Identification and analysis of the GAox family genes in rice
A previous study reported about 21 GAox genes in rice , but here we identified a total of 95 candidate OsGAox genes based on genome and transcriptome databases (Additional file 2: Table S1). We analyzed these genes and result showed that 95 candidate OsGAox genes contain 19 of the 21 reported genes, except GA2ox2 and GA2ox10 (Fig. 1a), because they lacked a common domain DIOX_N (PF14226) based on the results of alignment. Therefore, we further analyzed the 19 genes (Table 1).
To better understand the distribution of rice GAoxs on chromosomes, the map of genes distributed across chromosomes was created with Mapchart. Our study showed that the 19 genes were unevenly distributed on 7 chromosomes. There were four GAox genes both on chromosome 1 and chromosome 4. Five GAox genes were located on chromosome 5. Two GAox genes were mapped on chromosome 3 and chromosome 7. Only one GAox gene was found on chromosome 2 and chromosome 8, respectively (Fig. 1b). The information, including names, entry ID, number of deduced amino acid, molecular weights, predicted subcellular localization, group classification and theoretical pI were summarized in Table 1. Data showed that protein length of the identified OsGAox family genes ranged from 301 (GA20ox6) to 446 (GA20ox4) amino acids (aa) in length, with an average of 363 aa. The molecular mass ranged from 32.10 to 47.63 kDa, and the pI ranged from 5.25 (GA20ox8) to 7.44 (GA2ox6). Most of OsGAox family genes were predicted to be located in the nucleus and cytoplasm analyzed with WoLFPSORT  and TargetP , which was consistent with the previous studies . Results also suggested some of them can be transported into mitochondria or chloroplasts, implied that these organelles might be also involved in the GA metabolism in plants.
Phylogenetic analysis of the OsGAox gene family
In Arabidopsis, sixteen GAox genes (seven GA2oxes, four GA3oxes and five GA20oxes) have been identified previously . In order to determine evolutionary relationships of GAoxes in rice and Arabidopsis, the phylogenetic tree was constructed using the neighbor-joining (NJ) method from alignments of the GAox complete protein sequences of 16 AtGAoxes and 19 OsGAoxes (Fig. 2). The tree generated four distinct subgroups and also revealed that the phylogenetic representation of Arabidopsis and rice GAox proteins was quite different. Among the 35 proteins, 5 OsGAox and 5 AtGAox belonged to C19GA2ox subfamily, 8 OsGAox and 5 AtGAox to GA20ox subfamily, 2 OsGAox and 4 AtGAox to GA3ox subfamily and 4 OsGAox and 2 AtGAox to C20GA2ox subfamily. The presence of two subgroups of putative GA2ox (C19 or C20 GA classes) was also confirmed by C20GA2ox splitting from C19GA2ox in the phylogenetic tree. Four subfamilies (GA20ox, GA3ox, C19GA2ox and C20GA2ox) were shared in both two species, suggesting that these four subfamilies might be widespread in plant GA metabolism. Results showed that the diversity of GA20oxes is more abundant in rice, while the diversity of GA3oxes is more abundant in Arabidopsis, which might lead to various GA metabolism between monocots and dicots. Furthermore, the numbers of GA20ox and GA2ox genes were greater than GA3ox in both two species, indicating that GA20ox and GA2ox had undergone a more dynamic evolutionary route than GA3ox and thus resulted in more functional redundancy. Overall, the GAox genes shared some common characteristics in monocots and dicots evolutionary relationship, so the related studies on them could interact and put each other forward.
