Identification of MADS-box genes in foxtail millet.
Two BLAST methods were used to identify 89 MADS-box genes in foxtail millet. The genes were named SiMADS1–SiMADS89 according to their chromosomal positions. Their genetic characteristics, including coding sequence (CDS) length, protein molecular weight, pI and subcellular localization, were determined. Among the 89 SiMADS-box proteins, SiMADS7 and SiMADS74 were the smallest with only 60 amino acids, and the largest one (SiMADS21) had 483 amino acids. The molecular weight of the proteins ranged from 6.78 to 53.18 kDa, and pI ranged from 4.41 (SiMADS21) to 11.39 (SiMADS35). Predicted subcellular localization revealed 1 in the endoplasmic reticulum, 11 in the mitochondria, 19 in the chloroplast, 14 in the cytoplasm, and 44 in the nucleus (Table S1).
Phylogenetic analysis and classification of the SiMADS-box genes.
The phylogenetic relationships of the 89 SiMADS-box proteins was studied by multiple sequence alignment. MADS-box gene classification of A. thaliana and rice was referenced. The foxtail millet MADS-box phylogenetic tree classified the SiMADS-box genes into two subgroups: 37 M-type genes and 52 MIKC-type genes. The SiMADS-box genes were similar to those of the monocotyledon rice, and could be further divided into 17 groups: 14 in the MIKC-type gene subfamily, and 3 in the M-type gene subfamily (Fig. 1). Interestingly, the FLC-like subfamily did not contain SiMADS-box genes, in contrast to both Arabidopsis 25 and rice 1. We also investigated the number of MADS-box genes in maize 26, sorghum 26, rice 1, Arabidopsis 25 and Brassica rapa 27. The numbers and proportions of MADS-box genes in each subfamily are shown in Table 1. C4 plants (foxtail millet, maize, sorghum) had similar proportions of MIKC-type MADS-box genes as C3 plants. However, there were significant differences between the M-type genes in C4 vs. C3 plants. The proportion of Mα subfamily members in C4 plants was much higher than in C3 plants. On the other hand, the proportion of Mβ and Mγ subfamily members in C4 plants was much lower than in C3 plants. These results suggest that after the divergence of C3 and C4 plants, more members of the Mα subfamily were retained by the C4 plants. Therefore, members of the Mα subfamily may have played a major role in the evolution of C4 plants.
Gene structure, motif composition, and protein-interaction predictions for the SiMADS-box gene family.
Exon and intron structures of SiMADS-box genes were obtained by comparing their CDSs with the corresponding genomic DNA sequences. In general, the exon/intron structures of different members of the same subfamily should be similar. As can be seen from Fig. 2, all SiMADS-box genes contained one or more exons. The mean number of exons in the MIKC-type subfamily was about 5.3, and that in the M-type subfamily was about 1.6, suggesting that MIKC-type genes had stronger differentiation ability and more rapid expansion. In the M-type subfamily, the lowest average number of exons, 1.25, was found for the Mγ branch. Among the 37 M-type genes, 19 (51.35%) contained only one exon, i.e., these genes had no intron structure. This may be further evidence for significantly different rates of evolution for the Mγ and other M-type branches. However, the average number of exons for the MIKC* branch of the MIKC-type subfamily was largest, as high as 9. The foxtail millet MADS-box genes were found to contain mainly the MADS, MADS-SRF, MADS-MEF2 and K-box domains.
The online MEME program was used to analyze the motifs of the 89 SiMADS-box proteins, and a structural diagram of these proteins was constructed. We identified 10 conserved motifs (Table S2). The motif composition of SiMADS-box proteins in the same subfamily was similar, and differed in different subfamilies. As can be seen from Fig. 2, the motif order for the MIKC-type subfamily was mainly 10-1-8-2-9, and these motifs could be conserved. The conserved motif order for the Mα subfamily was suggested to be 4-3-7-6-5, where the differences in the conserved motif patterns might be related to the proteins' specific functions, but this requires further elucidation.
These results showed that motif sequences patterns in a same subfamily are basically similar, which may also indicate that these proteins have similar functions or participate together in some pathways. We therefore turned to STRING 28 to predict interactions among the 89 SiMADS-box proteins. The results are shown in Fig. 3. Among the 89 SiMAD-box proteins, 32 were predicted to interact with each other. Among these putatively interacting proteins, 31 were MIKC-type, and only 1 (SiMADS60) was M-type. The higher number of interactions predicted for the MIKC-type MADS-box proteins might be related to their function.
Chromosomal distribution and synteny analysis of the SiMADS-box genes.
