Change in Auxin and Cytokinin Levels Coincides with Altered Expression of Branching Genes during Axillary Bud Outgrowth in Chrysanthemum

In the production and breeding of Chrysanthemum sp., shoot branching is an important quality aspect as the outgrowth of axillary buds determines the final plant shape. Bud outgrowth is mainly controlled by apical dominance and the crosstalk between the plant hormones auxin, cytokinin and strigolactone. In this work the hormonal and genetic regulation of axillary bud outgrowth was studied in two differently branching cut flower Chrysanthemum morifolium (Ramat) genotypes. C17 is a split-type which forms an inflorescence meristem after a certain vegetative period, while C18 remains vegetative under long day conditions. Plant growth of both genotypes was monitored during 5 subsequent weeks starting one week before flower initiation occurred in C17. Axillary bud outgrowth was measured weekly and samples of shoot apex, stem and axillary buds were taken during the first two weeks. We combined auxin and cytokinin measurements by UPLC-MS/MS with RT-qPCR expression analysis of genes involved in shoot branching regulation pathways in chrysanthemum. These included bud development genes (CmBRC1, CmDRM1, CmSTM, CmLsL), auxin pathway genes (CmPIN1, CmTIR3, CmTIR1, CmAXR1, CmAXR6, CmAXR2, CmIAA16, CmIAA12), cytokinin pathway genes (CmIPT3, CmHK3, CmRR1) and strigolactone genes (CmMAX1 and CmMAX2). Genotype C17 showed a release from apical dominance after floral transition coinciding with decreased auxin and increased cytokinin levels in the subapical axillary buds. As opposed to C17, C18 maintained strong apical dominance with vegetative growth throughout the experiment. Here high auxin levels and decreasing cytokinin levels in axillary buds and stem were measured. A differential expression of several branching genes accompanied the different hormonal change and bud outgrowth in C17 and C18. This was clear for the strigolactone biosynthesis gene CmMAX1, the transcription factor CmBRC1 and the dormancy associated gene CmDRM1, that all showed a decreased expression in C17 at floral transition and an increased expression in C18 with continuous vegetative growth. These results offer a case study for Chrysanthemum, showing an altered cytokinin to auxin balance and differential gene expression between vegetative growth with apical dominance and transition to generative growth with loss of apical dominance and axillary bud outgrowth. This suggests a conservation of several aspects of the hormonal and genetical regulation of bud outgrowth in Chrysanthemum. Furthermore, 15 previously uncharacterised genes in chrysanthemum, were described in this study. Of those genes involved in axillary bud outgrowth we identified CmDRM1, CmBRC1 and CmMAX1 as having an altered expression preceding axillary bud outgrowth, which could be useful as markers for bud activity.


