Roles of DgBRC1 in Regulation of Lateral Branching in Chrysanthemum (Dendranthema ×grandiflora cv. Jinba)

The diverse plasticity of plant architecture is largely determined by shoot branching. Shoot branching is an event regulated by multiple environmental, developmental and hormonal stimuli through triggering lateral bud response. After perceiving these signals, the lateral buds will respond and make a decision on whether to grow out. TCP transcriptional factors, BRC1/TB1/FC1, were previously proven to be involved in local inhibition of shoot branching in Arabidopsis, pea, tomato, maize and rice. To investigate the function of BRC1, we isolated the BRC1 homolog from chrysanthemum. There were two transcripts of DgBRC1 coming from two alleles in one locus, both of which complemented the multiple branches phenotype of Arabidopsis brc1-1, indicating that both are functionally conserved. DgBRC1 was mainly expressed in dormant axillary buds, and down-regulated at the bud activation stage, and up-regulated by higher planting densities. DgBRC1 transcripts could respond to apical auxin supply and polar auxin transport. Moreover, we found that the acropetal cytokinin stream promoted branch outgrowth whether or not apical auxin was present. Basipetal cytokinin promoted outgrowth of branches in the absence of apical auxin, while strengthening the inhibitory effects on lower buds in the presence of apical auxin. The influence of auxin and strigolactons (SLs) on the production of cytokinin was investigated, we found that auxin locally down-regulated biosynthesis of cytokinin in nodes, SLs also down-regulated the biosynthesis of cytokinin, the interactions among these phytohormones need further investigation.


Introduction
Branching types of plants have evolved over the history of life to adapt to changing climates or environmental conditions. The two processes mainly involved in the formation of lateral branches are the initiation of axillary meristems (AM) and the outgrowth of branches. During postembryonic development, the shoot apical meristem (SAM) generates the entire aerial part of a plant body [1], which can be divided into several phytomers consisting of leaf, stem segment, and AM [2]. The AMs are initiated from the axils of leaves according to the detached [3] or de novo [4] meristem models; after initiation, the AMs develop into axillary buds. Whether the axillary buds remain dormant or grow out to yield branches is a key decision for branching types of plants which is determined by various environmental and developmental stimuli [5].
During the initiation of AMs, there are at least two pathways involved at different developmental phases. During the vegetative growth phase of Arabidopsis, LAS [6], RAX [7], and ROX [8] are required to initiate AMs. The arabidopsis mutant supershoot (sps) generates several meristems in each leaf axil, which correlates with increased levels of Z-type cytokinins [9].
Another important branching event is the outgrowth of lateral buds. The repression of lateral buds by auxin production in the main shoot is known as apical dominance [10]. Auxin transports basipetally after being synthesized in young leaves, eventually inhibiting branch outgrowth [11,12]. However, auxin has an indirect function in this process, as auxin never enters the axils [12,13]. There are two leading hypotheses explaining this. The second messenger hypothesis is that auxin has been shown to regulate the production of cytokinins or SLs, both of which regulate branches outgrowth in the nodes locally [14,15,16]. However, there are evidences that buds could be activated by the efflux auxin produced in the buds [17,18], moreover, auxin transport capacity in main stem is assumed limited [18], eventually competition of auxin transport between apices and buds leads to apcial dominance, which is named as the auxin transportation canalization hypothesis. Both hypotheses are supported by computational model [19,20,21] and experimental evidences [16,22,23,24].
Strigolactones (SLs) are newly defined hormones involved in the inhibition of lateral branching in several species [25,26]. Furthermore, mutants associated with greater branch production have been shown to exhibit deficiencies in SL synthesis or signaling, including the max mutant in Arabidopsis [27,28,29,30], dad mutant in petunia [31,32,33], rms mutant in pea [34,35,36,37], and dwarf mutant in rice [38,39]. The interactions between auxin and SLs in regulation of lateral branching are complicated, SLs may act by dampening auxin transport [18,28,40], or they may act downstream of auxin [16], or be independent from the status of stem auxin [41] to regulate lateral branching. SLs also interacted with auxin and cytokinin in other developmental events such as adventitious root formation [42], root-hair elongation [43] or stimulation of secondary growth [44].
In addition to auxin and SLs, cytokinin also plays a role in promoting the outgrowth of branches in nodes locally [45,46,47]. Since auxin and cytokinin could regulate the biosynthesis and signaling of each other, it was proposed that auxin and cytokinin maintain homeostasis during plant development [15,48,49]. The interaction and feedback loop within these phytohormones provide a robust balance for the whole system [41,50,51,52].
After perceiving the endogenous signals or environmental stimuli, the axillary buds respond and make a decision on whether to grow. A transcription factor known in various species as teosinte branched1 (TB1), branched1 (BRC1), or fine culm1 (FC1), contains a TCP domain and is regarded as a candidate which can act locally to prevent buds outgrowth in maize [53], rice [54], sorghum [55], tomato [56] and Arabidopsis [5,57]. TB1 is thought to contribute to the evolution of teosinte to maize, which resulted from a profound increase in apical dominance [53,58]. The functional role of TB1/BRC1/FC1 is conserved in preventing branch outgrowth in both monocots and dicots, while the TB1 in maize also plays a role in internode elongation and inflorescence development [5,53]. Moreover, BRC1 in pea and FC1 in rice act downstream of the SL pathway, and their functions are essential for SL mediated inhibition of bud outgrowth [59,60]. New evidence has proven that another class I HD-Zip transcriptional factor, GRASSY TILLERS 1(GT1), is a local regulator of tillering and consequently influences lateral branching in maize, moreover, the expression of GT1 was under the control of tb1 [61].
The branching pattern of ornamental plants determines their esthetic appeal, and hence, their commercial value. Chrysanthemum (Dendranthema grandiflorum) is one of the important standard cut flowers, and requires manual decapitation or removal of lateral branches to maintain its architecture, which comprises one-third of the production cost. We previously reported that strigolactones regulate lateral branching in the presence of auxin source in chrysanthemum [40], and that DgIPT3 isolated from chrysanthemum engages in cytokinin biosynthesis and lateral branching [62]. Other studies provided approaches to control the lateral branching of chrysanthemum, such as the transformation of antisense DgLsL [63,64]. The genetic network of branching in chrysanthemum needs further elucidation to provide breeders with new methods to cultivate novel cultivars with ideal traits. Here we build on this understanding by identifying the roles of DgBRC1 in regulating lateral branching under endogenous and exogenous stimuli. Additional investigations describe the interactions among auxin, cytokinin, and SLs in the regulation of shoot branching.