The chromosomal positions of the SiMADS-box genes are shown in Fig. 4A. Chromosome VIII had the least number of MADS-box genes (4), and chromosome V, the most (15). Referring to the closely related genes in A. thaliana falling into each other in 200 kB as tandem duplicates 29, only one pair of tandem duplicates (SiMADS69 and SiMADS70) were found among the foxtail millet MADS-box genes (Fig. 4A). The number of linked genes in linkage group (LG) III/V was higher than that in other linkage groups, whereas the distribution of SiMADS-box genes was largest in LG IX (5). There were 11 pairs of fragment duplicates among SiMADS-box genes (Fig. 4B, Table S3), many more than the number of tandem duplicates. In addition, the genes with fragment duplications belonged to the MIKCC branch. This may further explain why this is the largest branch of the foxtail millet MADS-box gene family, with the highest number of genes (49) and the highest proportion (55.06%) in each subfamily. Moreover, the proportion of this subfamily was higher than that in other species, even sorghum and corn, which are also C4 plants (Table 1). At the same time, the analysis of MADS-box gene structure also found that the MIKCC branch, which is a relatively young branch from an evolutionary point of view, has a large number of exons. These results suggest that some SiMADS-box genes may have been generated by gene-replication events, which may have been the main driving force for SiMADS-box gene evolution.
Collinearity analysis was conducted between M-type SiMADS-box genes and other plants (A. thaliana, Brassica rapa, Tartary buckwheat (Fagopyrum tataricum), Brachypodium distachyon, rice, and maize) (Fig. 5). The three dicotyledonous plants showed no collinearity with M-type SiMADS-box genes. There were only a few collinear genes between the M-type genes in foxtail millet, and those in Brachypodium distachyon, rice, and maize (Fig. 5, Table S4). Analysis of the collinearity map between MIKC-type SiMADS-box genes and those of other plants revealed highest collinearity with maize (71), followed by rice (62), Brachypodium distachyon (53), A. thaliana (7), Tartary buckwheat (6) and Brassica rapa (2). Further analysis of these collinear genes revealed that seven MIKC-type SiMADS-box genes (SiMADS20, SiMADS36, SiMADS52, SiMADS80, SiMADS81, SiMADS86, and SiMADS87) exist in both monocotyledons and dicotyledons. Among these seven genes, only two (SiMADS52 and SiMADS81) were included in the genes with fragment duplications. In addition, two MIKC-type genes (SiMADS35, SiMADS66) and two M-type genes (SiMADS15, SiMADS78) showed collinearity only with C4 plants (maize). In general, foxtail millet exhibited the highest collinearity with maize, suggesting that these C4 plants may have a close genetic relationship. In addition, for both M-type and MIKC-type SiMADS-box genes, the number of genes with collinearity to the monocots was much higher than that with collinearity to the dicots.
Evolutionary analysis of MADS-box proteins from foxtail millet and several other species.
We investigated the evolutionary relationship of M-type MADS-box proteins in the monocotyledons rice (31 genes), Brachypodium distachyon (17 genes) and maize (17 genes), and the dicotyledons A. thaliana (55 genes), T. buckwheat (26 genes) and Brassica napus (30 genes) (Fig. 6A, Table S5). According to the phylogenetic tree, the M-type MADS-box proteins could be divided into six subfamilies, labeled a–f. The motifs of the M-type MADS-box proteins were analyzed by online MEME analysis. Motifs 7, 1 and 2 were conserved and distributed almost alternately in the whole subfamily. However, there were large differences among subfamilies. Subfamily a had the largest distribution of SiMADS-box members (19); their conserved motif order was 8-2-4, but they lacked motif 7 − 1. The main order of the b and c subfamily motifs was 7-1-2-4. The motifs of subfamily d were mainly 7 − 1, and subfamily e did not contain any SiMADS-box members. The main order of the motif of subfamily f was 7-1-3-6-10.
To explore the evolutionary relationship between MIKC-type MADS-box proteins of foxtail millet and the monocotyledonous plants rice (38 genes), Brachypodium distachyon (29 genes) and maize (16 genes), and dicotyledonous plants A. thaliana (43 genes), T. buckwheat (38 genes), and Brassica napus (33 genes), we constructed a phylogenetic tree (Fig. 6B). There were six subfamilies (a–f) in the MIKC-type MADS-box family. Compared to the M-type MADS-box proteins, the motif distribution of the MIKC subfamily was more conservative. Its motif appeared most frequently in the order 5-1-4-7-6-2-3, and it was distributed in almost all subfamilies. However, there were also differences in some subfamilies, in particular subfamily f where motif 6 seemed to be replaced by 10. In addition, there was a new motif 9, which may confer unique physiological functions on this subfamily.
Expression patterns of the SiMADS-box genes in different foxtail millet tissues.
To study the physiological function of the SiMADS-box genes, the spatiotemporal expression of some members of the gene family was detected by qRT-PCR. Accumulation of eight MIKC-type and four M-type SiMADS-box gene transcripts in roots, stems, leaves (young leaves/mature leaves), peel and fruit in the middle stage of grain-filling was detected. Each of these genes belonged to a different subfamily. Since the function of MADS-box genes in floral organs is relatively clear, this study focused on the role of MADS-box genes in other tissues. These genes were expressed in all organs, but some were predominantly expressed in only a few tissues (Fig. 7A). Among them, SiMADS17 and SiMADS46 were highly expressed in roots, and SiMADS02, SiMADS52 and SiMADS60 were highly expressed in stems. In addition to SiMADS46 and SiMADS60, other genes were highly expressed in young leaves. Six genes (SiMADS12, SiMADS26, SiMADS28, SiMADS33, SiMADS37, SiMADS46) were highly expressed in mature leaves. Four genes (SiMADS02, SiMADS12, SiMADS33, SiMADS60) were highly expressed in the peel. All genes except SiMADS67 were highly expressed in the fruit, and in particular, the relative expression values of MIKC-type genes SiMADS33 and SiMADS37 were extremely high. These results, exhibiting differential expression patterns of SiMADS-box genes in different tissues of foxtail millet, indicated that the SiMADS-box genes have multiple functions in foxtail millet growth and development.
