A natural frameshift mutation in Campanula EIL2 correlates with ethylene insensitivity in flowers

Background The phytohormone ethylene plays a central role in development and senescence of climacteric flowers. In ornamental plant production, ethylene sensitive plants are usually protected against negative effects of ethylene by application of chemical inhibitors. In Campanula, flowers are sensitive to even minute concentrations of ethylene. Results Monitoring flower longevity in three Campanula species revealed C. portenschlagiana (Cp) as ethylene sensitive, C. formanekiana (Cf) with intermediate sensitivity and C. medium (Cm) as ethylene insensitive. We identified key elements in ethylene signal transduction, specifically in Ethylene Response Sensor 2 (ERS2), Constitutive Triple Response 1 (CTR1) and Ethylene Insensitive 3- Like 1 and 2 (EIL1 and EIL2) homologous. Transcripts of ERS2, CTR1 and EIL1 were constitutively expressed in all species both throughout flower development and in response to ethylene. In contrast, EIL2 was found only in Cf and Cm. We identified a natural mutation in Cmeil2 causing a frameshift which resulted in difference in expression levels of EIL2, with more than 100-fold change between Cf and Cm in young flowers. Conclusions This study shows that the naturally occurring 7 bp frameshift discovered in Cmeil2, a key gene in the ethylene signaling pathway, correlates with ethylene insensitivity in flowers. We suggest that transfer of the eil2 mutation to other plant species will provide a novel tool to engineer ethylene insensitive flowers. Electronic supplementary material The online version of this article (doi:10.1186/s12870-016-0786-4) contains supplementary material, which is available to authorized users.


Background
Ethylene is a gaseous phytohormone involved in regulating processes of horticultural importance encompassing flower development, fruit ripening, abscission and leaf and flower senescence [1]. In the ethylene signal transduction pathway, ethylene perception is facilitated via a copper co-factor present in receptor proteins integrated in the endoplasmic reticulum (ER) [2,3]. In Arabidopsis, receptor proteins comprising EThylene Response 1 (ETR1) and Ethylene Response Sensor 1 (ERS1) or ETR2, ERS2 and Ethylene INsensitive 4 (EIN4) have been characterised. They differ by the functionality of their kinase domains [4][5][6]. Ethylene receptors exist as dimers and physically interact with the negative regulator Constitutive Triple Response 1 (CTR1) [7]. The kinase activity of CTR1 is directed towards the C-terminal of Ethylene INsensitive 2 (EIN2), a positive regulator of the ethylene response [8][9][10]. Ethylene binding to receptors deactivates CTR1 and results in a dephosphorylation of EIN2 [10]. Subsequently, the C-terminal of EIN2 is cleaved and translocated from ER to the nucleus [11,12] where Ethylene INsensitive 3/Ethylene Insensitive 3-Like (EIN3/EIL)-dependent transcription and activation of the ethylene response occur [13][14][15].
Postharvest quality of many ornamental plants is sensitive to ethylene during production and distribution [16,17]. In climacteric plants, flower development is controlled by intrinsic rise in ethylene production and respiration which promotes flower development and senescence. Plant species with climacteric flower senescence are sensitive to exogenous ethylene and may exhibit accelerated petal or flower wilting upon exposure. Commercially important climacteric ornamental plants include carnations, orchids, Kalanchöe, Campanula and roses [18][19][20][21]. Endogenous ethylene production may arise due to natural floral development but also in response to stress, elevated CO 2 production [22] or increased auxin production [23]. Hence, ornamental plants are often treated with chemical inhibitors blocking ethylene signaling to improve postharvest quality and prolong flower longevity [24].
Genetic approaches designed to reduce ethylene sensitivity in flowers have modified signaling via the ethylene signal transduction pathway. The etr1-1 ethylene receptor mutant from Arabidopsis fails to bind ethylene [25]. Expression of etr1-1 in Petunia and Campanula carpatica flowers [26][27][28] results in ethylene insensitivity, delayed senescence and postponed flower abscission. Also, transgenic Petunia expressing reduced levels of PhEIN2 displayed delayed flower senescence [29]. However, to date genetic approaches successfully prolonging flower longevity have resulted in transgenic plants [30].
Campanula is an economically important ornamental plant, used as indoor potted plant, garden plant, as well as cut flower. The Campanula genus consists of The position of the deletion in Cmeil2 is underlined. The alignment was produced in Clustal Ω [62] approximately 415 species [31]. Here, we characterise expression patterns of flower expressed ERS, CTR and EIL genes in response to floral development and exogenous ethylene in three ornamental species of Campanula; C. portenschlagiana (Cp), C. formanekiana (Cf) and C. medium (Cm). The ethylene insensitivity identified in flowers of C. medium correlates with the occurrence of a natural mutation in the open reading frame of EIL2. This finding holds promise for new breeding strategies towards ethylene insensitive ornamental plants.

