The constitutive expression of the chrysanthemum gene CmERF110 in Arabidopsis thaliana affects lateral branching


 Background: Producers of cut flower chrysanthemum are obliged to manually remove lateral buds, a procedure which consumes one third of the total production cost. The formation of lateral buds in ‘SEI NO ISSEI’ is suppressed when the plants are exposed to high temperatures, but the molecular basis of this phenomenon is not well understood. Here, the transcriptome of buds formed by decapitated chrysanthemum plants grown under a high temperature regime was characterized with a view to revealing which genes known to be involved in a pathway determining shoot branching were induced/repressed by the treatment. Results: The transcriptomic data was acquired using RNA-Seq technology, based on the Illumina HiSeq™ 2000 platform. Four libraries were generated from pooled lateral buds of decapitated ‘SEI NO ISSEI’. To predict the potential functions of unigenes in the ‘SEI NO ISSEI’ buds, after assembly, we performed seven functional database annotations. 132,396 unigenes were assembled, of which 79,116 unigenes were annotated in the seven functional databases. The percentage of unigenes annotated in the NR, NT, Swiss-Prot, KEGG, COG, Interpro, and GO databases were 54.59%, 42.96%, 39.05%, 41.84%, 20.72%, 37.25%, and 13.01%, respectively. Multiple differentially expressed transcription factors and auxin-related genes were identified in buds of decapitated ‘SEI NO ISSEI’ under 28°C/23°C or 38°C/33°C. Constitutively expressing CmERF110 in wild type Arabidopsis thaliana produced a bushy plant. A quantitative analysis of gene expression changes in the CmERF110 transgenic A. thaliana plants showed that the presence of the transgene altered the abundance of transcript produced by TIR1, ARF2, ARF16, IAA3 and IAA9 (encoding auxin signaling proteins) and PIN1, AUX1, LAX1, LAX2 and ABCB1 (auxin transport). Conclusions: This study reported a highly complete transcriptome from the buds of decapitated ‘SEI NO ISSEI’. Differential abundant transcripts during high temperature treatments were identified and validated by qPCR, many of these differentially abundant transcripts as key players in bud outgrowth. These include known members of the AP2/ERF, MYB, WRKY, bHLH families and auxin-related genes. CmERF110 regulated shoot branching when its encoding gene was heterologously expressed in A. Thaliana. The regulation acted through the auxin-related genes.


Background
The ability of the plant shoot to form branches has a major effect on the plant's architecture, a trait of major importance in the context of crop domestication and improvement [1,2]. Axillary meristems are composed of a group of cells which have retained their meristematic potential [3]. Following their initiation, these structures develop into axillary buds, which either remain dormant or develop into a new branches, depending a number of both internal and/or external cues [4]. Three forms of bud dormancy have been recognized: paradormancy occurs when growth ceases because of physiological factors external to the bud; endodormacy when growth is regulated by internal physiological factors; and ecodormancy when external environmental factors are most important [5]. The major environmental cues regulating the induction and release of bud dormancy are temperature and light, although the extent of the influence and the importance of crosstalk between temperature-and light-regulated signaling pathways appear to be somewhat species-dependent [6]. The relevant physiological factors include phytochrome, sugar and phytohormones, which are basically associated with direct phenotypic changes when plants perceive environmental signals [7]. The phytohormones 5 associated with bud growth and development include abscisic acid, ethylene, gibberellic acid, cytokinin, strigolactone and auxin.
Chrysanthemum is a valuable ornamental species [22]. Producers of cut flowers are required to manually remove lateral buds, a costly and energy-consuming procedure. In the cultivar 'SEI NO ISSEI', the formation of lateral buds has been shown here to be suppressed when the plants are exposed to a high temperature regime. In order to reveal the molecular basis of this suppression, the transcriptome of this cultivar was characterized, which resulted in the recognition that the gene encoding a specific ethylene response factor was up-regulated in response to a high temperature exposure. By expressing this gene heterologously in Arabidopsis thaliana, it was possible to determine that its product regulates shoot branching, 6 acting through the auxin-related genes. The conclusion was that it may be possible, through the manipulation of this gene within chrysanthemum, to achieve a branchless chrysanthemum plant, which would be of substantial value to breeders of chrysanthemums directed at the market for cut flowers.