Gene structure and conserved motif analysis of GAox genes
To support the phylogenetic analysis, we performed gene structure analysis of GAox family members from Arabidopsis and rice. As shown in Figure 3, the number of exons was conserved, ranging from 1 to 3 exons in AtGAox and OsGAox genes. We also investigated intron phases with respect to codons. Most of the first intron was a phase 2 intron, suggesting that splicing events occurred after the second nucleotide. The second intron was generally a phase 0 intron, suggesting that splicing events occurred after the third nucleotide. This result revealed that the splicing phase was also highly conserved during the evolution of GAox genes both in Arabidopsis and rice. However, intron phases of rice were with a bias in favor of phase 0 compared to Arabidopsis, which indicated that the ancient introns were dominantly of phase 0 so as to favor intron average length and influence the evolution of GA oxidase genes in rice shuffling [25-27]. To characterize the structures of GAox proteins in rice, we further analyzed conserved motifs in detail by using the MEME motif search tool. The sequence logos and E values for ten motifs were presented, which named from 1 to 10 in turn by E values (Fig 4a). Result also suggested these conserved motifs were in different positions and had different width, implied the expansion of the functions during the evolutionary process in rice. Subsequently, we analyzed the distribution of these motifs in proteins. Results showed that motif 1, 2, 3, 4, 5, 6 and 8 were shared by most of the GAox proteins, while other motifs were absent in specific subfamilies (Fig 4b). In C19GA20ox subfamily, motif 7 was absent in all the members except for GA2ox8, while motif 9 was only shared in this subfamily, which implied that these two motifs, 7 and 9 may have special functions in C19GA2ox subfamily (Fig 4b). Moreover, motif 10 was shared by all C20GA2ox and GA3ox, implied the conserved functions of C20GA2ox and GA3ox (Fig 4b). Differences among motif distributions might explain the sources of functional divergence in GA oxidases in evolutionary history. Consequently, the detailed functions of these motifs needed to be further explored.
Expression patterns of OsGAoxes
To provide the clues for functional studies of OsGAoxes, we used FPKM values to represent their expression profiles in different tissues of rice and some of them were validated by qRT-PCR in this study . Because of the lack of the corresponding probe of GA3ox2, we analyzed expression patterns of the other 18 genes (Additional file 2: Table S2). Our results showed that all the GAox genes were almost expressed and displayed different expression levels in various tissues (Fig. 5). It was worthy to note that some GAox genes, especially GA2ox3, GA2ox7, GA2ox8 and GA20ox6, were highly expressed in panicle. This result indicated that these four genes may play key roles in panicle development. Furthermore, both GA2ox7 and GA20ox6 exhibited a high expression in all tissues, which revealed that they might play an important role during the whole developmental processes in rice. On the contrary, GA20ox8 expressed at a relatively lower expression in all tissues, implying that it might have a functional redundancy to others. Based on the above results, we next verified the expression pattern of these five genes by qRT-PCR. Results showed these investigated 5 genes exhibited same expression pattern compared to RNA-seq results (Additional file 1: Figure S1). Overall, these results represented that each GAox gene possessed special expression pattern in various tissues. Therefore, the studies on potential functions of these genes in different developmental stages in rice need to be done in future.
GAox gene expression alteration in response to exogenous GA
In this study, to explore the manner of these 19 GAoxes in response to exogenous GA, we treated 2-week seedlings of rice with GA3 and PAC (Paclobutrazol, biosynthetic inhibitor of the endogenous gibberellin) and analyzed the expression profiles of all 19 OsGAox genes by qRT-PCR. As expected, the seedlings treated with GA were higher than those non-treated seedlings, and the seedlings treated with PAC were lower than those non-treated seedlings (Additional file 1: Figure S2). This result showed that GAs indeed promoted the growth of plant. We further investigated the 19 OsGAox genes to get insight into the response event. Results showed only six of them exhibited obviously altered expression under the treatment of GA3 and PAC, GA2ox3, GA2ox5, GA2ox7, GA2ox9, GA20ox2 and GA20ox4 (Fig. 6a). Our expression profiling results showed that the expression of GA2ox3, GA2ox5, GA2ox7, GA2ox9 genes were dramatically increased under GA3 treatment and reduced under PAC treatment, indicated that most of genes of GA2 subfamily may be involved in the process of GA deactivation in response to exogenous GA. Interestingly, only two genes of GA20 and GA3 subfamilies had intense response to PAC treatment, GA20ox2 and GA20ox4, revealed that maybe only a few of the genes involved in bioactive GA synthesis played roles to regulate GA level under GA treatment. However, with the exception of these six genes, there was no obvious up-regulation or down-regulation expression of other genes (Fig. 6b). We speculated that this reaction may be probably limited to our experimental material. Overall, those results indicated that GA had a great effect on the height of rice, and not all OsGAox genes were involved in the gibberellin homeostasis. Some genes are the main regulatory genes, but some genes which were not involved in the gibberellin homeostasis may play a role in other aspects of rice developmental progress. After we figure out the function of one certain gene, our results can be utilized to up-regulate or down-regulate the gene by GA4 or PAC treatment to control the growth of rice.