ARF transcription factors in G. arboreum and G. hirsutum
The genome sequences of G. raimondii and G. arboreum provide us data resources to conduct a genome-wide screen of the ARF genes in the extent diploid progenitors of the allotetraploid G. hirsutum. In the previous studies, Sun et al., (2015) identified 35 ARF genes in G. raimondii [31]. To mine more ARF transcription factors in cottons the conserved domain (Pfam ID: PF06507) was used to hmmersearch against the G. arboreum and G. hirsutum genome databases, which resulted in 36 and 73 genes in G. arboreum and G. hirsutum genomes, respectively. The 36 G. arboreum ARF genes were designated GaARF1–GaARF20, and the 73 G. hirsutum ARF genes in A- and D-subgenomes were designated as GhARF1A/D–GhARF21A/D (Table 1). As those of Arabidopsis, cotton ARF proteins are composed of three domain regions, including DBD (DNA-binding Domain), MI (Middle Region) and CTD (C-terminal Domain) (Additional file 1: Figure S1).
Phylogenetic analysis of Gossypium ARF proteins
To illustrate the evolutionary relationships among the cotton ARFs, a phylogenetic tree was constructed using the protein sequences of 144 cotton ARFs, which were clustered into four clades (I–IV). The highest number of Gossypium ARFs are found in clade III and I, followed by clade IV and II (Fig. 1).
Overall, the expected diploid-polyploid topology is reflected in the tree for each set of orthologous/homoeologous genes, indicating general preservation during divergence of diploids and through the polyploid formation. We found that the number of ARF genes in G. hirsutum are approximately twice that in G. raimondii and G. arboreum, with one At or Dt homoeologous copy corresponding to one ortholog in each of the diploid cottons. Further, as shown in Fig. 1, the orthologous paired genes of the A genome (G. arboreum) and At sub-genome, or from the D genome (G. raimondii) and Dt sub-genome, tend to be clustered together and share a sister relationship.
Divergence of ARF genes in allotetraploid G. hirsutum and its diploid progenitors
The ARF genes in the two diploid species were then compared with G. hirsutum At- and Dt-subgenome homoeologs (Table 1). To explore the evolutionary relationship and possible functional divergence of ARF genes between the allotetraploid cotton and its extend diploid progenitors, the nonsynonymous substitution (Ka) and synonymous substitution values (Ks) and the Ka/Ks ratios for each pair of the genes were calculated (Table 1). By comparing the Ka and Ks values of 66 orthologous gene sets between the allotetraploid and its diploid progenitor genomes, we found that the Ka and Ks values are higher in the Dt subgenome than in the At subgenome (Fig. 2). These results indicate that GhARF genes in the Dt subgenome tend to have experienced faster sequence divergence than their At counterparts, suggesting an inconsistent evolution of ARF genes in the two subgenomes (Fig. 2).
In addition, the Ka/Ks ratios of one Dt-subgenome genes (GhARF3b_D) and five At-subgenome gene (GhARF2e_A, GhARF3c_A, GhARF4b_A, GhARF16b_A and GhARF17b_A) are greater than 1 (Table 1), suggesting that these genes have under positive selections after divergence of G. hirsutum from diploid ancestors, and may have gained new functions.
Expression analysis of GhARF genes in different cotton tissues
The expression profile of a gene family can provide valuable clues to possible functions of each genes. Analysis of 73 GhARF genes showed that most genes have different spatial expression patterns. For instance, GhARF1, GhARF2a, GhARF2b and GhARF2c were expressed in all the tissues of cotton examined (Additional file 2: Figure S2), whereas GhARF3a and GhARF3c were expressed preferentially in the pistils and ovules. Compared to GhARF5b, GhARF5a showed higher expressions in the root, pistil and ovule organs. Transcripts of GhARF3c and GhARF4a, GhARF9a and GhARF9b were most abundant in stem and root, respectively. Over half of GhARF genes showed a relatively high level of transcript accumulation in leaf. Notably, there are more than ten genes (including GhARF1, GhARF2a, GhARF2b, GhARF8a, GhARF9a, GhARF10b, GhARF11, GhARF16a, GhARF18 and GhARF19) that were highly expressed in cotton fiber cells at the fast elongation stage (5 dpa). In summary, most of the GhARF genes were up-regulated in ovule, fiber, vegetative and other tissues (Additional file 2: Figure S2).