Results
The effect of high temperature on bud outgrowth in 'SEI NO ISSEI' Decapitated 'SEI NO ISSEI' plants exposed to the 28°C/23°C treatment presented an acrotonic pattern of bud out-growth along their stems, whereas those exposed to 38°C/33°C produced very few buds (Fig. 1a). After a four day exposure to two temperature regimes (28°C/23°C or 38°C/33°C), there was no difference in the bud burst rate. However, when decapitated chrysanthemum was exposured to 28°C/23°C or 38°C/33°C on the seventh day, the bud burst rate was significantly different in the fifth and sixth nodes (the first node n-1 refers to the first bud of decapitated chrysanthemum from top to bottom). Buds formed by plants exposed for ten days to 28°C/23°C exhibited a significantly higher bud burst rate than those of plants exposed to 38°C/33°C (Fig. 1b). In the former plants, bud growth was acrotonic, as also shown by the pattern of leaf development, whereas bud growth under the higher temperature regime was inhibited and no acrotonic gradient was observed.
The length of a bud of different position was two to twelve fold greater in plants exposed for ten days to the lower temperature regime than in those exposed to the higher temperature regime (Fig. 1c).

Transcriptome sequencing and read assembly
In order to study the differentially expressed genes of buds of decapitated 'SEI NO

Unigene functional annotation
To predict the potential functions of unigenes in the 'SEI NO ISSEI' buds, after  (Table 3). Among them, the NR database had the largest number of unigenes, so we counted the species distribution according to the NR annotation results. The most common species was Vitis vinifera, which accounted for 16.71% of species distribution, followed by Solanum tuberosum (8.54%), Theobroma cacao (6.95%) and Erythranthe guttata (6.00%). The proportion of other species was 61.81% ( Figure S1).

8
To clarify the potential functions of unigenes in the buds of 'SEI NO ISSEI', we calculated its functional classification based on the COG annotation results. The COG annotation divided unigenes into 25 categories, the most of which was "general functional prediction only" (7601, 16.89%), followed by "replication, recombination and repair" (4309, 9.57%) and "transcription" (4187, 9.30%).
To further understand the potential functions, GO term enrichment analysis was performed. A total of 85,084 unigenes were divided into 53 functional groups, which mainly included biological process (23), cellular component (16) and molecular function (14). In biological process, the main functional groups were "metabolic process" (8904) and "cellular process" (8337). In cellular component, the main functional groups were "cell" (6393) and "cell part" (6393). Among the molecular function, "catalytic activity" (9984) and "binding" (8266) were most ( Figure S3). The results suggested that these main categories might play an important role in buds.
In addition, we calculated its functional classification based on the results of the KEGG database. A total of 58,140 unigenes were clustered, including cellular processes, environmental information processing, genetic information processing, human diseases, metabolism and organismal systems. The three main pathways were the "global and overview maps" in metabolism (12,893,22.18%), the "translation" in genetic information processing (5,562, 9.57%), and the "carbohydrate metabolism" (5,163, 8.88%) ( Figure S4).

Differentially transcribed of auxin-related genes in buds
Auxin plays a key role in bud growth and development, we analyzed auxin-related differential genes from RNA-seq data. There were 32 differential transcriptions associated with auxin, 21 genes in the signalling pathway and 11 genes in the transport pathway ( Fig. 3 and Table S3). Among 21 transcripts associated with auxin signalling process, the abundance of TIR1, ARF2, ARF4, ARF5, ARF8, ARF16, ARF18, ARF19, IAA3 and IAA9 transcript were low in buds outgrowth (n-24h and n-96h) compared to other time points (h-24h and h-96h), while PIN1, PIN2, AUX1, LAX1, LAX2 and ABCB1 (associated with auxin transport process) were significantly upregulated in buds outgrowth compared to inhibited buds.

