Systematic Analysis of Gibberellin Pathway Components in Medicago truncatula Reveals the Potential Application of Gibberellin in Biomass Improvement

Gibberellins (GAs), a class of phytohormones, act as an essential natural regulator of plant growth and development. Many studies have shown that GA is related to rhizobial infection and nodule organogenesis in legume species. However, thus far, GA metabolism and signaling components are largely unknown in the model legume Medicago truncatula. In this study, a genome-wide analysis of GA metabolism and signaling genes was carried out. In total 29 components, including 8 MtGA20ox genes, 2 MtGA3ox genes, 13 MtGA2ox genes, 3 MtGID1 genes, and 3 MtDELLA genes were identified in M. truncatula genome. Expression profiles revealed that most members of MtGAox, MtGID1, and MtDELLA showed tissue-specific expression patterns. In addition, the GA biosynthesis and deactivation genes displayed a feedback regulation on GA treatment, respectively. Yeast two-hybrid assays showed that all the three MtGID1s interacted with MtDELLA1 and MtDELLA2, suggesting that the MtGID1s are functional GA receptors. More importantly, M. truncatula exhibited increased plant height and biomass by ectopic expression of the MtGA20ox1, suggesting that enhanced GA response has the potential for forage improvement.


Phylogenetic Analysis and Chromosomal Locations of MtGAox Genes
To establish the evolutionary relationships within the MtGAox genes, a phylogenetic tree was constructed using the 39 GAox proteins from M. truncatula and A. thaliana, including 5 AtGA20ox, 8 MtGA20ox, 4 AtGA3ox, 2 MtGA3ox, 7 AtGA2ox, and 13 MtGA2ox. Based on the phylogenetic analysis, the 39 GAox proteins were divided into four subfamilies ( Figure 1): the gibberellin biosynthesis gene family contained the GA20ox and GA3ox genes, and the gibberellin deactivation GA2ox genes could be divided into C19-and C20-GA2ox subgroups.
Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 3 of 20 total of 23 candidate genes were obtained in the M. truncatula genome, including 8 MtGA20ox, 2 MtGA3ox, and 13 MtGA2ox (Table S1). These MtGAox genes were named based on their closest Arabidopsis orthologs. The gene names, accession numbers, locations, coding sequence (CDS) lengths and the number of the deduced amino acid are summarized in Table S1.

Phylogenetic Analysis and Chromosomal Locations of MtGAox Genes
To establish the evolutionary relationships within the MtGAox genes, a phylogenetic tree was constructed using the 39 GAox proteins from M. truncatula and A. thaliana, including 5 AtGA20ox, 8 MtGA20ox, 4 AtGA3ox, 2 MtGA3ox, 7 AtGA2ox, and 13 MtGA2ox. Based on the phylogenetic analysis, the 39 GAox proteins were divided into four subfamilies (Figure 1): the gibberellin biosynthesis gene family contained the GA20ox and GA3ox genes, and the gibberellin deactivation GA2ox genes could be divided into C19-and C20-GA2ox subgroups.
The 23 MtGAox genes were distributed among 8 chromosomes ( Figure S2A). The chromosome 02, 01 and 08, 07, 03 and 04 had six, four, three, and two genes, respectively, while the chromosome 05 and 06 each contained only one.  (Mt). The proteins are grouped into four distinct groups. The phylogenetic tree was constructed using 39 protein sequences from A. thaliana (16) and M.truncatula (23) on iTOLv5 (Interactive Tree of Life; https://itol.embl.de/) online website. The red dots of different sizes (Min size: 5 px; Max size: 30 px) represent the bootstrap values, ranging from 0.5 to 1. The branches covered by different color represent the four group proteins.

Gene Structures and Conserved Motifs of MtGAox in M. truncatula
To get a better understanding of the structural diversity of the MtGAox genes, the structures (exon/intron organization) of MtGAox were analyzed ( Figure S2B). MtGA20ox and MtGA2ox genes The 23 MtGAox genes were distributed among 8 chromosomes ( Figure S2A). The chromosome 02, 01 and 08, 07, 03 and 04 had six, four, three, and two genes, respectively, while the chromosome 05 and 06 each contained only one.