Introduction
Chrysanthemum morifolium (Ramat), called the florist's chrysanthemum, is an economically important horticultural crop.Chrysanthemums show a great variation in sizes and shapes: garden and potted plants are highly branching, while cut flowers frequently show limited branching and sometimes even removal of axillary buds is required to obtain single flowered stems.Modification of plant architecture through shoot branching is a relevant factor in breeding and production of chrysanthemum.Shoot branching or axillary bud outgrowth in herbaceous shoots is regulated by a complex interaction of external factors (light, temperature, nutrients and pruning) and plant hormone signalling [1][2][3].The most prominent hormones associated with the regulation of bud outgrowth are auxins, strigolactones and cytokinins [3].Auxins are widely regarded to be responsible for apical dominance, the phenomenon where the growth of the vegetative shoot apex exerts a control over the outgrowth of axillary buds [4].Removal of the shoot apex and floral transition, releases the apical control over lateral bud outgrowth [5].Auxin regulation of apical dominance can be explained by the young expanding leaves and the shoot apex that produce auxin, which is transported through the stem towards the roots in a polar auxin transport stream facilitated largely by the auxin transport protein PIN1 [6,7] in the basal membranes of xylem parenchyma cells.On its way to the roots the auxin exerts an inhibition of the axillary bud outgrowth.
Since the basipetal transport does not deliver auxin into the axillary buds directly, an indirect action of auxin is suggested [8].In literature the indirect inhibition by auxin is explained by two non-mutually exclusive models: the second messenger model [9] and the canalisation model [10].
The canalisation model explains the inhibition of axillary bud outgrowth by the polar auxin stream in the stem that acts as an auxin sink.The shoot apex and the axillary buds are auxin sources that compete with each other for the ability to export auxin to the sink.Evidence for this model comes from the observations in Arabidopsis that strigolactones inhibit axillary bud outgrowth by reducing PIN1 mobilisation, as such restricting polar auxin transport [11,12].
The second messenger model states that a signal downstream of auxin is responsible for the inhibition of bud outgrowth.Both cytokinins and strigolactones control shoot branching downstream of auxins, and thus may be considered as secondary messengers.Cytokinins have a positive effect on the outgrowth of axillary buds.This is supported by observations in pea of stimulation of bud outgrowth upon exogenous application of cytokinins [13] and increasing cytokinin biosynthesis in stems and axillary buds at the time of outgrowth of axillary buds [14].As a response to auxin signalling, the biosynthesis of cytokinins is inhibited in Arabidopsis and pea [15,16], while its degradation is promoted in pea [16].Like auxins, strigolactones inhibit axillary bud outgrowth, which was shown in Arabidopsis, rice and pea [17,18] and the biosynthesis of strigolactones is upregulated by auxin in Arabidopsis and pea [19,20].
In this way the physiological regulation of shoot branching involves the activity of many genes involved in the local axillary meristem maintenance and in the pathways of auxin, cytokinin and strigolactones (Fig 1).The formation of axillary meristems in Arabidopsis, involves the lateral suppressor gene LAS [21], REVOLUTA (REV) and PHABULOSA (PHAB) [22].SHOOT MERISTEMLESS (STM) is another gene involved in the formation of axillary meristems [23] and can be used as an early marker for axillary meristem initiation [24].In Euphorbia esula, dormant adventitious buds were shown to have upregulated STM expression after defoliation treatment to induce bud growth [25].
A central regulator of axillary bud outgrowth is the transcription factor BRC1 (BRANCHED1) that is expressed locally in the axillary bud and inhibits outgrowth [26].Many of the regulatory pathways affect shoot branching by acting on the expression of BRC1.In Arabidopsis and pea, strigolactones promote BRC1 expression while cytokinins inhibit BRC1 expression [27,28].BRC1 is also involved in the floral transition as it is under control of the florigen pathway with a proposed interaction between FT (FLOWERING LOCUS T) and BRC1 whereby BRC1 is inactivated, promoting branching at floral transition [29].A dormancy marker similar to BRC1 is DRM1 (DORMANCY ASSOCIATED PROTEIN 1).DRM1 expression was shown to decrease in pea [30] and Arabidopsis [31] after the release of dormancy by decapitation.In wheat and sorghum, high expression of DRM1 and BRC1 have been reported in dormant axillary buds [32,33].
Auxin (IAA) biosynthesis in Arabidopsis involves the activity of the TAA (TRYPTOPHAN AMINOTRANSFERASE) and YUCCA genes [34].For auxin transport TIR3 (TRANSPORT INHIBITOR RESPONSE 3) encodes a protein that is required in the mobilization of PIN1 proteins in the cell membrane [35].Tir3 mutant plants showed reduced auxin transport and increased shoot branching in Arabidopsis [36].Auxin signalling involves TIR1 (TRANSPORT INHIBITOR RESPONSE 1), AXR1 (AUXIN RESISTANT 1) and AXR6 (AUXIN RESISTANT 6)that are required to form an F-box complex for ubiquitination of Aux/IAA repressors [37,38] leading to activation of auxin response genes [39].AXR1 was shown to be responsible for the auxin induced inhibition of axillary bud growth in Arabidopsis [40] and is required for the upregulation of strigolactone biosynthesis and downregulation of cytokinin biosynthesis [20].Aux/IAA proteins are degraded by the TIR1 F-box complex as a response to auxin [41][42][43].The expression of these Aux/IAA genes is upregulated by auxin and is useful as a marker for auxin signalling [44,45].
Strigolactone regulation of shoot branching involves the MORE AXILLARY GROWTH (MAX) genes in Arabidopsis and it's homologues RAMOSUS (RMS), DECREASED APICAL DOMINANCE (DAD) and DWARF (D) in pea, petunia and rice respectively.MAX3 and MAX4 convert carotenoid precursors and MAX1 that is involved in the final step of strigolactone biosynthesis [8,46,47].Strigolactone signalling involves an F-Box complex with MAX2 and DWARF14 (D14), allowing the degradation of SMAX1-like proteins to enable strigolactone response [48,49].
For cytokinin biosynthesis IPT (ISOPENTENYLTRANSFERASE) is the rate limiting step in the pathway [50].In Arabidopsis, the cytokinin receptors are histidine kinase (HK) proteins that include HK4, HK3 and HK2.These cytokinin receptors relay a phosphor signal to RESPONSE REGULATOR proteins (RR) that act as transcription factors for downstream cytokinin responses [51].Type A RR genes have been shown to be required for axillary bud activation by cytokinin [52].
In chrysanthemum only few genes involved in shoot branching have been isolated (Fig 1).CmBRC1's activity has been described in chrysanthemum with high expression in inhibited axillary buds and downregulation when buds were activated [53].A LATERAL SUPPRESSOR LIKE (LsL) gene was isolated in chrysanthemum [54] and plants transformed with the antisense LsL gene showed reduced branching and increased IAA content in the shoot tip while transformation with a sense LsL construct showed increased branching and reduced IAA content in the shoot tip [55].The strigolactone biosynthesis gene CmMAX3 showed upregulated expression in plants with inhibited axillary buds by phosphorous (Pi) starvation [56].CmMAX4 was studied in the strigolactone regulation of shoot branching and its expression was reported to be upregulated by exogenous auxin treatment [57].CmMAX2 was found to be expressed in axillary buds and stems and is necessary for the response to strigolactone, including inhibition of shoot branching [58].Overexpression of the CmIPT3 gene in chrysanthemum resulted in enhanced branching [59].
Aside from looking at hormone levels directly, the expression of these genes is interesting as a potential indicator of axillary bud activity.Because of the limited number of genes available in chrysanthemum, several key genes involved in the shoot branching pathways were isolated in this study and their gene expression levels were determined during branching.Two distinctive chrysanthemum genotypes with different branching patterns were analysed.In the first genotype, C17, the transition from vegetative to generative growth released apical dominance, while the second one, C18, remained vegetative.Axillary bud outgrowth measurements were related to levels of auxin and cytokinin in apex, stem and axillary bud samples in order to investigate the hormonal regulation during axillary bud outgrowth.The expression of candidate branching genes was analysed in the same samples to show the underlying transcriptional control during bud growth.

Plant Material and Growth Conditions
Chrysanthemum morifolium (Ramat) cut flower genotypes C17 and C18, provided by Dekker Chrysanten BV., The Netherlands) were used in this study.Both genotypes are commercial cultivars that have an unrelated genetical background in their breeding history.C17 is a split type chrysanthemum, forming an inflorescence meristem at the shoot apex once a number of vegetative leaves are produced.This results in a release from apical dominance with outgrowth of subapical axillary buds.C18 is a non-split type with vegetative growth during long day (LD) conditions and generative growth under short day (SD) conditions.A diagram of the growth habit of both genotypes under LD is shown in Fig 2 .Of each genotype 100 rooted cuttings were potted into 3L pots in standard peat substrate.Plants were grown at 20 cm spacing in a standard greenhouse under natural LD conditions and average day/night temperature of 20 ± 2°C.