Verification of RNA-seq data by qRT-PCR
In the library of bud outgrowth, fourteen differentially expressed genes were selected for qRT-PCR to test the reliability of RNA-seq data. The qRT-PCR assays largely validated the RNA-Seq based identification of differential transcription ( Fig.   4 and 5). We selected four transcription factors, including Unigene28455_All (ERF110), Unigene9775_All (MYB4), CL2694.Contig5_All (WRYY40) and CL3647.Contig6_All (bHLH36) and ten auxin-related genes (TIR1, ARF2, ARF16, IAA3 and IAA9), which were all related to shoot branching and potential candidate genes for regulating chrysanthemum branching.

The phenotype of A. thaliana plants heterologously expressing CmERF110
When  Fig. 6a and 6b). Axillary buds were rarely formed in the first three rosette leaf axils of WT plants, whereas nearly all rosette leaves carried buds or branches in the axils of WT/ERF110 (active buds), brc1 (inactive buds) and brc1/ERF110 (active buds) plants. A greater number of high order branches was produced by WT/ERF110, brc1 and brc1/ERF110 plants than by WT plants (Fig. 6g-j). WT/ERF110, brc1 and brc1/ERF110 plants formed a similar number of primary cauline leaf branches (CI) as did WT ones (Fig. 6k). WT/ERF110 and brc1/ERF110 plants produced significantly more secondary cauline leaf branches (CII) than did, respectively, WT and brc1 plants (Fig. 6l). The brc1/ERF110 plants formed more rosette branches (RI and RII) than did WT plants. The RI and RII phenotype of brc1 plants was weaker than that of WT/ERF110 ( Fig. 6m and 6n). The conclusion was that the product of BRC1 influenced the plants' RI and RII performance, while the constitutive expression of CmERF110 mainly affected CII as well as RI and RII.

The constitutive expression of CmERF110 in A. thaliana reprogrammed the transcription of auxin-related genes
The effect of constitutively expressing CmERF110 on the transcription of the auxinrelated genes was tested in 21 day old A. thaliana seedlings. The abundance of PIN1, AUX1, LAX1, LAX2 and ABCB1 transcript was from 1.30 to 3.14 fold higher in the transgenic plants than in WT plants, while TIR1, ARF2 , ARF16 , IAA3 and IAA9 were down-regulated by between 1.23 and 1.33 fold (Fig. 7). The suggestions that in the transgenic plants, the CmERF110 product participated in the shoot branching, acting through the auxin-related genes.