Gene Structures and Conserved Motifs of MtGAox in M. truncatula
To get a better understanding of the structural diversity of the MtGAox genes, the structures (exon/intron organization) of MtGAox were analyzed ( Figure S2B). MtGA20ox and MtGA2ox genes had the same structure, which all contain three exons. However, all the MtGA3ox genes had two exons. Exon-intron organization indicated that MtGA20ox and MtGA2ox genes were more closely related than MtGA3ox genes.
To obtain a better understanding of the protein sequence characteristics of the MtGAox, MEME search was used to analyze the conserved domains and motifs. Based on the composition of motifs, the MtGAox proteins contained nine different motifs (Figure 2; Figure S3). Motif 1, 2, 3, 4, and 6 were identified as the conserved motif, which was present in every MtGAox protein. Motif 1 and 2 are representative 2OG-FeII_Oxy domain and Motif 3 is representative DIOX_N domain, which are identified in all GAox proteins. Motif 5 was only identified in C-terminal of MtGA20ox proteins, indicating that members within a single subfamily usually exhibit similar motif composition. However, motif 8 was located in the C-terminal part of some members of the MtGA2ox protein family, demonstrating that this motif is related to the functions of these MtGA2ox proteins.

Expression Patterns of MtGAox Genes
The tissue-specific expression patterns of a gene family can provide information about its possible function. To understand the functional divergence of the MtGAox genes, we analyzed the expression profiles of the MtGAox genes in different organs/tissues by quantitative RT-PCR (qRT-PCR) and gene expression atlas (Figure 3; Figures S4-S5). The results indicated that GA biosynthesis gene MtGA20ox1, MtGA20ox4, MtGA20ox5, MtGA20ox7, and MtGA3ox1 were widely expressed in all tissues examined ( Figure 3A). While, MtGA20ox3 and MtGA20ox6 had higher expression levels in stems, MtGA20ox2, MtGA20ox8, and MtGA3ox2 had higher expression levels in

Expression Patterns of MtGAox Genes
The tissue-specific expression patterns of a gene family can provide information about its possible function. To understand the functional divergence of the MtGAox genes, we analyzed the expression profiles of the MtGAox genes in different organs/tissues by quantitative RT-PCR (qRT-PCR) and gene expression atlas (Figure 3; Figures S4 and S5). The results indicated that GA biosynthesis gene MtGA20ox1, MtGA20ox4, MtGA20ox5, MtGA20ox7, and MtGA3ox1 were widely expressed in all tissues examined ( Figure 3A). While, MtGA20ox3 and MtGA20ox6 had higher expression levels in stems, MtGA20ox2, MtGA20ox8, and MtGA3ox2 had higher expression levels in flowers, pods, and roots ( Figure 3A), respectively, implying important roles of these genes in certain organs or tissues. Most of the GA deactivation genes, such as MtGA2ox1, MtGA2ox2, MtGA2ox6, MtGA2ox8, MtGA2ox9, MtGA2ox10, MtGA2ox11, MtGA2ox12, and MtGA2ox13 were widely expressed in all organs tested ( Figure 3B). However, MtGA2ox4 and MtGA2ox7 were expressed at higher levels in flowers than those in other tissues ( Figure 3B), implying their possible involvement in flower development.

Identification of GA Receptor MtGID1 and MtDELLA Genes in M. truncatula
When GA is present, the GID1 receptors bind to GA and this binding promotes the interaction of GID1 with DELLA proteins, which are degraded in the 26S proteasome later ( Figure S6). To identify MtGID1 and MtDELLA proteins in M. truncatula, BLASTP searches were performed using 3 AtGID1 and 5 AtDELLA sequences. A total of three GA receptor proteins, MtGID1A, MtGID1B, and MtGID1C, and 3 MtDELLA proteins, MtDELLA1, MtDELLA2, and MtDELLA3, were identified