Plant Morphological Measurements
The start of measurements and sampling was in the first week of July 2014, 2 weeks after potting.The total plant height and length of axillary buds/shoots was measured for 10 plants of each genotype during 5 consecutive weeks, referred to as: V1, T2, G3, G4, G5 (C17) or V1, V2, V3, V4, V5 (C18) with vegetative growth indicated by V, generative growth by G and the transition to generative growth indicated by T. The timing of floral transition in C17 was known from previous bud outgrowth experiments (data not shown).

Sampling Plant Material
In total, 90 plants were sampled per genotype at the first two time points (V1 and T2 for C17, V1 and V2 for C18).Three replicates were sampled per time point, each replicate consisting of 15 plants.For C17 at V1, the shoot apex (the top part of the shoot above the first fully unfolded leaf) and the 30 nodal positions under the apex were sampled.At T2, the apex and 15 nodal positions under the apex were sampled, as the lateral buds below this zone had already grown out.For C18 at V1, the shoot apex and 22 underlying nodal positions were harvested, while at V2, the apex and 27 subapical nodal positions were taken.For each nodal position, the axillary buds and stems were sampled separately.Samples were frozen in liquid nitrogen and stored at -80°C.Before analysis, the samples were pooled based on the pattern of axillary bud outgrowth until week 5.They were subdivided to represent vertical positions along the stem that are at different distances and influences from apical dominance.From top to bottom these were named Apex, Zone A, Zone B and Zone C (Fig 2 and Table 1).Pooled samples were ground in liquid nitrogen and separate samples were weighed for gene expression analysis (100mg/sample) and UPLC-MS/MS (50mg/sample).

Gene Isolation
For CmMAX1 degenerate primers were developed based on MAX1 amino acid sequences from Arabidopsis, tomato, poplar, soy, Medicago, vine and Ricinus S1 Table .Furthermore, a BLAST search was done to the chrysanthemum transcriptome database of Xu et al. [62] using Arabidopsis genes known to be involved in branching (corresponding accession numbers are indicated in Table S.2).This resulted in orthologous sequences for CmMAX3, CmDRM1, CmSTM, CmRR1, CmHK3a, CmHK3b, CmAXR1, CmAXR2, CmIAA16, CmAXR6, CmIAA12, CmPIN1, CmTIR1 and CmTIR3.Reference genes were identified by BLAST against the NCBI database.These included CmACT2, CmATUB, CmUBQ10, CmUBC, CmEF1α, CmCACS, CmEXP5, Table 1.Different zones and the corresponding nodal position from which stem and axillary bud samples were harvested in Chrysanthemum C17 and C18.For C17 in V1 nodal positions 37 to 31 (Zone A V1) were pooled together to represent inhibited axillary buds under apical dominance.Positions 26 to 16 in week 1 (Zone B V1) contain axillary buds that are further away from the shoot apex and show more outgrowth.Positions 13 to 8 (Zone C V1) represent bottom axillary buds inhibited by correlative inhibition of middle axillary buds.In T2 nodal positions 45 to 42 (Zone A T2) were pooled because the buds in this section showed the strongest outgrowth after T2 and release from apical dominance.Nodal positions 41 to 38 (Zone B T2) were pooled because these buds showed a diminishing outgrowth after week 2 when compared to the nodal positions above them.Nodal positions 37 to 31 (Zone C T2) were pooled to represent buds that show inhibited outgrowth after week 2. For C18 in V1 nodal positions 24-20 represent subapical inhibited buds (Zone A V1), positions 19-14 represent buds that are less inhibited by apical dominance (Zone B V1) and positions 13-8 represent buds furthest away from apical influence.In V2 positions 30-25 represent inhibited subapical buds (Zone A V2), positions 24-20 (Zone B' V2) and 19-14 (Zone B" V2) represent less inhibited buds with zone B' and B" the same positions as zone A and B in V1.
Zone A Nodal positions directly under the shoot apex and under apical dominance control.For C17 these are the positions that show release from apical dominance and strongest bud outgrowth after floral transition in T2.

24-20 30-25
Zone B Nodal positions further away from the shoot apex and apical dominance.For C17 T2 these positions show weaker bud outgrowth after floral transition in T2.For C18 V2 zone B was subdivided in 2 pools.Pool B' are the same positons as zone A from V1. Pool B" are the positions from zone B V1.

19-14 B' 24-20 B" 19-14
Zone C Nodal positions furthest away from the shoot apex and apical dominance.For C17 T2 these are positions that show inhibited bud outgrowth after floral transition in T2.