Campanula sensitivity to ethylene
To understand the physiological variation in ethylene sensitivity among Campanula species, we used ethylene exposure tests in a postharvest environment. C. portenschlagiana, C. formanekiana and C. medium were selected due to their relevance as ornamental plants. Cp was found to be sensitive to ethylene from concentrations of 0.05 μL · L −1 . Individual flower sensitivity increased with flower age. In Cp, old flowers did not survive 0.05 μL · L −1 ethylene treatment for 48 h whereas young flowers maintained longevity in a 0.1 μL · L −1 ethylene environment for more than 48 h. Cp flowers, regardless of age, did not survive after 72 h of the high ethylene treatment (Fig. 1a, d). Less pronounced ethylene sensitivity was found in Cf where 26 % of old flowers wilted in response to 72 h of 0.05 μL · L −1 ethylene as opposed to 100 % of old Cp flowers. Increased ethylene concentration for the same period resulted in complete senescence of 4-day old Cf flowers (Fig. 1b, e). As in Cp, young Cf flowers were less ethylene sensitive than old flowers, however, 93 % of young flowers wilted in response to 0.1 μL · L −1 ethylene (Fig. 1e). Thus, flowers of neither Cp nor Cf could tolerate 72 h of 0.5 μL · L −1 ethylene, regardless of flower age. In contrast, Cm flowers were non-responsive to ethylene, they maintained both colour and turgor for 72 h in the 0.1 μL · L −1 ethylene environment (Fig. 1c, f ).