Identification of GA Receptor MtGID1 and MtDELLA Genes in M. truncatula
When GA is present, the GID1 receptors bind to GA and this binding promotes the interaction of GID1 with DELLA proteins, which are degraded in the 26S proteasome later ( Figure S6). To identify MtGID1 and MtDELLA proteins in M. truncatula, BLASTP searches were performed using 3 AtGID1 and 5 AtDELLA sequences. A total of three GA receptor proteins, MtGID1A, MtGID1B, and MtGID1C, and 3 MtDELLA proteins, MtDELLA1, MtDELLA2, and MtDELLA3, were identified (Table S2; Figure 4A; Figures S7 and S8;). Gene structure analysis showed that the MtDELLA genes had the same exon-intron organization with no intron, whereas the intron of MtGID1 genes ranged from 1 to 2 ( Figure 4B).  (Table S2; Figure 4A; Figures S7 and S8;). Gene structure analysis showed that the MtDELLA genes had the same exon-intron organization with no intron, whereas the intron of MtGID1 genes ranged from 1 to 2 ( Figure 4B).

Expression Patterns of MtGID1 and MtDELLA Genes
To gain better insights into the potential functions of MtGID1 and MtDELLA genes, their expression patterns were analyzed ( Figure 5; Figures S9 and S10). The qRT-PCR results indicated that MtDELLA1 and MtDELLA2 had similar expression patterns and were detected at low levels in all tissues, indicating that MtDELLA1 and MtDELLA2 are involved in many aspects of growth and development. Moreover, MtDELLA3 had higher expression level in pods compared to that in root and stems ( Figure 5A), implying that MtDELLA3 may play more important roles in pod development. The expression levels of MtGID1B and MtGID1C were higher in root than those in other tissues ( Figure 5B), demonstrating that MtGID1B and MtGID1C may play a significant role in

Expression Patterns of MtGID1 and MtDELLA Genes
To gain better insights into the potential functions of MtGID1 and MtDELLA genes, their expression patterns were analyzed ( Figure 5; Figures S9 and S10). The qRT-PCR results indicated that MtDELLA1 and MtDELLA2 had similar expression patterns and were detected at low levels in all tissues, indicating that MtDELLA1 and MtDELLA2 are involved in many aspects of growth and development. Moreover, MtDELLA3 had higher expression level in pods compared to that in root and stems ( Figure 5A), implying that MtDELLA3 may play more important roles in pod development. The expression levels of MtGID1B and MtGID1C were higher in root than those in other tissues ( Figure 5B), demonstrating that MtGID1B and MtGID1C may play a significant role in root development than other tissues. However, MtGID1A was widely expressed in all tissues tested ( Figure 5B), although at low level, suggesting that it may be involved in multiple developmental processes.

The GA Receptors, MtGID1s, Were Able to Interact with MtDELLA Proteins
In DELLA protein, two important and conserved motifs, DELLA and VHYNP, are involved in the interaction with GID1 proteins [55,56]. In addition, the SUMO-Interaction Motif (SIM), which is important for the recognition of SUMOylated DELLA proteins, is also highly conserved between GID1s [57,58]. Multiple sequences alignment showed that the three MtDELLA proteins had the typical DELLA and VHYNP domains ( Figure 6A; Figure S11) and the three MtGID1 receptors had the conserved DELLA interaction SIM motif ( Figure 6B; Figure S11). To verify that the MtDELLA and MtGID1 proteins are functional GA signaling components, we assessed the interaction between MtDELLAs and MtGID1s using the yeast two-hybrid system (Figure 7). We also performed the interaction tests between MtDELLAs and MtGID1s in opposite orientation and obtained the same results ( Figure S12). MtGID1s interacted with MtDELLA1 and MtDELLA2 in yeast cells in the presence of GA3, but not in GA3-free medium, while MtGID1s could not interact with MtDELLA3.
These results indicate that all three MtGID1s are functional GA receptors and MtDELLA1 and MtDELLA2 may play more important roles in the GA pathway. In DELLA protein, two important and conserved motifs, DELLA and VHYNP, are involved in the interaction with GID1 proteins [55,56]. In addition, the SUMO-Interaction Motif (SIM), which is important for the recognition of SUMOylated DELLA proteins, is also highly conserved between GID1s [57,58]. Multiple sequences alignment showed that the three MtDELLA proteins had the typical DELLA and VHYNP domains ( Figure 6A; Figure S11) and the three MtGID1 receptors had the conserved DELLA interaction SIM motif ( Figure 6B; Figure S11). To verify that the MtDELLA and MtGID1 proteins are functional GA signaling components, we assessed the interaction between MtDELLAs and MtGID1s using the yeast two-hybrid system (Figure 7). We also performed the interaction tests between MtDELLAs and MtGID1s in opposite orientation and obtained the same results ( Figure S12). MtGID1s interacted with MtDELLA1 and MtDELLA2 in yeast cells in the presence of GA 3 , but not in GA 3 -free medium, while MtGID1s could not interact with MtDELLA3. These results indicate that all three MtGID1s are functional GA receptors and MtDELLA1 and MtDELLA2 may play more important roles in the GA pathway.
MtDELLAs and MtGID1s using the yeast two-hybrid system (Figure 7). We also performed the interaction tests between MtDELLAs and MtGID1s in opposite orientation and obtained the same results ( Figure S12). MtGID1s interacted with MtDELLA1 and MtDELLA2 in yeast cells in the presence of GA3, but not in GA3-free medium, while MtGID1s could not interact with MtDELLA3. These results indicate that all three MtGID1s are functional GA receptors and MtDELLA1 and MtDELLA2 may play more important roles in the GA pathway.