Gene Expression Analysis
RNA was extracted with a modified CTAB protocol (Luypaert et al., submitted).RNA concentrationand qualitywas verified with the NanoDrop spectrophotometer S3 Table .(Isogen Life Sciences) and Experion™ automated electrophoresis S1 Fig. using the StdSense analysis kit (Bio-Rad) was performed on a subset of samples as described in De Keyser et al. [64].
RNA was treated with DNase I according to manufacturer's protocol (DNA-free™, Ambion).DNase treated RNA (800ng) was converted to cDNA with the iScript cDNA Synthesis kit (Bio-Rad).The no reverse-transcriptase controls (noRTs) were made using a mixture of DNase treated RNA and nuclease free water (1:1).cDNA and noRT samples were diluted (1:5) prior to use.Expression of branching genes was quantified using the LightCycler 1 480 (Roche).In white plates (Bio-Rad) a final volume of 10 μl, containing 2 μl cDNA, 300nM of both primers, and 5 μl of SensiFAST™ SYBR 1 No-ROX (Bioline) was added.PCR amplification and melting curve analysis was done as described in De Keyser et al. [64].Primer sequences are listed in S4 Table .;besides the isolated genes also genes already known to be involved in axillary bud outgrowth in chrysanthemum were used: , CmIPT3 (Acc.Nr.JQ711176.1)and CmMAX2 [58].NTCs (no template controls) and noRTs were included for all genes; technical replicates were omitted.All stem or apex/bud samples, respectively, were analysed on the same plate for every gene (sample maximisation method).Due to limitations in the amount of cDNA available, the RT-qPCR analysis was done on two separate batches of cDNA (same RNA but 2 individual cDNA synthesis steps).
With the first batch, GeNorm [65] analysis was done and the expression of CmBRC1, CmIPT, CmLsL, CmMAX1 and CmMAX2 was quantified; the second batch was used for expression analysis of CmRR1, CmAXR1, CmAXR2, CmAXR6, CmHK3a, CmHK3b, CmDRM1, CmIAA12, CmIAA16, CmMAX3, CmPIN1, CmSTM, CmTIR1 and CmTIR3; reference genes were analysed again on these cDNA samples as well.For all gene expression data analysis, qbase + software (Biogazelle) [66] was used.In case the difference between the Cq of noRT and samples was smaller as 5, samples were excluded from further analysis.Mean gene-specific amplification efficiencies S5 Table .were determined using LinRegPCR [67,68].Gene expression results are presented as Calibrated Normalized Relative Quantities, normalised using 3 validated reference genes for stem and apex/bud samples S6

C17 Initiates Axillary Bud Outgrowth at Floral Transition and C18 Shows Continuous Vegetative Growth and Apical Dominance
Two different genotypes were used for branching analysis.Genotype C17 shifted from vegetative growth with apical dominance (V1) to generative growth with loss of apical dominance at T2 (floral transition) and consecutive outgrowth of axillary buds (G3-G5).Genotype C18 showed vegetative growth throughout all time points V1-V5.(Fig 2).The total plant height increased for C17 from 39.9 ± 1.5 cm at V1 to 53.1 ± 1.5 cm at T2 and 60.9 ± 2.1 cm at G3. Afterwards total length remained stable (61.0 ± 2.2 cm at G4 and 61.5 ± 2.4 cm at G5). C17 was characterised by vegetative growth at V1 and floral transition at T2.Thereafter, the shoot apex no longer initiated axillary buds and total number of nodal positions stayed at 47 from G3 to G5 (Fig 3).The vegetative growth was marked by apical dominance and short inhibited axillary bud outgrowth near the shoot apex in V1 (positions 37 to 31) and T2 (positions 45 to 38).Once generative growth started at G3 to G5, apical dominance was released and a strong outgrowth of the axillary shoots under the apex (positions 47 to 42) was observed.At V1 and T2, the axillary buds in the middle positions were under less influence by the apex and had larger axillary buds (positions 30 to 16 for V1 and 37 to 31 for T2) with even bud outgrowth in T2 at positions 30 to 16.The positions 15 to 1 under the middle showed strongly inhibited bud outgrowth as they were influenced by correlative inhibition (actively growing buds inhibiting the outgrowth of underlying axillary buds).From G3 to G5, shoot outgrowth decreased from position 41 to 38 and was completely inhibited for the lower positions (37 to 1).
Plant heights for C18 during vegetative growth from V1 to V5 were 41.8 ± 3.1 cm, 58.3 ± 3.4 cm, 72.4 ± 3.3 cm, 83.6 ± 4.4 cm and 93.5 ± 8.1 cm during V1 to V5, respectively.The shoot apex did not undergo floral transition and continuously initiated new leaves.At V5, the number of nodal positions reached a maximum of 53.A strong apical dominance was manifested during the whole vegetative growth of C18 (Fig 4).This was revealed by the short axillary buds under the apex and an average axillary bud outgrowth of 1.27 ± 0.03 cm at V5.
The measurements of axillary buds were used to show the inhibition or activation of the nodal positions.Based on these measurements the nodal positions along the stem were subdivided in Zone A, Zone B and Zone C according to bud outgrowth activity and influences from apical dominance as described in Table 1.In C17 the nodal positions 30-1 at T2 (Fig 2 ) were not included because bud outgrowth had already occurred there and only samples of axillary buds were taken.These subdivisions (also indicated in Figs 2-4) were subsequently used in the hormone measurements and gene expression analysis.

Levels of IAA and CK Change Differently at Floral Transition in C17 and during Vegetative Growth in C18
Raw data of CK and IAA measurements are provided in the supplemental information S7 Table .One of the three biological replicates was excluded for IAA (in C17 T2 Bud and Stem Zone A) and for CK (in C17 T2 Stem Zone B and in C18 V2 Bud Zone A); for these samples only 2 biological replicates were averaged.
Overall, IAA levels were higher in C18 than in C17, especially in apex and stem (Fig 5A and 5B).For C17 at V1, IAA content in the shoot apex was significantly higher than in the axillary buds and significantly lower compared to the stem (Table 2).At T2 the IAA content in the apex had decreased and was similar to IAA levels in axillary buds and stem, except for zone B. Between different zones of axillary buds, IAA levels were higher in zone A at V1 but not different at T2. Between stem zones, IAA levels were similar at V1 but higher in zone C at T2 S8 Table.
At the transition to generative growth in C17, from V1 to T2, a decrease in IAA content was observed in the apex and axillary buds of zone A, while a slight increase was seen in zone B and C. In the stem a significant decrease in auxin level occurred in zone A and B (Fig 5A).
IAA content in the shoot apex of C18 was significantly higher than in the axillary buds but did not differ from the IAA content in the stem at V1 (Table 2).At V2, the IAA content in the shoot apex did not differ significantly from the axillary buds or stem (Table 2).Between zones of axillary buds, no differences were seen at V1 or V2.Between stem zones no differences were seen at V1 but at V2 zone B' and B" differed significantly from zone C S8 Table.
In C18 there was no significant difference in IAA levels between V1 and V2 in the apex and axillary buds.In the stem a significant decrease was observed in zone B" and zone C but the absolute auxin level remained high compared to the stem zone A and B of C17 (Fig 5B ).Cytokinin (CK) content was generally higher in C17 when compared to C18, especially in apex and axillary buds (Fig 5C and 5D).CK content in the shoot apex was significantly higher than in the axillary buds (except for zone A) and the stem in C17 at V1 (Table 2).At T2, CK content in the shoot apex remained significantly different from the axillary buds (except zone A) and stem (except zone B) (Table 2).Between different zones of axillary buds, CK levels   For C18 the CK content in the apex showed no significant differences with axillary buds and stem at V1 or V2 (except with zone C at V2). Between zones of axillary buds or stem no differences were seen at V1 or V2 S8 Table.
In C18, CK levels had a decreased trend from V1 to V2 throughout the axillary buds and stem with significant changes in bud Zone B' and stem zone C.     3).Additionally, the functional domains were indicated on the isolated sequences, corresponding to the same functional domains in the Arabidopsis sequences S2 Fig.For CmHK3, two different fragments CmHK3a and CmHK3b were isolated which shared a 96.99% identity in their nucleic acid sequence (Clustal Omega [69]).