Identification of key genes in ethylene signal transduction
Degenerate primers were used to identify expressed components in the ethylene signal transduction pathway in Campanula flowers. This allowed identification of partial homologs for ERS2, CTR1 and EIL. Campanula transcript fragments of ERS2 and CTR1 were translated to protein and named according to their closest relative in Arabidopsis (Additional files 1 and 2) whereas Campanula EIN3/ EIL homologs were named EIL1 and EIL2. Protein alignment of Campanula ERS2 showed high similarity within species in the identified region whereas Campanula CTR1 proteins differed (Additional files 1 and 2). In Cf and Cm, ERS2 was encoded by a single gene whereas Cp ERS2 was represented by two loci containing different introns but resulting in identical partial transcripts (Additional file 3). Sequencing showed some polymorphisms among the partial CTR1 transcripts. These could not be separated in RT-PCR reactions and may represent different alleles in the same locus.
Also EIL transcripts were identified by degenerate primers using flower cDNA as template. Sequencing identified two partial EIL1 homologs EIL1a and EIL1b in Cp and only one partial EIL1 homolog in Cf and Cm. In Cf the cDNA pool contained an additional EIL homolog, EIL2, however this transcript was not readily detectable in Cm cDNA using the degenerate primers. As Cf and Cm have different sensitivities towards ethylene (Fig. 1e, f), primers were designed to separate and amplify both EIL1 and EIL2 fragments from Cf and Cm genomic DNA. Interestingly, Cmeil2 was found to contain a deletion of 7 bp in the EIL2 ORF resulting in a frame shift in the corresponding protein  formanekiana and c C. medium (Cm) (Fig. 2). The 7 bp deletion in Cmeil2 was verified from independent gDNA extractions (data not shown). At the nucleotide level CfEIL2 and Cmeil2 shared 96 % identity to each other and 76 % identity to CfEIL1 and CmEIL1 respectively (Table 1). PCR reactions specific for EIL2 using Cp gDNA or cDNA did not amplify a product. To address the transcriptional regulation of the ethylene signal transduction pathway in Campanula, flower tissues were harvested at five developmental stages (from bud to fully expanded flower on day 4, Fig. 3). Transcripts of putative ERS2, CTR1 and EIL1 were expressed constitutively during flower development (Fig. 4). To determine whether Campanula ERS2, CTR1 or EIL1 were responsive to ethylene, transcriptional analysis were performed in young (1-day old) flowers exposed to 0.025 μL · L −1 or 0.050 μL · L −1 ethylene for 24 h. However, neither ERS2, CTR1 nor EIL1 transcripts were responsive to the applied ethylene treatments (Fig. 5). Expression patterns of CfEIL2 and Cmeil2 transcripts were analyzed by classic RT-PCR and RT-qPCR. CfEIL2 transcripts showed expression pattern and levels similar to those of CfEIL1 throughout flower development and in response to 0.025 μL · L −1 and 0.050 μL · L −1 ethylene for 24 h (Fig. 6). In contrast, Cmeil2 was detectable in trace amounts when analysed by RT-PCR. Quantitative analysis by RT-qPCR showed consistently very low levels of Cmeil2 through flower development and no transcriptional response to ethylene. Expression levels of CfEIL2 and Cmeil2 differed by more than 100-fold in young flowers (day 0, day 1) whereas the same comparison in old flowers (day 4) yielded only 40-fold changes. The variation in fold change was primarily due to nonsignificant increases in Cmeil2 transcript levels. Expression levels of CfEIL2 and Cmeil2 in response to ethylene were not found to be significantly different due to the large variation in CfEIL2 expression levels (Fig. 6b).
The eil2 frameshift mutation is unique for C. medium Alignment of the putative EIL2 protein fragment from Campanula with EIL protein sequences from other Fig. 7 Alignment of partial sequences from translated Campanula EIL proteins spanning 196 amino acids. Conserved domains previously described in EIN/EIL proteins are boxed, these are the basic domains (BDI-BDIII) and the proline-rich domain (PR). The conserved SALM motif in which Cmeil2 is mutated is marked with (· · · ·). Conserved aa among all EIN/EILs are marked below with an asterisk. Cmeil2 is presented in bold. The alignment were produced from partial EIN/EIL protein sequences of C. portenschlagiana (Cp), C. formanekiana (Cf), C. medium (Cm), Actinidia deliciosa (Ad), Arabidopsis thaliana (At), Cucumis sativus (Cs), Nicotiana tabacum (Nt), Solanum lycopersicum (Sl) and Vitis vinifera (Vv). Previously named proteins are presented by their species abbreviation followed by their name. The alignment was produced in Clustal Ω [62] plants confirmed the presence of three conserved domains found in other EILs. These domains comprise the two basic amino acids binding domains (BD I and BD II) and the proline-rich region (PR) (Fig. 7). At the protein level the putative EILs from Campanula were closely related to each other when compared to other EILs except for Cmeil2. Cmeil2 showed high homology to other EILs until the position of the frameshift. The sequence following downstream of the frameshift was only observed in Cm and did not show any homology to previously reported EIL proteins. Omission of the deletion from the Cmeil2 reading frame resulted in a protein that perfectly aligned with other EIL2 proteins (data not shown). Phylogenetic analysis using other plant EILs indicated a close relation among Campanula EILs and a clear phylogenetic separation of Campanula EIL1 and EIL2 proteins (Fig. 8). Some branch points in the phylogenetic analysis yielded low bootstrap values due to the size of the aligned fragment (200 amino acids) and the high identity among all the EILs (Fig. 8).

Characterization of eil2 in Campanula species and cultivars
To elucidate the natural occurrence of eil2 in Campanula, close relatives to C. medium were identified as C. hofmannii, C. alpina, C. alpestris and Edraianthus graminifolius and C. incurva as close relative to C. formanekiana [32]. As the Cmeil2 mutation disrupts an NlaIII restriction site in EIL2 a simple screen for the presence of the mutation was developed (Fig. 9). EIL2 PCR products digested with NlaIII resulted in either two or three DNA fragments depending on the presence or lack of the eil2 mutation, respectively. Results obtained via NlaIII digests were verified by sequencing. Interestingly, eil2 was found to be specific for Cm and did not occur in related Campanula species (Fig. 9a). Intraspecific NlaIII restriction analysis among Cm cultivars confirmed the occurrence of eil2 regardless of cultivar origin (Fig. 9b). Thus the reported frameshift mutation in Cmeil2 is specific for Cm and occurs in all tested Cm both among non domesticated and domesticated cultivars.