Apical dominance in chrysanthemum
While considering the multi-branching nature of chrysanthemum plants, the branching characteristics of chrysanthemum cultivar 'Jinba' grown in a greenhouse were recorded during different growth stages. We recorded the length of all buds along the whole plant when the plant heights were 45 cm, 65 cm and 85 cm, respectively. The positions of buds were numbered acropetally, where bud 1 was the lowest bud ( Figure S1). All the buds on the 45 cm high plants remained dormant, and started to activate at the lower positions (around buds 21 to 26) when the plants reached 65 cm. Gradually, several active buds elongated and produced branches (around buds 19 to 27 in 85 cm high plants). Interestingly, the lowest buds usually maintained dormancy or expanded by a few leaves without further elongation during the entire growth period. Thus, the activation of axillary buds in chrysanthemum was associated with the developmental status and the distance from the shoot apex, exhibiting the same pattern observed in other species [2]. Additionally, a basipetal elongation of lateral branches was observed during the reproductive phase ( Figure S2).
The classic experiment of apical dominance showed that removal of the SAM resulted in the activation of lower AMs and production of lateral branches [10]. For the assays below about decapitation or isolated stem segments, the buds/branches were numbered basipetally where bud 1 was the top bud whose attached leaf expanded to 10 mm below the shoop tip, and the bud 2 was the bud below bud 1. Following decapitation of chrysanthemum, the top three buds elongated. Fifteen days postdecapitation, buds 1 and 2 had the same growth rate and yielded 17 mm branches, whereas buds 4,6 were activated but did not show obvious elongation due to correlative inhibition ( Figure S2) [65].