12
The capacity to form branches is an important determinant of crop productivity.
Shoot branching includes the formation of axillary buds in the axil of leaves and subsequent outgrowth of the buds [29]. The presence of an apical shoot meristem inhibits lateral bud outgrowth, and removal of this meristem initiates lateral bud outgrowth [30]. The outgrowth of lateral buds in the chrysanthemum cultivar 'SEI NO ISSEI' was found to be significantly compromised when the plants were exposed to a high temperature regime. In Rosa, temperature has also been shown to influence the budburst gradient along the stem [31] [43], which laid the foundation for us to study the burst of chrysanthemum buds under the contrast temperature (CL142.Contig2_All) and ABCB1 ( CL2492.Contig2_All) were all down-regulated ( Fig. 3 and Table S3). Physiological experiments conducted many years ago established auxin as a key regulator of axillary shoot branching [8]. TIR1 is an F-box protein, which acts as an auxin receptor [17]. It has been shown in tomato that a deficiency of ARF2a transcript results in the promotion of axillary shoot formation [32] and 13 ARF16 has been established as auxin-inducible [33] [45]. Several proteins are known to regulate auxin transport: these include the auxin efflux carrier PIN and the auxin influx facilitators AUX1 and LAX [34][35][36] [49][50][51]. ABCB1 is expressed in shoot and root apices [37], where it functions primarily to introduce auxin synthesized in the meristem into long-distance polar auxin transport streams [38][39][40] [53][54][55]. Our experimental results were consistent with the previous studies, indicating that under high temperature treatments the inhibition of bud outgrowth in chrysanthemum may be regulated by auxin-related genes.
A large number of transcription factors act in plants to regulate their growth and development [41]. Here, the ERF family became the particular focus of the transcriptomic analysis, because these factors are prominent in the response to abiotic stress [42,43], as well as being involved in numerous signal transduction pathways [44]. The sequence of five of the unigenes classed as being differentially transcribed as a result of the high temperature treatment identified them as likely members of the ERF family: both were more abundantly transcribed under h-96h than under n-96h ( Fig. 2 and Table S2), consistent with their products exerting control over bud outgrowth under high temperature conditions. The poplar gene EBB1 is known to encode an AP2/ERF transcription factor; plants over-expressing it exhibit early budbreak and those under-expressing it suffer from delayed budbreak [45]. The chrysanthemum ERF transcription factor CmERF053 has been shown to regulate shoot branching [46], while the product of the Larix kaempferi AP2/ERF gene LkAP2L2 exerts a pleiotropic effect on both branching and seed development [47]. The A. thaliana gene EBE encodes an AP2/ERF transcription factor, which has a notable effect on shoot architecture [48]. Finally, the peach transcription factor PpERF3b promotes precocious side-branching [49]. These previous studies suggested that ERFs played essential roles in the control of plant bud development, so a more in-depth investigation regarding this important group of temperaturerelated genes is required. CmERF110 was up-regulated by the high temperature regime inhibiting bud outgrowth in 'SEI NO ISSEI', however, its constitutive expression in A. thaliana induced a bush-like phenotype. Based on the above results, we supposed that the function of CmERF110 in both A. thaliana and chrysanthemum, may regulate different downstream genes or form protein complex with different protein interactions to regulate the shoot branching through affecting the auxin-related genes, but the mechanism needs further experimental verification.
Such as, auxin has been shown to enhance root-growth inhibition under aluminum (Al) stress in Arabidopsis . However, in maize (Zea mays), auxin may play a negative role in the Al-induced inhibition of root growth [50]. Previously, TSRF1, an ethylene response factor (ERF) protein from tomato, binds to GCC box in the promoters of pathogenesis-related genes positively regulates pathogen resistance in tomato and tobacco [51,52], but negatively regulates osmotic response in tobacco [52].
However, overexpression of TSRF1 enhances the osmotic and drought tolerance of rice by modulating the increase in stress responsive gene expression [53]. Our study also showed that the function of CmERF110 in chrysanthemum and Arabidopsis was opposite, and the previous studies further demonstrated that our findings were reliable.
In A. thaliana, the BRC1 product suppresses bud formation, since in the loss-offunction brc1 mutant, axillary meristem formation is promoted, the buds develop more rapidly and the plant branches more profusely [54,55]. When a transgene comprising CmERF110 driven by the CaMV 35S promoter was introduced into the brc1 mutant, the resulting plants produced more branches than when the same transgene was introduced into a WT background, although the latter transgenic formed more branches than did either brc1 or WT plants (Fig. 6). brc1 mutants has a significantly higher number of rosette branches (RI and RII) than WT plants, and brc1 plants has a similar number of primary and secondary cauline leaf branches (CI and CII) as the wild type [54]. Consistent with previous studies, the loss of BRC1 influenced the number of rosette branches formed, while the CmERF110 transgene affected both the number of secondary cauline leaf branches and rosette branches.
A qRT-PCR analysis identified a number of genes which were transcriptionally reprogrammed as a result of constitutively expressing CmERF110 in A. thaliana. The abundance of PIN1, AUX1, LAX1, LAX2 and ABCB1 transcript was higher in the transgenic plants than in the WT ones, so the transgenic lines exhibited a multibranched phenotype (Fig. 6). The high expression level of auxin transport protein in lateral buds is positively correlated with shoot branching. For example, the expression of RhPIN1 is upregulated during the process that sucrose promotes lateral buds [56]. In Arabidopsis, the expression level of PIN1 is significantly upregulated [57]. In decapitated peas, the growth of lateral buds are promoted and rapidly upregulate the expression of PsAUX1 and PsPIN1 [58]. In addition, the auxin signalling genes are also involved in the development of shoot branching.
Increasing the R/FR in Arabidopsis can promote branching and inhibit the expression of IAA3 [59]. The lateral bud outgrowth of leafy spurge is accompanied by downregulation of ARF2 expression [60]. In tomato, ARF2a acts downstream of IAA3 and IAA9 during the development of axillary shoots [32], SlARF2a inhibits lateral bud growth, and the down-regulated strains of IAA3 and IAA9 show an increased branching phenotype [61,62]. Similar expression patterns of IAA3, IAA9 and ARF2a suggest that they may function closely in axillary buds [32]. In our study, a further 16 effect of constitutively expressing CmERF110 in A. thaliana was to lower the abundance of TIR1 transcript, and reduced that of IAA3 and IAA9, with a knock-on effect on the transcription of ARF2 and ARF16 (Fig. 7). These transcriptional changes were consistent with the previous studies, thereby driving branch induction to completion.