Characterization of eil2 in hybrids of Cf × Cm
As Cm is homozygote for eil2 whereas Cf is homozygote for EIL2 the performance of eil2 in heterozygote plants were evaluated by ethylene exposure tests in C. formanekiana × C. medium hybrids. The presence of eil2 in Cf × Cm hybrids (A-E) was verified using the NlaIII screening system (Fig. 10). Young flowers were exposed to high ethylene concentrations of 5.0 μl · L −1 for 72 h and flower responses were scored in categories of no response, signs of senescence and complete senescence (Table 2)

Ethylene sensitivity in Campanula
Ethylene sensitivity of flowers is a recurring problem affecting breeders, producers and costumers of ornamental Solanum lycopersicum (Sl) and Vitis vinifera (Vv). Previously named proteins are presented by their species abbreviation followed by their name. The phylogenetic analysis were produced from MEGA version 6 [63] plants [17]. Thus characterisation of physiological and molecular variations in economically important Campanula species are much needed. In the present study, ethylene sensitivity among C. portenschlagiana, C. formanekiana and C. medium were found to depend on physiological and genetic factors. Ethylene sensitivity in flowers were dependent on genotype as Cp was highly sensitive, Cf had an intermediate level of sensitivity and Cm was found to be insensitive to high concentrations of ethylene (Fig. 1, 10 heterozygote Cf × Cm hybrids shows that all hybrids contain both Cf and Cm specific DNA fragments (c-g, hybrids A-E). Hybrids were produced according to [57]   sensitive to low amounts of 0.05 μL · L −1 ethylene, whereas all flowers of Cp were sensitive to 0.1 μL · L −1 ethylene. Doubling the ethylene concentration resulted insignificant increase in senesced Cf flowers from 26 to 100 % and also in complete loss of flower longevity in young Cf flowers (Fig. 1). In the same experimental settings all flowers of Cm were insensitive to ethylene. A similar correlation between flower age and ethylene sensitivity has been observed in Pelargonium peltatum [33]. This age dependent increase in sensitivity may also be connected to pollination as some plant species induce flower senescence upon pollination [34][35][36].
Whereas Cf flowers are ethylene sensitive in a concentration dependent manner (Fig. 1e, Table 2) not even a 50-fold increase in ethylene concentration reduced flower longevity in Cm. This indicates that ethylene insensitivity of Cm is independent of ethylene concentrations (Fig. 1f, Table 2).

Constitutive expression of ERS2 and CTR1 in flowers
Expression of ERS2 was found to be constitutive in Campanula during floral development and transcripts in young flowers were also unresponsive to low concentrations of ethylene (Figs. 4b and 5b). In both Arabidopsis and roses ethylene receptors are encoded by five genes [6,[37][38][39], some of which exhibit differential expression in response to exogenous ethylene and are regulated during flower development tissues [40,41]. Also, exogenously applied ethylene does not affect levels of Dianthus caryophyllus ERS2 in petals but this gene is regulated by flower development [42]. Collectively, results obtained in other plants indicate that ethylene receptor families comprise multiple members. The genome of Cp was found to contain two homologs of ERS receptors whereas only one gene/transcript was identified in Cf and Cm. As gDNA was also used as template in the cloning reactions mRNA levels in flower tissues should not be the determining factor. Thus all three Campanula species likely encode additional ethylene receptors which were not identified here due to primer specificities.
Similarly as for ERS2, all Campanula CTR1 transcripts were constitutively expressed during flower development and did not respond to application of exogenous ethylene (Figs. 4c and 5c). The same pattern was observed in roses where RhCTR1 and RhCTR2 were constitutively expressed throughout flower development; however both RhCTR transcript levels increased in response to ethylene [43]. In some plant species the genomic structure of CTR1 is highly complex. Banana and tomato CTR1 exist in a 15 exon 14 intron structure yielding complete ORFs of 11.5 and 12 kb, respectively [44,45]. Even with a long elongation time in PCR reactions we were not able to identify the full-length sequence of Campanula CTR1. However small polymorphisms detected in the partial Cp, Cf and Cm CTR1 transcripts indicate that these plants may be heterozygote in the CTR1 locus or that an additional copy of CTR1 exists. Two and four CTR homologs have been identified in roses and tomato, respectively [43,46]. Hence an additional CTR1 in Campanula is not unlikely.