Isolation of BRC1 homologue from chrysanthemum
To study the role of BRC1-like transcription factors during the development of lateral branches, two cDNA clones contained the TCP domain were isolated from the axils of chrysanthemum leaves. The1362 bp clone encodes a 335 amino acid protein, and the 1545 bp clone contains an extra 183 bp segment adjacent to the 39 UTR which introduces a stop codon, truncating the protein at 318 amino acids ( Figure 1, Figure 2D); these two segments were named as DgBRC1-1 and DgBRC1-2, respectively. In DgBRC1-2, only one amino acid is encoded in the un-spliced intron I and then a stop codon occurs ( Figure 1). Both copies contain the conserved TCP and R domains, and the ECE motif typical for the CYC/ TB1 clade of the TCP family ( Figure S3) [5,66]. Two segments were cloned from genomic DNA; one has the same sequence as DgBRC1-2, whereas the other has a 49 bp intron (intron II) ( Figure 1). Because that these clones were almost identical, even in the 39UTR region, so they were alleles from the same locus. As a hexaploid, chrysanthemum is thought to be a hybrid originated from several ancestries such as D. vestitum, D. indicum and D. nankingense [67], so there may be several alleles with same or different functions in the locus. We performed a phylogenetic analysis with a selected set of class II TCP factors from various plant species to explore the evolutionary relationship among DgBRC1 genes. The nucleotides sequences coding the conserved TCP domains were aligned and maximum likelihood analyses were performed. As expected, DgBRC1 belongs to the CYC1subclade of the CYC/TB1 clade of the class II TCP family along with other BRC1-like genes such as PsBRC1, SlBRC1a, SlBRC1b, and AtBRC1 (Figure 3), all of which were shown to retain the function of tb1 in regulating lateral branching [5,56,57,60]. The basic helix-loop-helix (bHLH) motif of the TCP domain was predicted to promote DNA binding and protein-protein interactions [66,68], and some TCP proteins have been shown to be targeted to the nucleus [5,69,70]. According to our analysis of subcellular localization, DgBRC1-1-GFP localized to nuclei, but DgBRC1-2-GFP was dispersed all over the cells (Figure 2). Compared with DgBRC1-2, DgBRC1-1 has a 17 amino acid tail; we mutated the nucleotides coding these 17 amino acids by inserting 2 nucleotides, resulting in a frameshift mutation ( Figure 2D). Interestingly, DgBRC1-1D17-GFP still accumulated in nuclei in 20 of 25 cells observed (data not shown), and were widely dispersed in another 5 cells ( Figure 2C). To conclude, the peptide in the C-terminal of DgBRC1-1 is partly necessary for nuclear localization, however, the requirements may not be sequence specific.

DgBRC1 is mainly expressed in dormant axillary buds
To investigate the expression pattern of DgBRC1 during the vegetative phase of chrysanthemum, DgBRC1 transcripts were  . Branches with support of 50 or more are indicated. AtTCP4 is in an outlying group. Clades are named according to [70] [75]. Accession no. is listed in Table S1. doi:10.1371/journal.pone.0061717.g003 quantified by Real Time PCR. DgBRC1 was mainly expressed in the nodes containing axillary buds, which supported their roles in shoot branching ( Figure 4). DgBRC1 was weakly expressed in the stem, leaf, and main shoot, while its expression in root was barely detectable. The highest level of expression was in the first node below the main shoot, and remained high in nodes 2 to 4 ( Figure 4). The DgBRC1-1 transcripts levels were also detected in different tissues, whose expression pattern was similar to total DgBRC1 (data not shown).

Functional conservation of the DgBRC1s
To determine whether DgBRC1 is functionally conserved, the DgBRC1 ORFs were overexpressed from the Cauliflower Mosaic Virus 35S promoter (35S::DgBRC1-1, 35S::BRC1-2) in the Arabidopsis WT and brc1-1 mutant. Multiple independent transgenic lines were generated with each construct, and those showing Medelian segregation patterns 15:1 in T2 lines were taken to homozygosity for detailed analysis. The numbers of rosette and cauline branches with a length of at least 3 mm were scored 10 days post-anthesis (DPA). Figure 5A represents typical lines for each construct with 35S::DgBRC1-1 (35S::DgBRC1-2 lines were similar and not shown). Overexpression of DgBRC1 reduced the number of rosette branches from 7.6 in brc1-1 to 4.5-5.7, which was indistinguishable from wild-type (WT) plants with an average of 4.6 branches ( Figure 5B, Table S2). Overexpression of DgBRC1 in WT inhibited the total growth of WT plants, and reduced the number of rosette branches from 4.6 to 1.8-2.1 (Table S2). These results indicated that both variants retain conserved functions of regulating lateral branching.