The Transcript Level of MtGAoxs is Regulated by a Feedback Mechanism
Previous studies indicated that feedback regulation controls the concentrations of active GAs in higher plants and the gibberellin metabolism genes also respond to exogenous GA treatment [59,60]. To investigate whether the expression level of MtGA20ox, MtGA3ox, and MtGA2ox are regulated by a feedback mechanism, leaves were treated with GA3 and gibberellin inhibitor PAC, Figure 7. Interaction tests between MtDELLA and MtGID1 proteins in the yeast two-hybrid system. Yeast transformants are spotted onto the control medium (SD/-Leu/-Trp) and selective medium (SD/-Leu/-Trp/-His/-Ade). The initial concentration of the yeast cells spots on SD/-Trp/-Leu (panel 1) and SD/-Trp/-Leu/-His/-Ade medium (panels 2 and 5) were OD600 = 0.2. Then, the yeast cells were diluted 10 and 100 times and were plated onto selective medium (panels 3, 4, 6 and 7) containing 20 µg/mL X-α-gal with or without GA 3 (10 −5 M).

The Transcript Level of MtGAoxs Is Regulated by a Feedback Mechanism
Previous studies indicated that feedback regulation controls the concentrations of active GAs in higher plants and the gibberellin metabolism genes also respond to exogenous GA treatment [59,60]. To investigate whether the expression level of MtGA20ox, MtGA3ox, and MtGA2ox are regulated by a feedback mechanism, leaves were treated with GA 3 and gibberellin inhibitor PAC, and the transcriptional changes were analyzed using qRT-PCR (Figure 8). qRT-PCR results showed that the transcript levels of MtGA20ox1, MtGA20ox2, MtGA20ox6 and MtGA3ox1 were downregulated after 24 h treatment of 50 µM and 100 µM GA 3 ( Figure 8A), indicating that these genes respond quickly to exogenous GA 3 treatment. On the opposite, the expressions of MtGA20ox3, MtGA20ox4, MtGA20ox5, MtGA20ox8, and MtGA3ox2 were upregulated when treated with 100 µM GA 3 ( Figure 8A), implying that upregulation of these genes may compensate for the downregulation of MtGA20ox1, MtGA20ox2, MtGA20ox6, and MtGA3ox1 genes to balance the GA level in plant. However, the transcript levels of MtGA20ox5, MtGA20ox8, MtGA3ox1, and MtGA3ox2 were upregulated when treated with 100 µM PAC ( Figure 8A), indicating that these genes may play important roles in GA synthesis. Meanwhile, GA treatment promoted the transcription of some MtGA2ox genes, such as MtGA2ox2, MtGA2ox3, MtGA2ox4, MtGA2ox5, MtGA2ox6, MtGA2ox7, MtGA2ox10, and MtGA2ox13 ( Figure 8B), demonstrating that these genes may have key roles in GA inactivation. After treatment with PAC, the downregulated genes were MtGA2ox1, MtGA2ox2, MtGA2ox4, MtGA2ox10, MtGA2ox11, and MtGA2ox12 ( Figure 8B). These data showed that MtGA20oxs are responsible for the regulation of bioactive GA synthesis and MtGA2oxs regulate the degradation of GA.