Occurrence and expression of EILs in Campanula
The EIN3/EIL family encodes transcription factors mediating the initialization of the physiological ethylene response [13,47]. In the three Campanula species investigated here approximately 600 bp of a flower expressed EIL homologue were identified, EIL1. All EIL1 transcripts were consistently expressed through flower development and did not respond to applied ethylene (Figs. 4d and 5d). In the small flowered Cp two close homologs of EIL1 were identified by sequencing as EIL1a and EIL1b, however the two could not be separated during expression analysis. In contrast, the two large flowered Campanula (Cf and Cm) both contained the ORF of EIL2, a gene not detected in Cp. CfEIL2 gene expression was like CfEIL1 constitutive throughout flower development and nonresponsive to exogenous ethylene (Fig. 6c, d). In Cm, eil2 was not expressed due to the deletion of 7 bp in the ORF (Figs. 2 and 6). Translational analysis of the partial Cmeil2 protein indicated that the 7 bp deletion would introduce a frameshift prior to what should have been the proline rich domain in Cmeil2 (Fig. 7). On nucleotide level CfEIL2 and Cmeil2 share 96 % identity ( Table 1), showing that they are close homologs. The close homology among CfEIL2 and Cmeil2 was supported by phylogenetic analysis (Fig. 8). Translation of the putative Cmeil2 indicate that the frameshift mutation in eil2 disrupts the putative major DNA binding domain in EIL2 in front of the proline rich region and simultaneously introduces a stop codon 50 amino acids further downstream in the eil2 protein sequence. Hence, if translated the protein would be truncated to approximately 40 % of the expected size when compared to AtEIN3 (Fig. 7). The position and functionality of EIL DNA binding domains have been characterised in Arabidopsis EIL3 and in cucumber EIN3 [48,49].
Whether the promoters of CfEIL2 and Cmeil2 share the same specificity remains to be shown, however Cmeil2 may have been expressed in flowers at one point as traces of the transcript was observed in RT-PCR and in RT-qPCR (Fig. 6). Finally, additional EIL homologs may be present in Campanula as 4-6 homologs have been identified in Arabidopsis, tomato and tobacco [13,[50][51][52].
Previous studies of EIN3/EIL homologs have shown constitutive expression in Paeonia, however in Dianthus caryopyllus DcEIL1/2 and DcEIL3 transcripts in petals and styles increased rapidly after ethylene treatment of flowers and then gradually declined [53,54]. The constitutive expressions of ERS2, CTR1 and EILs in the ethylene signal transduction pathway in Campanula indicate that flowers are capable of a fast physiological response in the presence of ethylene. This correlate well with earlier results where Campanula has been described as an ethylene sensitive species [20]. Furthermore, the lack of transcriptional response to ethylene exposure could be a combination of regulatory steps on the protein level [55,56].
The Cmeil2 phenotype was inherited as a recessive trait In the present study, the eil2 frameshift mutation was identified not only in Danish domesticated Cm but also in non domesticated specimens of Cm and in the closely related C. medium var. calycanthema (Fig. 9). This suggests that the eil2 frameshift must have occurred in an earlier ancestor of Cm. None of the closely related Campanula species C. alpina, C. alpestris, C. hofmannii or E. graminifolius contained eil2 (Fig. 9). In hybrids of Cf × Cm, heterozygote eil2 resulted in very similar ethylene responses as observed in Cf as only one of five hybrids showed an intermediate phenotype shifted towards increased ethylene insensitivity when compared to Cf (Table 2). This indicated that the eil2 phenotype was restored by the presence of a wild type EIL2 and was inherited as a recessive trait. Homozygote eil2 in Cf could not be obtained via crossings as both male and female parts of Cf × Cm hybrids were sterile. Results obtained here are the first to describe the effects of an ein3/eil mutation in flowers. Also, no reports exist of approaches where EIN3/EIL genes have been knocked out or silenced via gene modification. Therefore the eil2 phenotype described in Cm cannot be directly compared to related phenotypes in flowers of other plant species. In Arabidopsis, the closest homolog to Campanula EIL2 is EIN3 (AtEIN3). In Arabidopsis, ein3 mutants are well characterised and they too are inherited in a recessive manner [13]. Also phenotypes in ein3 or eil mutants in Arabidopsis are only described in seedlings or mature rosettes. To our knowledge ethylene sensitivity and floral development has not been described in Arabidopsis ein3 or eil mutants. However, in support of our results are data from tomato where expression of LeEIL1, LeEIL2 or LeEIL3 antisense transcripts results in ethylene insensitive buds [51]. Collectively, this study is the first to correlate ethylene insensitivity in flowers to an ein/eil mutant phenotype.