Gene Expression Analysis
Rt-qPCR was done using the 15 isolated genes and 4 genes previously isolated in chrysanthemum: CmBRC1 [53], CmLSL [54], CmIPT3 (Acc.Nr.JQ711176.1)and CmMAX2 [58].These genes represent key signalling functions in bud development and the pathways of auxin, cytokinin and strigolactones (Fig 1B).In the expression analysis of CmMAX3 there was amplification in the no-RT for the majority of the samples and the ΔCp cut-off-value of 5 was not met.Therefore CmMAX3 expression was excluded from the analysis.Similarly, one of the three biological replicates was excluded for CmAXR1 (in C17 V1 Bud Zone C), CmHK3a (in C17 V1 Bud Zone A), CmTIR3 (in C17 V1 Bud zone A and in Bud Zone C V1 and T2), CmAXR2 (in C18 V2 Bud zone B") and CmIPT3 (in C18 V2 Bud Zone A); for these samples only 2 biological replicates were averaged.
The expression of these genes is presented separately for C17 and C18 with significant differences between V1 and T2/V2 indicated on the figures (Figs 7-12).Raw CNRQ expression values are available in S10 Table .Additional comparisons were made between the apex and zones of axillary buds and between the zones of the stem at each time point and between time points for C17 (S11-S13 Tables) and C18 (S14-S16 Tables).
Expression of bud development genes.The expression of the bud development related genes CmBRC1 and CmDRM1 showed a different pattern in C17 compared to C18.A decreased trend from V1 to T2 was seen in the expression of CmBRC1 (Fig 7A ) and CmDRM1 (Fig 7C ), while an increased trend was observed in C18 (Fig 7B and 7D).This decreased expression in C17 was most pronounced in the stem with significant changes in all zones for CmBRC1 and in zone B and C for CmDRM1.In C18 increased expression of CmBRC1 was significant in axillary bud zone A and B', while CmDRM1 expression increased significantly in apex, axillary bud zone A and stem zone B" and C.
CmLsL expression decreased in C17 from V1 to T2 in the axillary buds of zone A, B and C and in the stem zone B (Fig 7E).In C18 CmLsL expression increased in shoot apex and in the axillary buds of zone C from V1 to V2 (Fig 7F).CmSTM expression increased both in C17 (from V1 to T2 in apex, bud zone C and stem zone A,B) and in C18 (from V1 to V2 in apex, bud zone B, C and stem zone A).
Expression of strigolactone genes.The strigolactone biosynthetic gene CmMAX1 was expressed differently in C17 and C18.A decreasing trend from V1 to T2 in C17 (Fig 8A and  8E) and an increased expression from V1 to V2 in C18 (Fig 8B and 8F) was observed.Decreased expression in C17 was significant in bud zone B and stem zone A and C for CmMAX1.Increased expression in C18 was significant in bud zone A and stem zone B" for CmMAX1.The expression of the strigolactone signalling gene CmMAX2 increased in C17 from V1 to T2 and in C18 from V1 to V2.
Expression of cytokinin genes.Expression of cytokinin biosynthesis gene CmIPT3 in C17 showed an overall decrease from V1 to T2 (Fig 8A).In C18 no significant difference in expression was observed between V1 and V2, except in zone A and B of the stem (Fig 8B).Cytokinin receptor genes CmHK3a and CmHK3b (2 fragments of the CmHK3 gene) showed similar expression patterns in C17 and C18 but with higher expression of CmHK3a in C18 compared to CmHK3b.In C17 expression increased in apex and decreased in the stem (Fig 8E and 8G).In C18, an increased trend was seen from V1 to V2 in apex, bud and stem (Fig 8F and 8H).The cytokinin The auxin perception genes CmTIR1 and CmAXR1 had a similar expression pattern in C17 and C18.Their expression in C17 from V1 to T2 remained low or largely unchanged (Fig 11A and 11C), while in C18, an increased trend was seen from V1 to V2 (Fig 11B and 11D).CmAXR6 expression in C17 showed little change from V1 to T2 with a significant increase in stem zone A and a significant decrease in stem zone C (Fig 11E).In C18 no significant changes were seen from V1 to V2 (Fig 11F).
The auxin response genes CmAXR2, CmIAA16 CmIAA12 showed a largely similar pattern of expression with little change in C17 and an increased trend in C18.This was most evident