Accumulation of DgBRC1 transcript is regulated by apical dominance and planting density
To determine the effects of apical dominance on DgBRC1 transcript levels, the classical decapitation assay was conducted. As mentioned earlier, the top three buds were released predominantly after decapitation ( Figure S2). To determine whether the outgrowth of lateral branches correlates with down-regulation of DgBRC1, transcript levels were analyzed in node groups 1 and 2 (node 1+2, top two nodes), and node groups 3 and 4 (node 3+4) before any visible sign of bud outgrowth. After decapitation, the transcript levels of DgBRC1 in nodes decreased dramatically 1 h following decapitation, and then almost attained pre-decapitation levels by 48 h after decapitation ( Figure 6A). We conclude from these results that DgBRC1 transcription was down-regulated rapidly when the inhibitory effect of SAM on lateral buds was released.
The response of plants to high planting density is known as shade avoidance syndrome [71,72], which includes a decrease of lateral branching [73]. In Arabidopsis, high planting density was found to regulate the outgrowth of branches partly through BRC1 [5,74]. To test whether DgBRC1 transcript levels were sensitive to planting density, chrysanthemum seedlings were planted in 1 or 9 plants per pot (9 cm69 cm69 cm), respectively. No matter in nodes 1+2 or 3+4, the DgBRC1 transcript levels in high density conditions (9 plants per pot) increased significantly compared with

Auxin, bud outgrowth and DgBRC1 transcripts
To investigate transcript levels of DgBRC1 regulated by auxin, plant growth regulators (PGR) were applied on two-bud segments including bud 3 and bud 4 cultured in a split-plate system ( Figure  S4) [46]. DgBRC1 transcript analysis was performed on both nodes 3 and 4 following 4 hours of treatment with 5 mM NAA applied apically and 5 mM NPA applied basally; the elongation of branches was recorded 10 days after treatment. In the intact plants no activation of the buds at nodes 3 or 4 was observed over the course of the experiment, so the growth of branches in intact plants was not shown in Figure 7. Compared with the two-bud segments without any PGR applied (control), 5 mM NAA in the apical media block was found to slightly inhibit the outgrowth of bud 3, and did not alter the growth of bud 4 ( Figure 7A, D); the DgBRC1 transcript levels in node 3 and node 4 were restored to the levels in buds of intact plants, which did not correlated with the observed changes in outgrowth of bud 3 and buds 4 ( Figure 7G). 5 mM NPA in the basal media block had modest effects on outgrowth of branches ( Figure 7A, D) and DgBRC1 transcript levels when compared with the control ( Figure 7G). Interestingly, transcription of DgBRC1 in node 3 was higher than in node 4 after NPA application, which could be explained by the inhibition of the PATS by NPA from bud 3 to bud 4.

Cytokinin, bud outgrowth and DgBRC1 transcripts
To investigate the effect of cytokinin on transcripts of DgBRC1, 5 mM synthetic cytokinin Benzylaminopurine (BAP) was added to the apical or basal media blocks. Nodes 3 and 4 were sampled separately 4 hours after PGR treatment, and the elongation of branches were recorded 10 days thereafter. Both apically and basally applied BAP promoted elongation of both buds, especially buds near to the application position ( Figure 7B, E). Basally applied BAP on two-bud sections modestly reduce the DgBRC1 transcript levels, while apically applied BAP did not down-regulate transcription of DgBRC1 in both nodes ( Figure 7H). These results indicated that cytokinin promoted elongation of lateral branches, but it was not correlated with the transcripts of DgBRC1. DgBRC1 may be regulated post transcriptionally, or there may be other pathways independent of DgBRC1 which can respond to cytokinin.