GA Has a Positive Effect on Growth and Biomass of M. truncatula and Alfalfa
To further investigate the in vivo function of GA, wild-type Medicago plants were transformed with 35S:MtGA20OX1 ( Figure S13A). Fourteen regeneration lines were obtained and eleven were positive by PCR analysis ( Figure S13B). Quantitative real-time PCR data showed that transcripts of MtGA20OX1 were significantly increased in these transgenic plants, and the overexpression lines (OE) 3 and 9 were selected for further analysis ( Figure S12C). To evaluate the effects of MtGA20OX1 expression on Medicago development, the following traits were measured: leaf size, plant height, and biomass. Under the same growth conditions, the MtGA20OX1 overexpression lines exhibited larger cotyledon and leaf size, compared with those of the controls (Figure 9A-C). Moreover, the plant height of MtGA20OX1 transgenic lines was measured. The MtGA20OX1 overexpression lines showed an increase in plant height, compared with that of the controls (Figure 9D-E). Then, we evaluated total above-ground biomass yield of 10 weeks old wild-type and MtGA20OX1 transgenic plants. The average fresh and dry weight of the MtGA20OX1 transgenic plants had a significant increase in total biomass, compared with those of wild-type ( Figure 9F-G).

Transcription Analysis of Cell Development Related Genes in Transgenic Lines
To explore the molecular mechanism of the increasing biomass of MtGA20OX1-overexpressing plants, the expression level of cell division related genes were examined. The results showed that the transcript levels of MtCYCB1.1, MtCYCB1.2, and MtCYCD2.1 were up-regulated in MtGA20OX1-overexpressing plants ( Figure 10A-C), whereas MtKRP1, MtKRP2, and MtKRP3 genes encoding the cyclin-dependent kinase inhibitor, were down-regulated ( Figure 10D-F). These results suggest that overexpression of MtGA20OX1 might promote plant growth by controlling cell division.

Transcription Analysis of Cell Development Related Genes in Transgenic Lines
To explore the molecular mechanism of the increasing biomass of MtGA20OX1-overexpressing plants, the expression level of cell division related genes were examined. The results showed that the transcript levels of MtCYCB1.1, MtCYCB1.2, and MtCYCD2.1 were up-regulated in MtGA20OX1-overexpressing plants ( Figure 10A-C), whereas MtKRP1, MtKRP2, and MtKRP3 genes encoding the cyclin-dependent kinase inhibitor, were down-regulated ( Figure 10D-F). These results suggest that overexpression of MtGA20OX1 might promote plant growth by controlling cell division.

Discussion
Gibberellins (GAs) play a crucial role in regulating plant growth and development [9,[61][62][63]. However, to date, little was known about the gene families involved in GA biosynthesis, catabolism, and signaling in M. truncatula. In this study, 8 MtGA20ox, 2 MtGA3ox, 13 MtGA2ox, 3 MtGID1, and 3 MtDELLA genes were identified in M. truncatula through bioinformatics analysis. Moreover, we evaluated the developmental responses of Medicago and alfalfa to GA3 and PAC treatment.
The MtGAox genes were distributed among chromosomes with different densities. The number of MtGA20ox and MtGA2ox genes was 1.75-fold larger than that in Arabidopsis, suggesting that the MtGA20ox and MtGA2ox ancestor may have experienced gene expansion and there may be gene redundancies in functions [14,15,64,65]. However, the GA3ox subfamily included 4 AtGA3ox and 2 MtGA3ox proteins, suggesting that the MtGA3ox family have experienced gene loss during evolution compared with that of Arabidopsis. Gene structure analysis indicated that the exon-intron organization of GAox members in the same group were highly conserved in Medicago, Arabidopsis, and rice [66].
The patterns of gene structure diversity and motif composition can provide significant evidence of the evolutionary relationships of multi-gene families [67,68]. Gene structure and motif analysis revealed that the most closely related members of the MtGAoxs family showed similar exon-intron organization and motif distribution, implying that they is functional conservation in the gibberellin metabolic pathway. Specific motifs in amino acid sequences are important regions for their