Discussion
The axillary shoot growth observed in both chrysanthemum genotypes C17 and C18 was determined by apical dominance during vegetative growth.This was shown in C17 by the inhibition of axillary buds immediately under the apex, the larger buds in the middle part and the inhibited lower buds at V1 (Fig 2).In C18 the stronger apical dominance was evidenced by the overall inhibited axillary buds in the vegetative growth (Fig 4).These branching forms matched with previous axillary bud/shoot length measurements in chrysanthemum cut flowers [53,70].In C17 transition to the generative growth phase at T1 caused a release from apical dominance followed by the growth of axillary shoots directly under the apex.This concurs with previous measurements of subapical axillary bud outgrowth in chrysanthemum after release from apical dominance by decapitation and floral transition [53].At the end of the experiment, both genotypes exhibited their typical growth habitus, a split type for C17 and a continuous vegetative growth under long day for C18.
The hypothesis of apical dominance where a high auxin production in the shoot apex and subsequent basipetal auxin transport inhibits the outgrowth of axillary buds is supported by various studies on vegetative axillary buds [70][71][72].The reported IAA contents in the shoot apex, stem and axillary buds in this study were in agreement with this apical dominance mechanism (Fig 5).The higher IAA content measured in the shoot apex of C18 compared to C17 corresponded with the stronger apical dominance observed in C18.The floral transition initiated at T2 in C17 coincided with a decreased IAA content in the shoot apex and in the stem.In C18 no floral transition took place and correspondingly no drastic decrease of IAA content was observed from V1 to V2.
Cytokinins are known regulators of axillary bud outgrowth and cytokinin homeostasis is also influenced by auxin that can inhibit biosynthesis and promote degradation [27].The increased CK levels observed in the axillary buds and stem in zone A of C17 at the moment of floral initiation (T2) reflect this axillary bud outgrowth promoting role.In vegetatively growing C18 plants, CK levels decreased in the axillary buds and stem from V1 to V2.Furthermore, lower CK levels were measured in the shoot apex, axillary buds and stem of C18, which exhibited a stronger apical dominance than C17 (Fig 5).Like these observations, previous reports have shown increased cytokinin levels in axillary buds at the release from apical dominance by apex defoliation in tobacco plants [73] and by decapitation in pea plants [14].
The CK/IAA ratio offers a good indicator for the state of apical dominance and axillary bud outgrowth as was shown previously by Emery et al. [72] in lupin.In that case inhibited axillary buds showed a decreased CK/IAA ratio, whereas activated axillary buds showed increasing CK/IAA ratios.An increased CK/IAA ratio was also described in the outgrowth of axillary buds of tobacco following release from apical dominance [73].In chrysanthemum we observed a strong increase in CK/IAA ratio in C17 in the subapical axillary buds of zone A at the time of release from apical dominance by floral initiation.In C18, where floral transition is absent, there was a notable decrease of the CK/IAA ratio in the axillary buds (Fig 6).
In the subdivision of the different samples of axillary buds and stem into the zones A, B and C it could be remarked that in C17 the nodal positions at T2 do not match those in V1.These positions however where chosen to reflect a physiological change that was seen in the axillary bud outgrowth under the apex at the transition from vegetative growth to generative growth.Furthermore middle positions in T2 already showed bud outgrowth with side shoots.These nodes were omitted from the sampling to only include axillary buds.For this reason, the nodal positions in zone A at V1 are the same positions that become zone C in T2.The different physiological status that occurs at floral transition in T2 is however better reflected by the change from zone A V1 to zone A T2, with decreased auxin content and increased cytokinin content in axillary bud and stem.This change is not seen from zone A V1 to zone C T2 S8 Table .and this comparison would disregard information about the hormonal regulation that happens in the axillary buds under the apex that show a strong outgrowth after floral transition.This was also shown by the changes in gene expression that occurred from V1 to T2 in zone A, and not from zone A V1 to zone C T2 S13 Table ., coinciding with the changing hormone levels.
In C17 CmBRC1, CmDRM1 and CmMAX1 expression showed a decreased trend from V1 to T2, while in C18 their expression had an increased trend or remained constant from V1 to V2 (Figs 7 and 8).This indicates that CmBRC1 and CmDRM1 are downregulated at transition to generative growth and release from apical dominance in C17, while upregulation is seen in C18 during continuous vegetative growth.This corresponds with previous reports of high BRC1 expression in chrysanthemum [53] and high DRM1 expression in pea and Arabidopsis [30,31] in inhibited axillary buds and downregulation in activated buds.
It has to be noted here that in our results the most striking decrease in CmBRC1 expression at floral transition happened in the stem samples, while in the axillary buds only a slight decrease was observed in the axillary bud samples of zone A and B with even an increase in zone C.This observation can be seen as somewhat contradictory with earlier reports of the role of BRC1.BRC1 was found to be mainly expressed in dormant axillary buds in chrysanthemum and showed decreased transcript levels from 1 hour after release from apical dominance by decapitation [53].Interestingly, in the same study, 48 hours after decapitation, the BRC1 transcript levels had returned to the same levels as before the decapitation.This reported rapid downregulation of BRC1 could be a possible explanation why we could not see a strong decrease in the axillary buds, assuming that CmBRC1 levels had already reverted to the state before release of apical dominance by floral transition.On the other hand, it has been observed that in some cases the correlation between BRC1 expression and axillary bud activity was weak.In rice for instance, expression of the BRC1 orthologue FC1, is not reduced in dwarf10 (d10/ max4) mutants that show increased shoot branching [74].Also in maize the TB1 expression did not decrease in ccd8 (max4) mutants with increased branching phenotype [75].In our case it cannot be excluded that CmBRC1 expression is correlated with axillary bud activity since a decrease in expression is still seen in the stem.To further study this relationship it would require samples at short intervals within 24 hours of the floral initiation to account for the rapid changes in CmBRC1 expression.