Antagonism between auxin and cytokinin on lateral branching
To investigate the antagonistic effects of auxin and cytokinin on outgrowth of lateral branches and transcripts of DgBRC1 in chrysanthemum, two assays were designed. Firstly, both 5 mM NAA and BAP were supplied apically; secondly, apical 5 mM NAA and basal 5 mM BAP were supplied simultaneously. Comparing Figure 7F with 7D and 7E, basal cytokinin's promotion of growth in bud 4 was suppressed a bit by the apical auxin, but apical cytokinin's promotion of growth of bud 4 was countered by apical auxin. In another way, basal BAP weaken the inhibition by apical auxin, while apical BAP strengthened the inhibition of apical auxin on outgrowth of lateral branches. According to Figure 7I, apical applied auxin did not alter the transcripts of DgBRC1 no matter BAP was applied apically or basally.

Effect of Auxin and Strigolactone on DgIPT3 transcript in nodes
To test the effects of auxin and strigolactone on biosynthesis of cytokinin, transcript levels of DgIPT3 in nodes were investigated 6 hours after application of apical NAA, basal GR24 (a synthetic SL). According to Figure 8, the DgIPT3 transcript levels in two-bud segments (control, as decapitated plants) were enhanced significantly compared with those from intact plant, while DgIPT3 transcript levels were decreased to the level of intact plants when apical NAA was applied ( Figure 8). Basal GR24 reduced the DgIPT3 transcript levels compared with the control, especially for node 3, which had DgIPT3 transcript levels significantly different from those of node 4.
To conclude, apical auxin could reduce DgIPT3 transcript levels of in the nodes, implying that auxin reduces the biosynthesis of cytokinin in axillary buds locally. In addition, SLs could also reduce DgIPT3 transcript levels in the nodes.

Structure of DgBRC1 genes
The DgBRC1 genes described in this study possessed all the characteristics of TCP family members in other species, such as the TCP domain, R domain, and especially the ECE motif which is contained in the CYC/TB1 clade [75]. TCP genes containing a bHLH motif have been shown to be involved in regulation of plant growth and development [66,68,76]. Two subfamilies of TCP proteins are further subcategorized into class I [68], and class II [53,77,78] based on phylogenetic analyses. The class II TCP proteins could be further divided into two groups: the CINCIN-NATA (CIN) clade [79] and CYC/TB1clade [5,80]. In core eudicots, CYC/TB1 has been duplicated twice, generating three subclades: CYC1, CYC2, and CYC3 [75]. The CYC1 subclade mainly functions to regulate lateral branching [5,56,57]. The CYC2 subclade plays a role in determination of flower symmetry [80,81,82]. The function of CYC3 subclade is poorly understood, the BRC2 gene from Arabidopsis played a minor role in the outgrowth of lateral branches [5], and may be involved in coordination of growth among branches [74].
Interestingly, DgBRC1 also shares some rare characteristics with other members in this family, such as introns located within 39 UTRs, which has previously been reported for CYC in Antirrhinum [82], and three CYC2 genes in sunflower [83]. Introns in the 39UTR may engage in nonsense-mediated mRNA decay (NMD), which could identify and eliminate aberrant mRNAs or regulate gene transcription [84,85].
The TCP domain is considered necessary but not sufficient for DNA binding and protein-protein interactions [5,68,69,70,86]. In our assay, two alleles were found to differ in subcellular localization; the short one localized to the cytoplasm and nuclei, whereas the long one displayed nuclear localization. Most cells transformed by the mutated DgBRC1-1 in the extra 17 amino acids of the C-terminal still localized in the nuclei. These results indicated that the extra 17 amino acids were necessary for nuclear localization, but the sequence may be not specific.

Basipetal and acropetal cytokinin regulated lateral branching in different pathways
There is evidence that xylem cytokinins (mainly tZR type) are transported through the transpiration flow acropetally as longdistance signals, and phloem cytokinins (mainly iP type) are translocated either systemically or basipetally [87]. Xylem cytokinins were supposed to promote sustained outgrowth of lateral branches [21]. In our work, cytokinins supplied in apical or basal medium were supposed to transport in different directions. Both basipetal and acropetal cytokinin promoted the outgrowth of buds 4 when apical auxin was absent, seemed that the correlative inhibition was released. However, basipetal cytokinin strengthened the inhibition on lower buds manipulated by apical NAA, which was similar to results in other plants [46,88,89,90,91,92], while acropetal cytokinin weakened the repression of auxin on buds outgrowth. Basipetal and acropetal cytokinin also played different roles in regulation of DgBRC1 transcripts in short term. In consequence, cytokinins from different directions or sources may be involved in different pathways.

