Seventy-two MADS-box genes were identified from the current version of S. miltiorrhiza genome (Supplementary Table S1 and Fig. 1). These 72 SmiMADSes could be classified into two types according to their phylogenetic relationships with Arabidopsis and rice, as follows: type I (18, Supplementary Table S2) and type II (54). Type I was further categorised into Mα (12), Mβ (2) and Mγ (4). Type II was assigned to MIKC* (6) and MIKCc (48). MIKCc could be further assigned to 12 subfamilies (Supplementary Table S2). In comparison with MADS-box genes in Arabidopsis, the number of type I genes in S. miltiorrhiza greatly decreased, with 44 fewer than in Arabidopsis (Supplementary Table S2). This result suggested that type I SmiMADSes showed a higher gene death rate and that S. miltiorrhiza effectively used type I to regulate female gametogenesis and seed development. The number of type II genes in S. miltiorrhiza was relatively stable, with nine more than that in Arabidopsis. These findings showed that type II genes in S. miltiorrhiza were more conserved than type I genes, which might be due to the adaptive evolution of this plant.
MADS-box genes with different structures encode transcriptional regulators with various functions, especially during floral development23. In general, MADS-box genes are composed of four domains, namely, MADS, I, K and C24. Among them, MADS domain is the most conserved region. It is presumed that the I domain is responsible for protein dimer formation. The K domain is responsible for protein dimerization, and the C-terminal domain may be responsible for transcriptional activation and protein complex formation25. In terms of both the conserved domain and exon-intron structure of SmiMADSes, type I SmiMADSes were the relatively simple MADS-box proteins that contained one to three motifs (Fig. 3). Sixteen out of 18 type I SmiMADSes, except for SmiMADS11 and SmiMADS18, were intronless (Fig. 2). This simple structure might facilitate their role in evolution and floral organ development26. In contrast, the structure of type II SmiMADSes was more complex. Most of them contained three to six motifs; 96% (52/54) of type II SmiMADSes, except for SmiMADS21 and SmiMADS22, contained two or more introns. The variable number of coding exons in type II SmiMADSes could contribute greatly to the high adaptability of S. miltiorrhiza10. Type II SmiMADSes with multiple exon–intron structures may experience less selection pressure than type I SmiMADSes.
The 72 SmiMADSes were localised on 10 genome scaffolds (Fig. 4). The maximum number included 15 genes (20.8%) localised on scaffold GWHAOSJ00000023, followed by 13 (18.1%) on GWHAOSJ00000057, 12 (16.7%) on GWHAOSJ00000020 and 9 (12.5%) on GWHAOSJ00000084. Only one MADS-box gene was localised on scaffolds GWHAOSJ00000632 and GWHAOSJ00000836. MCScanX analysis revealed that SmiMADSes have a strong evolutionary homology (Fig. 5 and Supplementary Table S4). Nineteen collinear blocks in S. miltiorrhiza genome involved 29 MADS-box genes; of these, 16 blocks belonged to inter-scaffold duplications, and three blocks belonged to intra-scaffold duplications. If these scaffolds are from different chromosomes, so many inter-scaffold duplications may be the result of chromosomal cross over, exchange, recombination, transposable elements and long terminal repeat retrotransposons during the evolution of S. miltiorrhiza10. Additionally, to investigate the evolutionary position of SmiMADSes, an ML phylogenetic tree of MADS-box genes in rice, Arabidopsis and S. miltiorrhiza genomes was constructed, as shown in Fig. 1. The result showed that all S. miltiorrhiza MADS-box genes, except for the MIKC*, DEF + GLO and ANR1 subfamilies, were linked more tightly with A. thaliana than with rice. Many closely related orthologous MADSes, such as SmiMADS1 with AGL61, SmiMADS5 with AGL62, SmiMADS25 with AGL32, SmiMADS41 with AGL44 and SmiMADS45 and 46 with AG (Fig. 1), exist in S. miltiorrhiza and A. thaliana genomes. The interspecific collinearity analysis among S. miltiorrhiza and two model plants (O. sativa and A. thaliana) showed that more MADS-box homologous gene pairs are present between S. miltiorrhiza and A. thaliana than between S. miltiorrhiza and O. sativa (Fig. 6 and Supplementary Tables S5 and S6). This close relationship between S. miltiorrhiza and A. thaliana may be due to the fact that they are both dicotyledons. Ancestral MADS genes were present before S. miltiorrhiza and A. thaliana diverged.