Discussion
Gibberellins (GAs) play a crucial role in regulating plant growth and development [9,[61][62][63]. However, to date, little was known about the gene families involved in GA biosynthesis, catabolism, and signaling in M. truncatula. In this study, 8 MtGA20ox, 2 MtGA3ox, 13 MtGA2ox, 3 MtGID1, and 3 MtDELLA genes were identified in M. truncatula through bioinformatics analysis. Moreover, we evaluated the developmental responses of Medicago and alfalfa to GA 3 and PAC treatment.
The MtGAox genes were distributed among chromosomes with different densities. The number of MtGA20ox and MtGA2ox genes was 1.75-fold larger than that in Arabidopsis, suggesting that the MtGA20ox and MtGA2ox ancestor may have experienced gene expansion and there may be gene redundancies in functions [14,15,64,65]. However, the GA3ox subfamily included 4 AtGA3ox and 2 MtGA3ox proteins, suggesting that the MtGA3ox family have experienced gene loss during evolution compared with that of Arabidopsis. Gene structure analysis indicated that the exon-intron organization of GAox members in the same group were highly conserved in Medicago, Arabidopsis, and rice [66].
The patterns of gene structure diversity and motif composition can provide significant evidence of the evolutionary relationships of multi-gene families [67,68]. Gene structure and motif analysis revealed that the most closely related members of the MtGAoxs family showed similar exon-intron organization and motif distribution, implying that they is functional conservation in the gibberellin metabolic pathway. Specific motifs in amino acid sequences are important regions for their functions. All of the MtGAox proteins contained the conserved 2OG-FeII_Oxy and DIOX_N domains.
Gene expression patterns analysis can be used to predict the molecular functions of genes in the development of different organs or tissues. In our study, the GA biosynthesis genes, MtGA3ox2, MtGA20ox3, MtGA20ox2, and MtGA20ox8, were specifically expressed in root, stem, flower, and pod. Further, additional studies are needed to explore the function of these genes in specific organs in more detail. However, the GA deactivation genes were more widely expressed in all organs, demonstrating that they may play critical roles in processes involved in growth and development of M. truncatula. MtDELLA1, MtDELLA2, and MtGID1A were highly expressed in all organs, indicating that these genes might play more important roles in GA signaling transduction in M. truncatula.
Previous studies have indicated that the feedback regulation mechanism controls the concentration of active GAs in higher plants [69,70]. In most plants, GA20ox and GA3ox genes were involved in the final steps of GA biosynthesis and the expression of GA20ox and GA3ox was downregulated by exogenous GA treatment [59,70]. In contrast, the transcript of GA2ox genes was upregulated by GA treatment [65]. In Vitis Vinifera L. and Camellia sinensis (L.) O. Kuntze, the expression levels of VvGA20ox1, VvGA20ox2, CsGA20ox2, and CsGA20ox3 were downregulated under GA 3 treatment and the same was observed for the expression of MtGA20ox1, MtGA20ox2, and MtGA3ox1 in M. truncatula. In addition, most of the genes encoding MtGA2ox proteins were upregulated by GA 3 treatment and only two genes, MtGA2ox2 and MtGA2ox4, can respond to both GA 3 and PAC treatments. Our results demonstrated that the feedback regulation of GA in M. truncatula is similar to that of previous studies, indicating that the feedback regulation mechanism is conserved and the MtGAox genes are functional components of the GA metabolic pathway [69].
The GID receptors and DELLA proteins are master components of GA signaling [29,31,71]. Phylogenetic analyses showed that the Medicago genome contains three GA receptors (MtGID1A, MtGID1B, and MtGID1C) and three DELLA proteins (MtDELLA1, MtDELLA2, and MtDELLA3). The number of MtGID1s is identical to those in Arabidopsis and tomato, indicating that the evolutionary mechanisms of MtGID1s are more conservative than MtDELLAs [32,35,57]. In rice, GID1 interacts with DELLA protein SLR1 in a GA-dependent manner in yeast cells [22]. Moreover, AtGID1s can interact with AtDELLA proteins yeast cells in the presence of GA 4 [33]. In M. truncatula, MtGID1s physically interacts with MtDELLA1 and MtDELLA2 in the yeast two-hybrid system in the presence of GA 3 . Although MtDELLA2 and MtGID1C had a weak interaction at the initial concentration (10 • ) in the absence of GA 3 , further dilution and configuration tests showed that the interaction between MtDELLA2 and MtGID1C is mainly in a GA-dependent manner. These results demonstrate that MtGID1s, MtDELLA1, and MtDELLA2 are functional GA signal components. However, MtGID1s and MtDELLA3 cannot interact with or without GA 3 . Several studies have elucidated that GA regulation of nodulation depends on DELLA proteins in legumes [72][73][74]. It is likely that MtDELLA3 may participate in symbiotic nodulation in a GA-independent manner.
In tomato plants with compound leaves, exogenous GA application resulted in less complex leaves with smooth leaf margins [47,75]. However, leaf complexity is not suppressed by GA treatment in M. truncatula and alfalfa, whose leaves consist of three leaflets, indicating that the effects of GA on leaf complexity may be species-dependent. Furthermore, GA promotes stem elongation in M. truncatula and this function is conserved in multiple species [54,[76][77][78][79][80]. It has long been known that GA controls plant growth and development by regulating cell expansion, cell elongation, and cell division. In this study, the MtGA20OX1-overexpressing plants showed increased leaf size compared to the wild-type. The cyclin-dependent protein kinase genes were up-regulated in transgenic lines, while, the inhibitor genes were down-regulated. In the future, genetic improvement of alfalfa may be achieved by enhancing GA biosynthesis or signals, such as ectopic expression of GA20ox and receptors GID1s or mutation in GA2oxs and DELLA.

