to AtSTM expression levels in these lines, which were low. The additional leaflets were formed at the base of the terminal leaflet and resulted from a secondary morphogenesis. This suggests that AtSTM leads to additional leaflets through the extension of the meristematic activity, allowing more leaflets to emerge, and not from the division of the lateral leaflets into two structures. In M. truncatula, the terminal leaflet derives from the terminal zone where auxin maxima are located through the activity of SMOOTH LEAF MARGIN1 (SLM1), the PIN1 ortholog in M. truncatula [74]. Lateral leaflets result from the marginal blastozone activity and the formation of local auxin maxima that depend on SGL1 activity [74]. The tetrafoliate pattern seen in AtSTM transgenic lines likely results from a defect in auxin distribution in the terminal zone. This leaf patterning is also found in M. truncatula plants inactivated for HEADLESS (HDL) or MtREVOLUTA1 (MtREV1), the putative orthologs of A. thaliana WUSCHEL and REVOLUTA, of which mutants are altered in auxin homeostasis [75,76]. The ectopic expression of AtSTM could rescue the formation of lateral leaflets in the sgl1 mutant. These data show that AtSTM could substitute for SGL1 via an independent pathway to form complex leaves. This suggests that AtSTM could bypass the requirement for SGL1 during the formation of compound leaves in M. truncatula, indicating shared functions between these proteins, a conclusion further reinforced by the study

Discussion
Here, we compared the activity of two transcription factors, AtSTM and SGL1, in M. truncatula. Our analysis is based on transgenic plants that were able to grow in a greenhouse and therefore expressed AtSTM at low levels. This allowed us to investigate the activity of AtSTM during flower development.
An increase in the leaflet number was only occasionally observed following AtSTM ectopic expression in wild-type M. truncatula. This limited effect of AtSTM could be linked to AtSTM expression levels in these lines, which were low. The additional leaflets were formed at the base of the terminal leaflet and resulted from a secondary morphogenesis. This suggests that AtSTM leads to additional leaflets through the extension of the meristematic activity, allowing more leaflets to emerge, and not from the division of the lateral leaflets into two structures. In M. truncatula, the terminal leaflet derives from the terminal zone where auxin maxima are located through the activity of SMOOTH LEAF MARGIN1 (SLM1), the PIN1 ortholog in M. truncatula [74]. Lateral leaflets result from the marginal blastozone activity and the formation of local auxin maxima that depend on SGL1 activity [74]. The tetrafoliate pattern seen in AtSTM transgenic lines likely results from a defect in auxin distribution in the terminal zone. This leaf patterning is also found in M. truncatula plants inactivated for HEADLESS (HDL) or MtREVOLUTA1 (MtREV1), the putative orthologs of A. thaliana WUSCHEL and REVOLUTA, of which mutants are altered in auxin homeostasis [75,76]. The ectopic expression of AtSTM could rescue the formation of lateral leaflets in the sgl1 mutant. These data show that AtSTM could substitute for SGL1 via an independent pathway to form complex leaves. This suggests that AtSTM could bypass the requirement for SGL1 during the formation of compound leaves in M. truncatula, indicating shared functions between these proteins, a conclusion further reinforced by the study of the floral phenotype of p35S:AtSTM sgl1 plants.
Our data revealed an unexpected effect of AtSTM on floral development, as AtSTM could induce petal identity. The effect of AtSTM on petal identity was moderately visible in an SGL1 wild-type background, as only few chimeric petaloid floral organs were formed, but was dramatic in an sgl1 mutant background. Indeed, all p35S:AtSTM sgl1 flowers produced petals or petals with sepal sectors, while such organs were missing in sgl1. Although the increase in petal number could be in part due to the indeterminate state conferred by the sgl1 mutation, it nevertheless indicates that AtSTM can restore petal formation in an sgl1 mutant. The shape of sgl1 sepals was also restored following AtSTM expression, showing that AtSTM could substitute for other functions of SGL1 during flower development. The formation of petals in p35S:AtSTM sgl1 was correlated with an activation of MtAP1 and more notably of MtPI expression, suggesting that AtSTM could promote the expression of these floral organ identity genes to restore petals, and not through an indirect effect on floral meristem growth, for instance. Such a role for KNOX1 genes in the promotion of B function was not yet reported in either M. truncatula nor in A. thaliana [44,46,72].
On the contrary, p35S:AtSTM sgl1 flowers did not form stamens, and in contrast to sgl1 cauliflowers, which developed carpels, AtSTM sgl1 cauliflowers lacked carpels. MtAGb expression was only rarely detected in p35S:AtSTM sgl1 cauliflowers, in agreement with the lack of carpel identity. Interestingly, the expression of MtAGb was systematically detected in floral meristems beneath the outermost layers in sgl1 background. The localization and the low intensity of the MtAGb signal in sgl1 cauliflowers suggest that SGL1 influences MtAGb expression.
In Arabidopsis, a link for AtSTM with carpel identity was revealed with the analysis of plants compromised for AtSTM activity in line with AtSTM expression in flowers [5,6,71,73]. A more direct contribution to carpel identity was illustrated with the phenotype of A. thaliana KNOX1 overexpressors showing homeotic conversion of ovules into pistils. However, KNOX1 ectopic expression does not complement the ag mutant [52,72]. In line with these conclusions made in Arabidopsis, we observed that in M. truncatula, ectopic expression of AtSTM could not induce the C function in the absence of SGL1 activity. It is possible that in p35S:AtSTM sgl1 flowers, AG is playing a role related to floral meristem termination more than a function related to the specification of carpel identity.
The impact of AtSTM was more obvious both in leaves and flowers of the sgl1 mutant compared with wild-type SGL1 plants. This distinct impact could suggest that the STM pathway is more effective in the absence of SGL1 activity. It is likely that SGL1 acts in part through the M. truncatula UFO ortholog, as it does in Arabidopsis and other legumes. Indeed, in pea and in Lotus japonicus defective in STAMINA PISTILLOIDA (STP) or in PROLIFERATING FLOWER ORGAN (PFO), the A. thaliana UFO orthologs lack petals and stamens and show a reduced carpel formation similar to sgl1 flowers [47,77]. On the other hand, AtSTM was shown recently to function together in A. thaliana with AP1 to specify floral meristem identity in part via UFO [78]. This suggests that SGL1 and STM pathways may converge on MtUFO and that a competition for UFO interaction or for targets shared between SGL1 and AtSTM could be the basis for the higher effect of AtSTM in the absence of SGL1.
Our work shows that AtSTM substitutes for SGL1 function in M. truncatula during both vegetative and reproductive development. A parallel has been proposed between compound leaflet primordia and common primordia formation. Both of these processes seem to require the maintenance of an indeterminate phase controlled by SGL1 [61]. While in leaves, SGL1 maintains the indeterminate state, in flowers, SGL1 acts in opposite by promoting the formation of common primordia. The capacity for AtSTM to substitute for SGL1 in both leaves and flowers underlines this parallel and the control of meristematic activity shared by these two transcription factors.