Future approaches to achieve ethylene insensitive plants
Previous studies have indicated that there may be specific functions for the individual EILs. Thus to alter the physiological response of flowers to ethylene, the right ortholog in each plant species has to be identified. In the framework of this study, the Cmeil2 frameshift mutation could potentially be transferred to EIL2 homologs of related species by conventional crossing or the deletion could be introduced by wide hybridisation among related species, assisted by embryo rescue techniques when necessary [57].
The frameshift in Cmeil2 is positioned in a highly conserved region, and it therefore holds potential for molecular breeding towards ethylene insensitive plants. However, the identification of the full genomic sequence in Cm and Cf and the full sequence of the translated gene product in Cf are essential steps in this process. Ultimately, we propose that the identified 7 bp deletion in Cmeil2 may be used to confer ethylene insensitivity to other plant species. This may be feasible via targeted mutagenesis techniques utilizing ZNF, TALENs [58] or CRISPR/Cas9 [59]. As a result, ethylene insensitivity may be transferred from Cm to other important climacteric ornamentals e.g., roses, Petunia, carnations, or even to edible climacteric crops such as broccoli and tomato.

Conclusions
We characterised the physiological and molecular responses among three Campanula species to exogenous ethylene. Key genes in the ethylene signal transduction pathway ERS, CTR and EIL1 were found to be constitutively expressed in Campanula and unresponsive to exogenous ethylene. However, EIL2 was found to be specific for the large flowered species C. formanekiana and C. medium, but was not expressed in Cm due to a 7 nucleotide frameshift in the coding region of Cmeil2. The natural mutation identified here in Cmeil2 correlates with the observed ethylene insensitivity in this species. This finding holds great potential for future breeding strategies towards ethylene insensitive plants.

Identification of putative ERS2, CTR1 and EIL genes
Based on the NCBI GenBank [60] sequence from Campanula carpatica (GenBank: AF413669) intron spanning primers were designed to amplify partial fragments of ERS2. ERS2 PCR on gDNA produced two ERS2 products in Cp and one ERS2 genomic fragment in Cf and Cm. In Cp, The two CpERS2 products were derived from two genes containing different intron sizes but coding for very similar transcripts (CpERS2a and CpERS2b, Additional file 3). A partial CTR1 was produced using degenerate primers aligning to conserved areas among Musa acuminata, Solanum lycopersicon, Arabidopsis thaliana and Rosa hybrida CTR1 sequences (Gen-Bank:JF430422, GenBank:AF096250, GenBank:NM_1804 29, and GenBank:AY032953). Sequencing showed some polymorphisms in CTR1 of Cp, Cf and Cm. This could indicate that more than one copy of CTR1 exist in Campanula. Partial EIL genes cloned via degenerate primers produced from Malus x domestica and Solanum lycopersicon EIL sequences (GenBank:GU732486 and Gen-Bank:NM_001247617). Campanula ACT was amplified using degenerate primers, sequenced and from this sequence specific primers were designed for expression analyses. Primers and PCR product sizes are presented in Additional file 3.
Genomic DNA from Cp, Cf, and Cm was isolated from flowers with DNeasy Plant Mini Kit (Qiagen) using 300 mg of plant material following manufacturer's recommendations. PCR reactions used 100-250 ng gDNA, 2 % (v/v) DMSO and polymerase LaTaq (Takara Bio Inc.) as manufacturer recommends. Reactions followed the program; 4 min 94°C, 33-35 cycles of [30 s 94°C, 1 min 60°C, 1 min 72°C] and a final 7 min elongation step at 72°C in a MyCycler (Biorad). Cloning of PCR-products was via TOPO TA Cloning® kit (Life Technologies Corp, Invitrogen) as recommended by manufacturer. Plasmids were purified by QIAprep Spin Miniprep kit (Qiagen) and sequenced by Eurofins MWG Operon. EIL2 PCR products were purified with QIAquick PCR purification Kit (Qiagen) and restriction analyses using NlaIII were done as supplier recommends (New England Biolabs).