The observed downregulation of CmMAX1 in C17 suggests that local strigolactone biosynthesis in the stem and axillary buds decreases during transition to generative growth and release from apical dominance, preceding bud outgrowth in the axillary buds under the apex.This is the first description of MAX1 in chrysanthemum.Given the role of strigolactones to inhibit axillary bud outgrowth [17] and also the regulation by MAX1 in Arabidopsis [76] and in Petunia [46], our results suggest that the release from apical dominance at floral transition observed in chrysanthemum coincides with a reduced strigolactone inhibition as shown by a reduced CmMAX1 expression.Expression levels of MAX1 during floral transition have not been previously reported but in red clover TpMAX1-1 expression as well as TpBRC1 expression were downregulated in nodal stem segments 24 hours after bud outgrowth was induced by node excision [77].
The gene involved in auxin transport, CmTIR3, showed an increased expression from V1 to V2 in C18 while expression remained largely unaltered in C17 (Fig 10C and 10D).This observation corresponds well with the reported increased branching phenotype of Arabidopsis tir3 mutants at floral transition [12].For CmPIN1 however, a general increase in expression was seen in C17 while the CmPIN1 expression in C18 remained constant (Fig 10A and 10B).It would indicate an increased auxin transport in C17 even though decreased auxin levels were observed here.This observation seems contradictory to what is known about auxin canalisation, where auxin promotes the transcription of PIN1 through a positive feedback loop [78,79].Furthermore it has been shown in Arabidopsis leaves that PIN1 expression was decreased in Tir3 mutant background, [80], so that a similar expression pattern of CmPIN1 to CmTIR3 could be expected.
The expression of auxin signalling genes CmTIR1 and CmAXR1 and showed an increase in C18 from V1 to V2 while in C17 this was not observed.The general expression of these genes was also higher in C18 compared to C17 ( Fig 11).For these genes however, previous reports have shown that auxin treatments did not increase their expression [38,81,82], making it less likely that the increase seen in C18 reflects a response to increased auxin levels.
CmAXR2 and CmIAA16 expression had an increased trend mainly in the stem of C18 (Fig 12).These Aux/IAA proteins function as repressors of auxin response but their transcription is induced by auxin [44,45].Therefore the upregulation could be sign of a feedback to the auxin levels that remained high in C18 compared to C17.This pattern was less pronounced in the expression of CmIAA12.
The cytokinin genes showed a somewhat counterintuitive gene expression (Fig 9).For CmIPT3 the decreased expression as seen in C17 stem and axillary buds from V1 to T2 should indicate decreased cytokinin biosynthesis.This is opposite from what would be expected by looking at the cytokinin levels that are increased in the top axillary buds and stem from week1 to week2.However in chrysanthemum CmIPT3 has been previously reported to be upregulated in nodes after decapitation [53].In Arabidopsis, IPT3 is mainly expressed in the phloem and pericycle cells [83] and is upregulated by high nitrate [84].Other cytokinin biosynthetic genes that are more important than IPT3 might be involved in the cytokinin biosynthesis at the nodal positions.Possible candidates are IPT1 and IPT2 that are associated with raised cytokinin levels and biosynthesis in pea shoots preceding bud outgrowth after decapitation [14].The cytokinin response genes CmRR1, CmHK3a and CmHK3b showed a general increased expression in C18 and some significant decreases in expression in C17.Like CmIPT3 these results are not what would be expected from the observed cytokinin levels that increased in C17 and decreased in C18.Although it has been reported in Arabidopsis that several cytokinin response genes were downregulated in treatments with exogenous cytokinin [85].A possible explanation could be that this cytokinin response can proceed without new protein synthesis.CmHK3a and CmHK3b represent 2 fragments that were isolated from the CmHK3 gene and they share a 97% identity.The expression pattern of both fragments was similar along the apex, axillary buds and stem of genotypes C17 and C18 but in C18 the expression levels of CmHK3a were generally higher than CmHK3b.This suggests that CmHK3a has a different expression from CmHK3b and that these fragments could represent different forms of the CmHK3 gene such as is the case with the alternatively spliced ZmHK3a and ZmHK3b in maize [86] or the homologous LjHK3a and LjHK3b loci in lotus [87].
CmLsL was previously shown to be involved with shoot branching in chrysanthemum [55].Overexpression resulted in increased branching and LsL expression was negatively correlated with IAA content in the shoot tip.From this we would expect to see an increase in CmLsL expression in C17 and a decrease in C18.In our experiments however we observed a general decrease in LsL expression in C17 which was not seen in C18 (Fig 9E and 9F).A possible explanation for this can be that LsL is mainly involved in the early vegetative growth and in the formation of axillary meristems [21,55], while our samples were taken later in vegetative growth with axillary meristems already established and developed to axillary buds.Similarly, STM also represents a gene that is involved in axillary meristem formation [23].This could account for the lack of differential expression that was observed with CmSTM as there was a general increased expression in both C17 and C18.