Roles of DgBRC1 in regulation of lateral branches
BRC1 loss-of-function mutants yielded multiple branches in various plant species [5,53,54,56,57,60], and transcripts of BRC1 were down-regulated upon release of apical dominance; furthermore, BRC1 has been shown to respond to other hormonal or environmental stimuli [5,57]. A strong correlation between bud repression and expression of BRC1 genes has been proposed [93]; therefore, BRC1 was thought to be an integrator of branching signals which could control the outgrowth of lateral branches [5].
In our work, DgBRC1 transcripts responded rapidly to the release of apical dominance by decapitation, in addition, 4 hours after auxin was applied to the stem segments apically, transcripts level of DgBRC1 returned to the level of intact plants. These results indicated that DgBRC1 transcripts were related with auxin regulation of buds outgrowth in short term. In another aspect, DgBRC1 transcripts were higher in high planting density, which indicated that DgBRC1 was also related with shade avoidance syndrome. However, DgBRC1 transcripts could not respond to cytokinin applied apically or basally at least in short term. Results from other studies suggest that, more branches were generated when pea brc1 mutants were decapitated or supplied with BAP, which indicated other BRC1-independent pathways could control the outgrowth of branches [60]. Furthermore, transcripts of FC1 remained high in active buds of a rice strigolactone mutant [38], and SbTB1 transcripts did not increase when strong buds arrest was caused by defoliation [94]. In conclusion, BRC1 may be not sufficient or necessary for outgrowth of lateral branches.

Auxin regulates cytokinin biosynthesis in nodes locally
Concentration of cytokinin in xylem exudates or axillary buds increased after decapitation [48,95,96,97], and auxin could regulate the biosynthesis of cytokinin, which makes cytokinin a good candidate as the second messenger of auxin [15,48]. Two hypotheses have been proposed regarding the sites at which auxin effects cytokinin biosynthesis. Since root is the main place of cytokinin synthesis, it was supposed that auxin effects biosynthesis of cytokinin in root, followed by cytokinin transport into axillary buds through xylem. To support this, transcripts of BrIPT3 and 7 in root were reduced by NAA incubation [98]; the biosynthesis of cytokinin was iPMP dependent in roots of Arabidopsis and tobacco, which could be down-regulated by auxin [15]. However, there is other evidence suggesting that the cytokinin promoting axillary bud outgrowth after decapitation was locally biosynthesized in the nodal stems rather than in the roots [48]. After decapitation, PsIPT1 and 2 transcripts of in nodal stems increased dramatically, but transcripts in axillary buds remained unaffected [48]. In rice, expression of OsIPT2, 4, 7 and 8 in the shoot apex (containing the SAM, axillary buds, young leaves and nodes) were clearly suppressed within 3 h following auxin treatment [59]. Here we showed that DgIPT3 transcripts in nodes were up-regulated after decapitation, and dropped back to the levels of intact plants when auxin was applied apically on decapitated plants, which indicates that auxin could regulate the biosynthesis of cytokinin in nodes locally.