GhARF2 and GhARF18 showed the highest expression in fiber (5 dpa) and both were located in the Clade I (Fig. 1), suggesting that they may function in cotton fiber development. Previous studies have demonstrated that ARF2 plays a role in transcriptional regulation in auxin-mediated cell division [29], leaf longevity [32], response to stress [33], regulation of fruit ripening [30] and so on. As GhARF2s shown pleiotropic effects on plant development [34], we decided to identify the major GhARF2s in regulation of cotton fiber elongation in subsequent experiments.
GhARF2 had a high expression pattern during fiber elongation process
There are nine ARF2 genes in G. hirsutum (GhARF2c_At not annotated), we first examined their expression profiles in different tissues in cotton (Fig. 3). Based on the published RNA-seq data (Zhang et al., 2015) GhARF2a, GhARF2b and GhARF2c genes had higher expression levels in various tissues than GhARF2d or GhARF2e (Fig. 3a). The transcripts of GhARF2b homoeologs (GhARF2b_At and GhARF2b_Dt) were enriched and abundant in cotton fiber cells (Fig. 3a), subsequent quantitative RT-PCR (qRT-PCR) confirmed the expression pattern (Fig. 3b). The highly up-regulated expression in fiber cell suggested that GhARF2b has been recruited to act primarily in cotton fiber.
GhARF2b overexpression represses cotton fiber elongation
To test the function of GhARF2b, we constructed the vectors to over-express and down-regulate GhARF2b_Dt in G. hirsutum by using the fiber-specific GhRDL1 promoter [7, 18, 35]. The expression levels of GhARF2b in transgenic cotton were clearly elevated in the overexpression lines according to qRT-PCR analysis; for example, the GhARF2b transcript abundance was about two-fold higher in the OE-3 than in the wild-type cotton fiber cells (Fig. 4a). However, GhARF2b did not stimulate fiber cell elongation, rather, it resulted in shorter fiber (Fig. 4b,c).
On the contrary, suppressing GhARF2b expression by RNAi resulted in longer fibers (Fig. 5a,b). The expression levels of GhARF2b in RNAi cottons in the RNAi lines were about 3 ~ 5-fold down-regulated in cotton fiber of 0DPA, 6DPA and 12DPA (Fig. 5c-e). Together, these data suggest that GhARF2b acted as a negative regulator of fiber cell elongation, at least when its expression exceeded the threshold. Alternatively, it may function in other aspects of cotton fiber development.
GhARF2b overexpression enhances cotton fiber initiation
Next, we examined the effects of GhARF2b up-regulation on cotton fiber initiation. The over-expression line OE-3 and RNAi line ds-2 were selected for analyses. The SEM with 60 × magnification of ovules of WT-R15, OE-3 and ds-2 collected at -1, 0, 1 DPA were observed (Fig. 6). The cotton fiber initiation of the − 1-DPA ovules did not present differences among the three types of cottons, however, the 0- and 1-DPA ovules of OE-3 and ds-2 lines showed higher and lower densities of fiber initials compared to the wild-type control (Fig. 6). Further, we magnified the SEM views of ovules to 500–700× (Fig. 7). Obviously, at the fiber initiation stage (0, 1 DPA), the fiber initial density of the OE-3 was increased by about 1.5-fold compared with that of the wild-type, in contrast, the fiber initial density of the ds-2 line was reduced (Fig. 7a-c). These results support a role of GhARF2b in promoting cotton fiber cell initiation.