Conclusions
Our results showed that at floral transition in chrysanthemum genotype C17 the release from apical dominance with subsequent outgrowth of axillary buds coincided with an increased CK/ IAA ratio.Conversely, in genotype C18, continuation of vegetative growth was accompanied by a decreased CK/IAA ratio.The expression of several genes, most notably CmBRC1, CmDRM1 and CmMAX1, showed a differential expression pattern that coincided with the changes in CK/IAA ratio, with decreased expression at floral transition in C17 and increased expression with continued vegetative growth in C18.Furthermore, expression of auxin genes CmTIR3, CmAXR2 and CmIAA16 corresponded with a higher auxin status in C18 during vegetative growth, compared to C17.The expression of CmBRC1, CmDRM1 and also CmMAX1, which was previously not yet reported in chrysanthemum, could be used as early indicators of bud outgrowth activity.

Fig 1 .
Fig 1. Key shoot branching regulatory pathways and involvement of the branching genes used in this study of axillary bud outgrowth.The left panel shows the shoot branching model from what is known in other species, while the right panel shows the branching genes that have been described in chrysanthemum with the genes that were isolated in this study underlined.① indicates the bud outgrowth regulation that follows the auxin canalisation model where auxin export is required for an axillary bud to grow out.② indicates the regulation of bud outgrowth according to the second messenger model where auxin indirectly regulates bud outgrowth through strigolactone and cytokinin signalling.doi:10.1371/journal.pone.0161732.g001

Fig 3 .
Fig 3.  Axillary bud lengths for chrysanthemum C17 in V1, T2 and G3 to G5. Mean bud/shoot length (±SE; n = 10) is presented on the xaxis for every nodal position presented on the y-axis with 1 being the node closest to the shoot base.The dotted lines mark the nodal positions that were sampled in V1 and T2 according to Table1.doi:10.1371/journal.pone.0161732.g003

Fig 4 .
Fig 4. Axillary bud lengths for chrysanthemum C18 in V1 to V5. Mean bud/shoot length (±SE; n = 10) is presented on the x-axis for every nodal position presented on the y-axis with 1 being the node closest to the shoot base.The dotted lines mark the nodal positions that were sampled in V1 and V2 according to Table 1.doi:10.1371/journal.pone.0161732.g004 At floral transition from V1 to T2 in C17, the CK content increased significantly in the axillary buds of zone A and C and in the stem of zone A (Fig 5C).

Fig 5 .
Fig 5. Hormone concentrations measured in two Chrysanthemum genotypes C17 and C18 at time point V1 and T2/ V2.IAA and cytokinin (CK) content of the shoot apex and the axillary buds or stem in different zones are presented.Data are means ± SE (n = 3).* indicates significance at the 0.05 level between V1 and T2/V2.doi:10.1371/journal.pone.0161732.g005 Fig 6.CK/IAA ratio for Chrysanthemum genotypes C17 and C18 at two time points V1 and T2/V2.Total cytokinin/IAA of the shoot apex and the axillary buds or stem in different zones are presented.Data are mean cytokinin levels (n = 3) divided by mean auxin levels (n = 3) + calculated SE. doi:10.1371/journal.pone.0161732.g006

Fig 7 .
Fig 7. RT-qPCR gene expression analysis of bud development genes for Chrysanthemum genotypes C17 and C18 at time points V1 and T2/V2.CNRQ for bud development related genes CmBRC1, CmDRM1, CmLsL and CmSTM in the shoot apex, axillary buds and stem samples.Data are means ± SE (n = 3) of non log-transformed CNRQ.Fold changes between V1 and T2/V2 are indicated above the grey bars with * indicating significance of the Kruskal Wallis test (p<0.05).doi:10.1371/journal.pone.0161732.g007

Fig 8 .
Fig 8. RT-qPCR gene expression analysis of strigolactone genes for Chrysanthemum genotypes C17 and C18 at time points V1 and T2/V2.CNRQ for strigolactone genes CmMAX1 and CmMAX2 in the shoot apex, axillary buds and stem samples.Data are means ± SE (n = 3) of non log-transformed CNRQ.Fold changes between V1 and T2/V2 are indicated above the grey bars with * indicating significance of the Kruskal Wallis test (p<0.05).doi:10.1371/journal.pone.0161732.g008

Fig 10 .
Fig 10.RT-qPCR gene expression analysis of auxin transport genes for Chrysanthemum genotypes C17 and C18 at time points V1 and T2/V2.CNRQ for auxin transport genes CmPIN1 and CmTIR3 in the shoot apex, axillary buds and stem samples.Data are means ± SE (n = 3) of non log-transformed CNRQ.Fold changes between V1 and T2/V2 are indicated above the grey bars with * indicating significance of the Kruskal Wallis test (p<0.05).doi:10.1371/journal.pone.0161732.g010

Fig 11 .
Fig 11.RT-qPCR gene expression analysis of auxin signalling genes for Chrysanthemum genotypes C17 and C18 at time points V1 and T2/V2.CNRQ for auxin perception genes CmTIR3, AXR1 and AXR6 in the shoot apex, axillary buds and stem samples.Data are means ± SE (n = 3) of non log-transformed CNRQ.Fold changes between V1 and T2/V2 are indicated above the grey bars with * indicating significance of the Kruskal Wallis test (p<0.05).doi:10.1371/journal.pone.0161732.g011

Fig 12 .
Fig 12. RT-qPCR gene expression analysis of auxin response genes for Chrysanthemum genotypes C17 and C18 at time points V1 and T2/V2.CNRQ for auxin response genes CmAXR2, CmIAA16 and, CmIAA12 in the shoot apex, axillary buds and stem samples.Data are means ± SE (n = 3) of non logtransformed CNRQ.Fold changes between V1 and T2/V2 are indicated above the grey bars with * indicating significance of the Kruskal Wallis test (p<0.05).doi:10.1371/journal.pone.0161732.g012 Table.; different normalisation factors were used for batch 1 and 2. All further analysis was done in MS Excel.
tering of the genes was done based on a Pearson correlation matrix.Discriminant analysis was performed in order to discriminate the global gene expression between tissues and time points.For statistical analysis of gene expression the log-transformed CNRQ values were used.

Table 2 .
Kruskal-Wallis tests comparing the hormone levels between the shoot apex and the different zones of the axillary buds and stems.Data are fold changes (Zone X/Apex) and the significant difference between means by Kruskal-Wallis test is indicated by * (p-value<0.05).Changes in CK/IAA ratio reflected the altered hormone levels in C17 at floral transition and in C18 during vegetative growth.In C17 the CK/IAA ratio increased from V1 to T2, most notably in the apex and in zone A and B of both buds and stems (Fig6A).The CK/IAA ratio in C18 decreased from V1 to V2 in the axillary buds and in the apex and stem (Fig6B).