The interactions among auxin, cytokinin, and SLs
Since the mutants of SLs production and signaling were investigated and then the characteristic of SLs was revealed [25,26], SLs appear to be involved in optimizing several plants growing and developmental events in lower and higher plants [42,43,44,99]. During these events, SLs were found to be interacted with other hormone to balance the homeostasis of plants [42,43,44,99]. In the events of lateral branching, the interactions between SLs and auxin were complicated, for example, SLs was supposed to regulate shoot branching by dampening auxin transport, which is supported by direct evidence that GR24 inhibited branching only in the presence of auxin in the main stem in Arabidopsis and chrysanthemum [23,40], and PIN1 accumulation in xylem parenchyma cells was reduced by GR24 [18,23]; other evidences indicated that, SLs acted as a second messenger to repress shoot branching [16], and SLs could repress lateral branching directly without requirement of apical auxin [16,99].
Auxin and cytokinin were shown to play antagonistic roles in regulation of shoot branching. Auxin could inhibit the biosynthesis of cytokinin in root [15,95] or stem [48], in our work auxin inhibited cytokinin biosynthesis in nodes, which was supposed to be more fast and effective to regulate lateral branching. Furthermore, cytokinin could regulate the biosynthesis of auxin [50], and phloem cytokinin could modulate the polar transport of auxin [100]. A homeostasis feedback loop via auxin and cytokinin biosynthesis, transports, and signaling is thought to regulate several important developmental events [46,90].
There has been no direct evidence about the interactions between SLs and cytokinin. In our study, when GR24 was applied basally in stem segments, the DgIPT3 transcripts levels were reduced in nodes compared with stem segments without any hormone supplied, which suggested that SLs may inhibit the biosynthesis of cytokinin. Interestingly, DgIPT3 transcripts level in node 3 was much lower than node 4, which may be explained as that SLs dampened the transport of auxin from node 3, and the accumulation of auxin further repressed the biosynthesis of cytokinin. In addition, auxin could inhibit the biosynthesis of SLs in stems [101], so it is equally possible that auxin induces SLs to repress cytokinins biosynthesis in our work. Therefore, auxin and SLs could be in a dynamic feedback loop to regulate cytokinin biosynthesis and to control lateral branching. From the above, auxin, cytokinin and SLs are in a dynamic feedback loop, and further studies are needed to reveal this network in regulation of branching.

Plant materials and growth condition
Chrysanthemum (Dendranthema grandiflorum cv. Jinba) plantlets were propagated under sterile conditions in jars containing MS agar medium [102], and then grown in tissue culture room at 24uC with a photoperiod of 16/8 h light/dark and a light intensity of 100-120 mmol m 22 s 22 . For the apical dominance characterization assay, plants were transferred into pots (9 cm69 cm69 cm) containing peat soil and vermiculite (1:1) in a green house at 24uC, with a photoperiod of 16/8 h light/ dark. For the planting density assay, seedlings were cultured with 1 plant per pot, or 9 plants per pot. Seeds of Arabidopsis thaliana Col-0 (WT) and brc1-1 were stratified for 3 d at 4uC, and then sown in cells containing peat soil and vermiculite (1:1); they were transferred to a chamber at 22uC with a photoperiod of 16/8 h light/dark.

RNA extraction and gene isolation
Total RNA was extracted from nodes with TRIzol Reagent (Invitrogen, 15596-026); cDNA synthesis was performed using SuperScript II reverse transcriptase (Invitrogen, 18064-022). Chrysanthemum TCP domain factors were amplified from cDNA prepared from nodes using PCR with two different pairs of degenerate primers (Table S3). After amplification of a fragment containing a partially conserved TCP and R domain, full-length cDNA of DgBRC1 was elongated by 59and 39 Rapid Amplification of cDNA Ends (RACE) PCR. Complete DgBRC1s from both cDNA and genomic DNA were cloned with primers specific to the 59UTR and 39UTR. Amplified fragments were cloned into the pMD18-T vector (Takara, D101A) and sequenced. Genomic DNA was extracted from young leaves using the CTAB method.

Sequence Alignment and Phylogenetic analyses
DNA Sequences from the start of the TCP domain to the end of the R domain were aligned with ClustalW2 (http://www.ebi.ac. uk/Tools/msa/clustalw2/) using default parameters [103], and visualized with Genedoc [104]. During phylogenetic analyses, TCP domain-coding DNA sequences were aligned with MUSCLE (http://www.ebi.ac.uk/Tools/msa/muscle/) using default parameters [103]. The test for best nucleotide substitution evolutionary model was done with jMODELTEST [105,106]. The best fit model (Akaike Information Criteria selection) was GTR+I+G (parameter for gamma distribution = 1.0913). Maximum likelihood (ML) tree reconstruction with the best model and 100 bootstrap pseudoreplicates was run in MEGA 5 [106,107]. Plants species and accession numbers are listed in Table S1.

Mutagenesis of DgBRC1-1
Mutagenesis through PCR-driven overlap extension of DgBRC1-1 was performed as previously described [108]. The nucleotides encoding the extra 17 amino acids in C-terminal of DgBRC1-1 were frameshifted, and then cloned into pEZS-NL for further use in subcellular localization observations.