Ethylene exposure experiments
To monitor flower development, individual buds were labelled one day before flower opening. This stage was termed day 0. In the following days newly opened flowers (day 1) and 4 days old flowers (day 4) were identified, tagged and used in subsequent experiments. This allowed two morphologically different stages to be monitored simultaneously throughout the ethylene exposure experiments. Ethylene exposure were conducted in a climate chamber in glass tanks with postharvest growth conditions; 20°C day/18°C night, 16-h photoperiod at 10-12 μmol m −2 · s −1 provided by cool-white fluorescent tubes (Philips Master TL-D-36 W/830). Each glass tank had a volume of 128 L and contained three plants. Flower labeling resulted in each glass tank containing three plants with a total of 15 labelled flowers for each developmental stage (day 1 and day 4). Except for Cm where 8-13 labeled flowers pr. growth stage were used. Ethylene concentrations of 0 μL · L −1 , 0.05 μL · L −1 or 0.1 μL · L −1 were obtained by injection of gaseous ethylene (Mikrolab Aarhus A/S) into sealed glass tanks. Flowers were monitored, tanks ventilated and ethylene reinjected every 24 h. For Fig. 1 a senescent flower was defined as a flower showing twisted or closed corolla or wilted. For Table 2, ethylene sensitivity of Cf × Cm hybrid flowers were classified in three categories; no symptoms, signs of senescence (partial wilting and discoloration of corolla) and complete senescence (complete wilting and full discoloration of corolla). The latter experiments were done in glass tanks with 5 μl · L −1 ethylene for 72 h. Cp and Cf experiments were repeated twice whereas Cm experiments were in three replicates.
Gene expression analyses were done at developmental stages: bud, day 0 (one day before flowering), day 1, day 2 and day 4 (Fig. 3). Plants were subjected to low ethylene concentrations of 0.025 μL · L −1 and 0.050 μL · L −1 ethylene for 24 h. Each tank contained three plants with labeled 1-day flowers. Flowers from each tank were pooled, harvested in liquid nitrogen, grinded and used for RNA extraction. Each experiment was performed in three replicates.
RNA extraction, cDNA synthesis and expression analysis RNA was extracted using RNeasy Plant Mini kit (Qiagen) with the following change to manufacturer's protocol: Cell lysis were done using RLT buffer with 0.01 % βmercaptoethanol (v/v) for 1 min at 56°C. RNA yield and purity (A260/A280 ratio > 2.0) was estimated by a Nanodrop TM 1000 Spectrophotometer (Thermo Fisher Scientific Inc.). RNA integrity was evaluated on 1.2 % agarose gels. Purified RNA was stored at -80°C. RNA was DNase treated with Amplification Grade DNase I (Invitrogen) and cDNA synthesis was done using iScript cDNA Synthesis kit as recommended (Biorad). In 20 μl reactions 0.8 μg RNA was used. No contamination of DNA in cDNA was verified in non RT samples. Expression analyses were done using 5-fold diluted cDNA and ExTaq as polymerase as recommended (Takara Bio Inc.) in MyCycler (Biorad). Primers and gene specific reaction settings used in expression analysis are presented in Additional file 4. RT-PCR program; 4 min 94°C, 25-32 cycles of [30 s 94°C, 1 min 55°C, 1 min 72°C] and 7 min 72°C. Reactions for RT-qPCR were performed on an ICycler instrument (Bio-Rad) by using the iQ SYBR Green Supermix (Bio-Rad) according to supplier's instructions [61] using the program; 95°C for 10 min, 50 cycles of [30 s at 95°C, 1 min at 57.5°C and 1 min at 72°C]. To evaluate the efficiency of qPCR, serial dilutions of cDNA were used to generate a standard curve. This resulted in R 2 values of 0.998 and 0.994 for ACT and EIL2 primer sets, respectively. Threshold cycles (Ct), (defined as cycle were the signal exceeds ten times the standard deviation of the baseline), for CfEIL2 and Cmeil2 were standardized to the corresponding Actin Ct (ΔCt). The relative quantification of target gene CfEIL2 and Cmeil2 between the different treatments was determined as 2^(−ΔΔCt). Values are based on three replicates.

Bioinformatics and statistics
Sequence identification and analysis were done using CLC sequence viewer (CLC bio), BLAST and Clustal Ω [62]. Phylogenetic analysis were conducted using MEGA version 6.06 [63]. Statistical analyses were done in Sig-maPlot v. 13 by one way analysis of variance using the Holm-Sidak method.