Conclusions
In summary, this study reported a highly complete transcriptome from the buds of decapitated 'SEI NO ISSEI'. Differentially expressed genes during high temperature treatments were identified and validated by qPCR, many of these differentially abundant transcripts as key players in bud outgrowth, including the AP2/ERF, MYB, WRKY, bHLH families and auxin-related genes. To our knowledge, the present study has identified that CmERF110 regulated shoot branching when its encoding gene was heterologously expressed in A. Thaliana. The regulation acted through the auxin-related genens. The gene therefore represents a potential candidate for manipulating the shoot branching behavior of chrysanthemum. What remains to be established is how CmERF110 regulates bud outgrowth, how it interacts with other gene products influencing bud growth, and whether shoot branching in chrysanthemum can be manipulated by over-expressing CmERF110.

cDNA library construction and RNA-Seq analysis
RNA was extracted using a Total RNA Isolation System (Takara Bio Inc., Otsu, Japan) following the manufacturer's protocol [24]. After extract total RNA and treated with DNase I, mRNA was isolated using oligo (dT) and fragmented in mixed fragmentation buffer. Using the mRNA fragments as templates, the first strand of cDNA was synthesized, then the second strand was synthesized by using DNA polymerase I, RNase H, dNTPs and buffer. Purified double strands cDNA was resolved with EB buffer for end reparation and single nucleotide A (adenine) was added to each 3' end. After that, the A-tailed fragments were connected with adapters. The suitable fragments were selected for the PCR amplification. The constructed library was tested with the Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR System, then the quality-tested samples were sequenced using the Illumina HiSeq 2000 platform (Illumina, San Diego, CA, USA) [22]. Raw reads were edited by removing low-quality, adaptor-polluted and high content of unknown base(N) reads [25], after which we got clean reads. Trinity software [26] was then used to assemble the transcriptome and the resulting unigenes were assigned a putative function based on homologs present in the NR (NCBI nonredundant protein), NT (NCBI non-redundant nucleotide), Swiss-Prot, KEGG (Kyoto Encyclopedia of Genes and Genomes) and COG (Cluster of Orthologous Groups) databases. Conflicting assignments were resolved by applying priority in the order NR, Swiss-Prot, KEGG and COG. Differentially expression genes (DEGs) were identified by imposing the criteria P value <0.05, false discovery rate (FDR) ≤0.0001 and |log 2 ratio| ≥1.0. All sequencing data have been deposited in the NCBI sequence read archive (www.ncbi.nlm.nih.gov/sra).

Quantitative real-time PCR (qRT-PCR) assay
Total RNA was extracted from the buds of chrysanthemum plants subjected to each of the various temperature treatments and from 21 day old A. thaliana plants carrying the transgene p35S::CmERF110. Each treatment was represented by three biological replicates. A 1 μg aliquot of each RNA was subjected to an RNase-free DNase I treatment, then converted to ss cDNA (after) using PrimeScript ® Reverse Transcriptase (Takara). The subsequent 20 μL qRT-PCRs (three technical replicates per biological replicate) each contained 10 μL SYBR Green PCR master mix (Takara), 10 ng cDNA and 0.2 μM of each primer; the reactions were given an initial denaturation (95°C/2 min) followed by 40      The transcriptional response after 24 h of high temperature treatment of a selection of four t 34 Figure 5 The transcriptional response after 96 h of high temperature treatment of a selection of four t 35 Figure 6 The phenotypic consequence of constitutively expressing CmERF110 in A. thaliana in the wild Figure 7 The abundance of transcript generated from auxin-related genes involved in the regulation o