Subcellular localization
For construction of the 35S::DgBRC1-GFP reporter plasmids, the ORFs of DgBRC1-1 and DgBRC1-2 were each cloned into binary vector pEZS-NL. Transformation into Onion (Allium cepa) was performed as described previously [109]. Onion peels were unfolded in water, and then viewed with an Eclipse C1si confocal microscope (Nikon); images were aquired using the EZ-C1 FreeViewer software (Nikon). GFP was visualized by using an excitation wavelength of 488 nm and a band pass 510 to 525 nm emission filter.

Generation of transgenic plants
For complementation experiments, the ORFs of DgBRC1-1 and DgBRC1-2 were each cloned into binary vector pBI121, fusing them with the 35S promoter. The resulting constructs were transformed into Arabidopsis thaliana mutant brc1 plants via Agrobacteriaum tumefaciens strain LBA4404 using the floral dip method [110,111]. Independent transformants were screened on MS medium containing 50 mg l 21 kanamycin. Independent homozygous T3 lines with single insertion sites were used for the branching phenotype analysis.

Split-plate and two-bud section system
The split-plate system was modified according to [46]. We used 30 ml MS medium for each 9 cm petri dish; after solidification, a 10 mm wide strip of the medium was removed from the centre of the plate. The volume of the media block remaining on each side was about 12.5 ml; 25 ml of 2.5 mM stock solutions containing various compounds were injected into one or both sides of the media blocks with a micro-pipette, and the final PGR concentration was 5 mM. BAP (Sigma B3408), NAA (Sigma N0640) and NPA (Dikma 46154) were dissolved in 70% ethanol, and GR24 (LeadGen Labs, Orangen, CT, USA) was dissolved in acetone. Stocks were stored at 220uC, and fresh stocks were made every month. Chrysanthemum seedlings were grown to10 cm high in sterile conditions, and then they were used for effects of PGRs on outgrowth of lateral branches and determination of DgBRC1 transcripts in nodes. After PGRs had diffused evenly throughout the media, two-bud sections containing buds 3 and 4 were cut from the chrysanthemum seedlings, and then inserted into media. The total length of the two-bud sections was about 1.7 cm, and the distance between node 3 and node 4 was about 1.0 cm. It was ensured that at least 4 mm of the cut tips on each side were embedded into the media. The petri dishes were then held vertically in the room where intact plants were cultured. For each treatment, 8 plants were measured for growth of lateral branches every 24 h for 10 days by aligning a ruler behind the plates. Nodes 3 and 4 were harvested separately 4 h or 6 h after treatment for analysis of DgBRC1 or DgIPT3 transcripts. For each treatment, 10 plants were analyzed; all experiments were repeated for 3 biological replicates.

Analysis of gene expression
Plant tissues were harvested and total RNA was isolated using TRIzol Reagent (Invitrogen). Traces of DNA were digested using 5 units of DNaseI (Takara, D2270A) for 30 min, followed by extraction by Phenol/Trichloromethane. cDNA was synthesized with the Quantscript RT Kit (Tiangen, KR103-04) using Nonadeoxyribonucleotide Mixture (Takara, D3802). The cDNA was diluted 1:4 with water, and quantitative PCR was performed in 20 ml reactions with SYBRH Premix Ex Taq TM II (Takara, DRR081A) and the Applied Biosystems 7500 real-time PCR system according to the manufacturer's instructions. Ct values were obtained with the 7500 Systems SDS software 2.0.1 (Applied Biosystems). A standard curve for each target gene was generated using a dilution series of known concentrations of plasmid vectors containing target genes and measured by the same quantitative PCR process. Target Ct values were converted to absolute transcript numbers per reaction, and normalized to 18S rRNA transcript levels. Three biological replicates were measured for each treatment, and each replicate represented the RNA extracted from a pool of 10 buds. Primer sets are given in Table S3. The primers for total DgBRC1 transcripts analysis were picked from the same part of DgBRC1-1 and DgBRC1-2, while the sense or antisense primers for DgBRC1-1 were selected exactly in the position where intron I was spliced.

Statistics
ANOVA followed by a Duncan's test (for more than two comparisons) or a t test (two comparisons) was used (SPSS 18.0) where differences between means were assessed and significance was determined at a = 0.05. Figure S1 The length of all lateral branches in chrysanthemum plants of 45