Plant Materials and Treatments
M. truncatula wild-type (R108 ecotype) and 35S:MtGA20OX1 seeds were scarified and germinated at 4 • C for one week. The geminated seeds were transferred to nursery seedling plate (4 × 4 × 4 cm Length, Width, Height) for 2 weeks in a light incubator (day: 24 • C, 16 h; night: 22 • C, 8 h; relative humidity: 75%). Then, the seedlings were transferred to pots (15 × 15 × 25 cm Length, Width, Height) with soil mix (soil: vermiculite = 3:1) and grown in a greenhouse at 24 • C (day)/22 • C (night) with 16 h (day)/8 h (night) photoperiod, and relative humidity at 70 to 80%. For GA and paclobutrazol (PAC) treatment, GA 3 and PAC were dissolved in ethanol at a stock concentration of 100 mM and 10 mM. One-month-old wild-type plants grown in pots were treatment with GA 3 and PAC at the final concentration of 100 µM and 10 µM for four weeks. Distilled water was used as a control. Each treatment was performed with 20 biological triplicates. To identify the GA metabolism and signaling genes in M. truncatula genome, we used 5 AtGA20ox, 4 AtGA3ox, 7 AtGA2ox, 3 AtGID1, and 5 AtDELLA protein sequences to BLAST the Medicago truncatula genome sequence (www.medicagogenome.org/). The Arabidopsis GA metabolism and signaling genes were obtained from TAIR (www.arabidopsis.org/). Totally, 8 MtGA20ox, 2 MtGA3ox, 13 MtGA2ox, 3 MtGID1, and 3 MtDELLA genes were identified in M. truncatula genome using blast with an E-value <1 × 10 −4 . The protein sequences of these genes were listed in Table S4.
Multiple protein sequences alignment was performed using Jalview 2.10.5 software (www.jalview. org/). The phylogenetic tree of DELLA and GID1 proteins from A. thaliana and M.truncatula was constructed by MEGA 7.0 using the Neighbor-Joining (NJ) method with the following parameters: Poisson correction, pair-wise deletion and bootstrap values in percentages with 1000 replicates. The phylogenetic tree of GAoxs proteins from A. thaliana and M.truncatula was constructed on iTOLv5 (Interactive Tree of Life; https://itol.embl.de/) online website.