Plant Growth and Plant Material
M. truncatula plants were grown in a greenhouse or in growth chambers under longday conditions (16 h light at 23 • C and 8 h dark at 15 • C). The wild-type (R108) and the sgl1-1 mutant M. truncatula lines have been described [42].
The pSGL1:GUS reporter construct was generated as follows. A 2.7 kb fragment corresponding to the SGL1 (Medtr3g098560) promoter sequence used in [42] (wild-type M. truncatula cv Jemalong) was amplified from the M. truncatula R108 ecotype using primers pSGL1-for, incorporating a BglII site, and pSGL1-rev, incorporating a BamHI site. The promoter was cloned into pCR Blunt II-TOPO vector to create pTOPO-pSGL1 and sequenced. The pSGL1 promoter was moved into the binary vector pCAMBIA 3301 in front of the β-glucuronidase (GUS) gene. For this, a BglII-BamH1 fragment containing the SGL1 promoter was ligated into pCAMBIA3301 cut with BamHI and BglII to replace the 35S promoter.
The pSGL1:HA-AtSTM construct was generated as follows (AtSTM, AT1G62360). pTOPOpSGL1 was cut with EcoR1-BamH1 to release the pSGL1 promoter, which was cloned into the pCAMBIA 3300 binary vector cut with EcoRI and BamHI to create pCAMBIA 3300 pSGL1. The alli2AtSTM plasmid harboring the triple hemagglutinin (HA) tag-AtSTM fusion under the double enhanced cauliflower Mosaic Virus 35S promoter was used as a template to amplify the HA-AtSTM fusion using primers AtSTM-for and AtSTM-rev incorporating BamHI and EcoRI sites, respectively. This fragment was ligated into the pALC vector (Syngenta Ltd., Jeolotts Hill, UK) cut with BamHI and EcoRI. The BamHI-XbaI fragment containing the HA-AtSTM fusion and the 35S terminator was cloned into pCAMBIA 3300 pSGL1 to create pCAMBIA pSGL1:HA-AtSTM 35S term.
The p35S:HA-AtSTM construct was generated as follows. The pSGL1 promoter sequence of the pCAMBIA pSGL1:HA-AtSTM 35S term was replaced with the 35S promoter sequence from pCAMBIA 3301 using the BglII and BamH1 sites. The pCAMBIA 3301 was cut with BamH1 and BglII to release the 35S promoter, and the pCAMBIA 3300 containing the pSGL1:HA-AtSTM construct was cut with BglII and BamHI to replace the pSGL1 promoter with the 35S promoter to create pCAMBIA p35S:HA-AtSTM 35S term. pSGL1-GUS, p35S:AtSTM and pSGL1:AtSTM constructs were introduced into A. tumefaciens GV3101. The pSGL1-GUS construct was used to transform M. truncatula R108 wild-type plant, while p35S:AtSTM and pSGL1:AtSTM constructs were used to transform M. truncatula R108 plants heterozygous for the sgl1-1 mutation. M. truncatula transgenic lines were created using a leaf disc protocol [79]. Transgenic calli were selected on media containing 3 mgL −1 Basta (glufosinate-ammonium). Primers are listed in Table S1.
Four independent pSGL1:GUS transgenic lines were analyzed for SGL1:GUS activity. The SGL1:GUS activity was detected in meristem, vascular tissue and young leaves in R108 M. truncatula ( Figure S1), which was similar to the activity of the SGL1 promoter isolated from the JemalongA17 ecotype [42], and in axillary meristem, young floral buds and carpels ( Figure S1).
Most of the transgenic plantlets expressing AtSTM were not viable when transferred to soil. RT-PCR were realized to compare the level of expression of AtSTM in transgenic lines. Total RNA was extracted from AtSTM transgenic lines expressing p35S:AtSTM (in vitro seedlings and transgenic plants grown in the greenhouse) using Tri reagent (Sigma-Aldrich, Saint-Quentin-Fallavier, France) and treated with DNAse I (Invitrogen, Waltham, MA, USA) according to the manufacturer's instructions. AtSTM levels were monitored using qAtSTM-F and qAtSTM-R primers. Primers specific for the M. truncatula UBIQUITIN gene (Medtr3g091400) were used as an internal control [80]. Only transgenic plantlets expressing AtSTM at low levels were viable in the greenhouse. Four p35S:AtSTM independent lines and three pSGL1:AtSTM lines were obtained. Of these, two independent p35S:AtSTM lines and one pSGL1:AtSTM based on their phenotype were chosen for further characterization. These plants showed reduced fertility. Plants homozygous for the p35S:AtSTM construct and heterozygous for sgl1 were obtained and confirmed by PCR genotyping [42].