We also examined the correlation between SiMADS-box gene expression patterns in foxtail millet root, stem, leaf, peel and fruit; most of the genes were positively correlated (Fig. 7B). Expression of the M-type gene SiMADS28 and MIKC-type gene SiMADS67 was significantly positively correlated, and their expression was also significantly positively correlated with MIKC-type genes SiMADS2, SiMADS12 and SiMADS26. These results indicated similar functions for M-type and MIKC-type SiMADS-box genes during plant growth and development. For example, expression of the M-type genes SiMADS17 and SiMADS78 showed a significant positive correlation, and their expression was also significantly positively correlated with MIKC-type genes SiMADS33, SiMADS37, SiMADS46 and SiMADS52, which were strongly expressed in fruit.
Expression patterns of MADS-box genes in foxtail millet during fruit development.
Foxtail millet fruit are rich in calcium, dietary fiber, polyphenols, fats, proteins and other nutrients 17,18. Tissue-specific expression of SiMADS-box genes was also found to be high in fruit. Therefore, we determined the expression level of SiMADS-box genes in peel and fruit before, during and after grain-filling (Fig. 8A). Two genes (SiMADS02 and SiMADS17) showed very low expression in the peel. Five genes (SiMADS28, SiMADS33, SiMADS46, SiMADS52, SiMADS60) were highly expressed in the peel, mainly at the early filling stage. M-type SiMADS60 also exhibited high expression in the middle filling stage, and M-type SiMADS28 in the late filling stage.
Compared to the peel, all genes were highly expressed in the fruit, eight genes in particular (SiMADS12, SiMADS26, SiMADS28, SiMADS33, SiMADS37, SiMADS46, SiMADS60, SiMADS67). Among these genes, there were differences in expression for those belonging to the MIKC-type and M-type. MIKC-type genes were mainly highly expressed in the early filling stage (SiMADS12, SiMADS26, SiMADS46) and middle filling stage (SiMADS33, SiMADS37), whereas M-type genes SiMADS28 and SiMADS60 were mainly highly expressed in the late filling stage.
We also studied the correlation between SiMADS-box gene-expression patterns in peel and fruit at different filling stages and found both positive and negative correlations (Fig. 8B). For example, the MIKC-type SiMADS12 and SiMADS26, which were highly expressed in the early stage of fruit-filling, were significantly negatively correlated with the M-type SiMADS28 and SiMADS60, which were highly expressed in the late stage of grain-filling. This also revealed that there were differences in expression of SiMADS-box genes of the MIKC-type and M-type. There were also some similarities, for example, a significant positive correlation between M-type SiMADS17 and SiMADS78 expression, which were also significantly positively correlated with expression of the MIKC-type genes SiMADS33 and SiMADS37, which were highly expressed in fruit at the middle stage of filling.
Expression patterns of SiMADS-box genes in response to different abiotic stresses.
To determine whether the expression of SiMADS-box genes is affected by different abiotic stresses, we analyzed the expression of 12 SiMADS-box genes under eight kinds of abiotic stress: acid, alkali, NaCl, polyethylene glycol (PEG), flooding, dark, heat and cold. We used qRT-PCR to analyze the expression patterns of these 12 genes in leaves, stems and roots under the different treatments. Some SiMADS-box genes were significantly induced, whereas others were suppressed (Fig. 9A). For example, SiMADS33 and SiMADS78 were significantly induced under several abiotic stresses (acid, alkali, dark, heat and cold treatments). In addition, some genes showed different patterns under different treatments, such as SiMADS28, the expression of which was significantly upregulated under alkali, salt and PEG treatments, but changed little under flooding. SiMADS60 expression was significantly upregulated in both heat and cold treatments, but its expression also did not change significantly under flooding. SiMADS33 and SiMADS67 tended to be highly expressed in roots, SiMADS78 tended to be expressed in roots and leaves, and expression of SiMADS28, SiMADS37 and SiMADS60 was significantly upregulated in roots, stems and leaves. In general, the expression levels of genes that tended to be expressed in roots peaked after 24 h of stress, whereas the expression levels of those that tended to be expressed in stems and leaves could be detected after 2 h of stress. A correlation between SiMADS-box gene-expression patterns under stress was also observed (Fig. 9B). Most SiMADS-box genes were positively correlated. For example, there was a significant positive correlation between the expressions of SiMADS33 and SiMADS28, SiMADS67 and SiMADS78, and between SiMADS02 and SiMADS17 (P < 0.05).