Floral organs are classified into four types, namely, sepals, petals, stamens and pistils (carpels). Their development is regulated by four different classes of floral organ identity genes according to the proposed ABCE model27. In summary, for Arabidopsis, sepal formation is specified by the class A genes AP1/AGL7 and AP228; petal formation is specified by the class A genes and two class B genes (AP3 and PI)29; stamen formation requires AP3 and PI together with the class C gene AG; and carpel formation is controlled by the AG alone30. Proper development of all four whorls of floral organs also requires the class E SEP genes (SEP1/AGL2, SEP2/AGL4, SEP3/AGL9 and SEP4/AGL3)31. All class A, B, C and E genes encode MADS-box transcription factors, except for AP2, which encodes a transcription factor belonging to the super-family of AP2/ERF proteins. From this, we conducted the RNA-seq analysis of RNA samples extracted from four flower tissues, namely, sepals, petals, stamens and pistils (carpels), to explore the tissue-specific expression of SmiMADSes and to find some genes that potentially regulate flower growth and development. In combination with the phylogenetic relationship in Fig. 1, we found that SmiMADS65 in SQUA subfamily had a high and specific expression in sepals and petals (Supplementary Table S9). Thus, it was predicted to be one of the S. miltiorrhiza A-class homeotic genes that regulate sepal and petal formation (Table 1). Three genes in DEF + GLO subfamily (SmiMADS26, SmiMADS28 and SmiMADS29) were strongly and specifically expressed in petals and stamens and can be recognised as the S. miltiorrhiza B-class homeotic genes that potentially regulate petal and stamen formation. The specific expressions of SmiMADS45 and SmiMADS46 (AG subfamily) were found in both stamens and carpels; these were regarded as the S. miltiorrhiza C-class homeotic genes that control stamen and carpel formation. SmiMADS70, SmiMADS71 and SmiMADS72 (SEP subfamily) were suspected to be the S. miltiorrhiza E-class homeotic genes because of their close relationships with Arabidopsis SEP1/AGL2, SEP2/AGL4, SEP3/AGL9 and SEP4/AGL3 and high expression in all parts of S. miltiorrhiza flowers (Table 1). Moreover, nine type II genes (predicted A-class: SmiMADS65 in SQUA subfamily; B-class: SmiMADS26, SmiMADS28 and SmiMADS29 in DEF + GLO subfamily; C-class: SmiMADS45 and SmiMADS46 in AG subfamily; E-class: SmiMADS70, SmiMADS71 and SmiMADS72 in SEP subfamily) were strongly expressed in all flowering stages of S. miltiorrhiza (Supplementary Table S11 and Supplementary Figure S5). At present, our team is conducting a functional validation analysis on the roles of the S. miltiorrhiza floral organ genes mentioned above to elucidate the molecular mechanisms regulating S. miltiorrhiza’s flower growth and development.
Table 1
MADS-box genes and their functions
MADS-box gene
|
Potential functions
|
Tissue-specificity of expression
|
Predicted homologous genes in S. miltiorrhiza
|
Homologous genes in other plants
|
FLC/AGL25
|
Control of flowering time; flowering suppressor35
|
widely expressed in floral organs before flowering except in shoot apex36
|
SmiMADS60-63
|
Cereal plants (wheat, barley, etc.) do not have FLC homologous genes37
|
SVP/AGL22
|
Control of flowering time; flowering suppressor38
|
mostly in apical meristem of inflorescence, buds and leaves before flowering39
|
SmiMADS33-34
|
Barley: BM1, BM10 and HvVRT240; Wheat: TaVRT241
|
SOC1/AGL20
|
Control of flowering time; flowering activator42
|
mostly in shoot apical meristem, leaves and flower buds43
|
SmiMADS57
|
Brassica: LF, MF1 and MF244
|
FUL/AGL8
|
Control of flowering time; regulate floral meristem differentiation; fruit development45
|
floral meristem, ovules, cauline leaves45
|
SmiMADS66-67
|
Paralogs of AP1
|
AGL24
|
Control of flowering time; flowering activator46
|
floral meristem
|
SmiMADS30-32
|
Paralogs of SVP
|
AP1/AGL7
|
A-class homeotic gene; regulates sepal and petal formation
|
sepals and petals47
|
SmiMADS65#
|
Snapdragon: SQUA and DEFH2848; Rice: OsMADS14, OsMADS15 and OsMADS1849
|
AP3
|
B-class homeotic gene; regulates petal and stamen formation
|
petals and stamens50
|
SmiMADS28#, SmiMADS29#
|
Snapdragon: DEF51; rice: OsMADS1652; maize: silky153
|
PI
|
B-class homeotic gene; regulates petal and stamen formation
|
petals and stamens
|
SmiMADS26#
|
Petunia: pMADS1 and GP54; rice: OsMADS2 and OsMADS49
|
AG
|
C-class homeotic gene; regulates stamen and carpel formation
|
stamens and carpels
|
SmiMADS45#, SmiMADS46#
|
Petunia: pMADS354; rice: OsMADS3 and OsMADS5855; maize: ZAG156
|
SEP1/AGL2, SEP2/AGL4, SEP3/AGL9, SEP4/AGL3
|
E-class homeotic gene; co-regulates floral development and activates B- and C-class genes
|
carpels, petals, stamens and sepals
|
SmiMADS70#, SmiMADS71#, SmiMADS72#
|
Petunia: FBP2 and FBP557; tomato: TM558; rice: OsMADS159
|
Note: Only those MADS-box genes related to floral development are listed. Unless otherwise stated, the genes mentioned were from A. thaliana. Tissue-specific expression patterns of the genes marked # in the upper right corner were confirmed by transcriptome data in this study. |
Flowering time is a key agronomic trait that directly influences crop yield and quality32. Many flowering time genes have been identified and characterised in the model plant A. thaliana, such as FLC, SVP, SOC1, FUL and AGL24. Herein, we predicted the flowering time genes in S. miltiorrhiza according to the phylogenetic relationships with corresponding Arabidopsis MADS-box members (Table 1). SmiMADS30-32 in the STMADS11 subfamily (homologous genes of AGL24) and SmiMADS57 in TM3 subfamily (homologous genes of SOC1) were supposed to be S. miltiorrhiza floral promotion factors, whereas SmiMADS33-34 in STMADS11 subfamily (homologous genes of SVP) and SmiMADS60-63 in FLC subfamily (homologous genes of FLC) may be inhibitors of flowering. Unfortunately, since we did not collect pre-flowering samples, these speculations could not be strongly supported by expression data. However, their expression levels were extremely low, or they were not at all expressed during the three periods after flowering (Fig. 7B).
Overall, MADS-box genes are actually involved in the growth and development of both vegetative and reproductive organs of S. miltiorrhiza. They are vital in all stages of S. miltiorrhiza flowering and precisely regulate the growth and development of various parts of flowers.