Phenotypic Observations
Leaves and flowers were observed under a binocular microscope (Nikon, SMZ1000) and imaged with a digital camera (ProgRes C10 plus ). M. truncatula meristems showing GUS activity were dissected and photographed using a LeicaMZ12 dissecting microscope fitted with an AxioCam ICc5 digital camera.

Scanning Electron Microscopy (SEM)
Three to eight-week-old plants were dissected to observe leaf and flower primordia. The samples were imaged using SEC DESKTOP SEM (Scanning Electron Microscope, (SNE-1500M), SEC, Suwon, Korea) at an accelerating voltage of 15 kV.

In Situ Localization of GUS Activity and In Situ Hybridization
GUS staining and tissue embedding have been described in [81]. RNA in situ hybridization with digoxigenin-labeled probes was performed as previously described [82]. The RNA antisense and sense probes of MtAP1 (Medtr8g066260) MtPIM, MtPI (Medtr3g088615) and MtAGb (Medtr8g087860) were generated using as cDNA templates a 426 bp fragment of MtPIM (282-707 from ATG), a 298 bp fragment of MtPI (504-801 from ATG) or a 215 bp fragment of MtAGb (558-773 from ATG), respectively, cloned into the pGEM-T Easy vector (Promega, Madison, WI, USA) and using the corresponding SP6 and T7 RNA polymerases in the vector for transcription. SP6 was used for transcription of RNA antisense probes and T7 for the sense. The in situ hybridization with control sense probes is presented in Figure S5.
Author Contributions: All authors made essential contributions to the project. V.P. performed most of the experiments; A.E. made the Medicago truncatula transgenics; T.C. performed the GUS assays, some leaf phenotypic characterization and some SEMs; B.A. provided technical assistance to V.P.; A.B. and F.M. performed the in situ analyses; P.L., P.R. and V.P. designed the research; V.P. and P.L. wrote the article. All authors have read and agreed to the published version of the manuscript.