Chromosome Location, Gene Structure, and Motif Detection
The information, including chromosome length, gene location, and gene length, of MtGA20ox, MtGA3ox, and MtGA2ox genes on the chromosome were obtained from the M. truncatula genome database. MapGene (http://mg2c.iask.in/mg2c_v2.1/) was used to visualize the chromosomal distribution of these genes in M. truncatula genome.
The exon-intron structures of the MtGA20ox, MtGA3ox, MtGA2ox, MtGID1, and MtDELLA genes were drawn using the online GSDS 2.0 website (http://gsds.cbi.pku.edu.cn/) by comparing the coding sequences with their corresponding genomic sequences. The conserved protein motifs were analyzed using the MEME online tool (http://meme-suite.org/) with the following parameters: maximum number of motifs of 20 and the optimum width from 6 to 200 residues.

RNA Extraction and qRT-PCR
To detect the relative expression levels of MtGA20ox, MtGA3ox, MtGA2ox, MtGID1, and MtDELLA genes in root, shoot, leaf, flower, stem, and pod, total RNA was extracted from these tissues on two-month-old wild-type plants. To detect the effects of GA 3 and PAC on the expression of MtGA20ox, MtGA3ox, and MtGA2ox, total RNA of leaves was extracted from plants after 24 h GA 3 and PAC treatment. To detect the relative expression levels of MtGA20ox1, MtCYCB, MtCYCD, and MtKPR, total RNA of leaves was extracted from two-month-old wild-type and overexpression plants.
Total RNA was extracted using a Plant RNA Kit (TransGene Biotech, Beijing, China) following the manufacturer's instructions. The concentration and quality of the extracted RNA were evaluated using Nanodrop 2000 Spectrophotometer (Thermo Scientific, USA). Reverse transcription of RNA to cDNA was performed with 1 µg total RNA using an iScript cDNA Synthesis Kit (Bio-Rad, Richmond, CA, USA). qRT-PCR was carried out in triplicate for each sample on a CFX Connect™ Detection System (Bio-Rad, Richmond, CA, USA) using TaKaRa SYBR Premix Kit (TaKaRa, Japan). MtUBI gene was used as the internal reference gene. The relative expression levels of the genes were calculated using the 2 -∆∆CT method. Primer sequences used for qRT-PCR analysis were listed in Table S3.

Vector Construction and Plant Transformation
To obtain the MtGA20OX1 overexpression construction, the 1149 bp CDS sequence was amplified using primer pair MtGA20OX1-F/MtGA20OX1-R. The MtGA20OX1 CDS sequence was transferred into the pB7WG2D vector by Gateway LR reaction (Invitrogen, Carlsbad, CA, USA). Then 35S:MtGA20OX1 destination construct was introduced into Agrobacterium strain EHA105. For stable transformation, leaves of wild-type were transformed with EHA105 strain containing the 35S:MtGA20OX1 vector. Primers used are listed in Table S3.

Yeast Two-Hybrid
The yeast two-hybrid assay was performed using the Gold System (Clontech, Mountain View, CA, USA). The coding sequence of MtDELLA1, MtDELLA2, and MtDELLA3 were cloned into the pGADT7 and pGBKT7 vectors, and the coding sequences of MtGID1A, MtGID1B, and MtGID1C were cloned into the pGADT7 and pGBKT7 vectors using the Gateway system. The bait and prey plasmids were co-transformed into yeast strain AH109. For the auxotrophic assay, yeast colonies were inoculated onto SD/-Trp/-Leu and SD/-Trp/-Leu/-His/-Ade plates with or without 10 −5 M GA 3 and incubated in the dark at 28 • C for 3 days. The initial concentration of the yeast cells spots on both SD/-Trp/-Leu and SD/-Trp/-Leu/-His/-Ade plates were OD600 = 0.2. Then, yeast cells were diluted 10 and 100 times and spotted on SD/-Trp/-Leu/-His/-Ade selective media for detail interaction test analysis. All primers used are listed in Table S3.

Statistical Analysis
Statistical analysis was performed using the SPSS 22.0 statistical software (International Business Machines Corporation, Amonk City, NY, USA). The data relating to expression level and size were subjected to one-way ANOVA of variance followed by a comparison of the means according to a significant difference tested at p < 0.05. The data are expressed as the mean ± SE.