Functional conservation and divergence of ve AP1/FUL-like genes in Marigold (Tagetes erecta)


 Background: Members of AP1/FUL subfamily genes play an essential role in the regulation of floral meristem transition, floral organ identity, and fruit ripping. At present, there have been insufficient studies to explain the function of the AP1/FUL-like subfamily genes in Asteraceae. Results: Here, we cloned two euAP1 clade genes TeAP1-1 and TeAP1-2, and three euFUL clade genes TeFUL1, TeFUL2, and TeFUL3 from marigold (Tagetes erecta). Expression profile analysis demonstrated that TeAP1-1 and TeAP1-2 were mainly expressed in receptacles, sepals, petals, and ovules. TeFUL1 and TeFUL3 were expressed in floral buds, stems and leaves as well as in productive tissues, while TeFUL2 was mainly expressed in floral buds and vegetative tissues. Transgenic Arabidopsis lines showed that overexpression TeAP1-2 or TeFUL2 resulted in early flowering, implying that these two genes might regulate the floral transition. Yeast two-hybrid analysis indicated that TeAP1/FUL proteins only interacted with TeSEP proteins to form heterodimers, and that TeFUL2 could also form a homodimer.Conclusion: In general, TeAP1-1 and TeAP1-2 might play a conserved role in regulating sepal and petal identity, just like the role of MADS-box class A genes, while TeFUL genes might display divergent functions. This study provides an insight into molecular mechanism of AP1/FUL-like genes in Asteraceae species.

owering time and in orescence structure [38,39]. However, overexpression of GSQUA2 (a FUL-like gene, homology to DEFH28) results in dwarf plant, early owering, and vegetative abnormality, but it does not affect in orescence structure [37].
Marigold (Tagetes erecta) is a popular ornamental plant and industrial crop, in whose ower lutein is rich. As a member of Asteraceae, marigold also has a typical capitulum. Compared to that of Chrysanthemum, the whole life cycle of marigold lasts only 2-3 months from sowing to owering. In addition, in the evolutionary history of Asteraceae family, marigold undergoes a long evolution process and it is located in a derived Calenduleae clade [42]. These characteristics make marigold a valuable material in the study of function of AP1/FUL genes. In this study, we cloned and characterized ve AP1/FUL-like genes in marigold, whose distinct expression patterns, protein interaction manner and different phenotypes of Arabidopsis transgenic lines might imply their divergent functions in regulating the oral meristem development, oral organ identity, and owering time.

Results
Isolation and phylogenetic analysis of TeAP1/FUL-like genes The full-length sequences of ve AP1/FUL-like genes were ampli ed by using gene-speci c primers with cDNA generated from buds as a template. In order to further identify the putative homologues of AP1 and FUL genes, we blasted nucleotide sequences of these ve genes against NCBI. The blast search results indicated that two different AP1-like genes and three different FUL-like genes were detected with the former designated as TeAP1-1 (Acc. No. MT394170), TeAP1-2 (Acc. No. MT394171), and the latter designated as TeFUL1 (Acc. No. MT394172), TeFUL2 (Acc. No. MT394173), and TeFUL3 (Acc. No. MT394174), respectively. Sequence analysis revealed that the putative proteins based on these ve genes coding sequences were composed of 246, 247, 235, 235, and 242 amino acids, respectively. The two putative AP1 proteins shared over 89% amino-acid identity, while the three putative FUL homologous proteins shared relatively low identity (Supplementary Table S3). Multiple sequence alignment and conserved analysis of AP1/FUL proteins indicated that all TeAP1/FUL proteins contained a conserved MADS domain, a I domain, a less conserved K domain, and a variable C-terminal domain (Fig. 1). The putative proteins of TeAP1-1 and TeAP1-2 possessed a typical euAP1-motif (CFPS) containing both an acidic domain and a farnesylation motif (CaaX, shown at their C termini).
In addition, characteristic FUL motif was shared by the three TeFUL proteins (Fig. 1).
To investigate the relationship between TeAP1/FUL genes and other members of AP1 and FUL clades, a phylogenetic analysis was carried out by using amino acid sequences of AP1/FUL clade from other plant species and O. sativa OsMADS1 belonging to SEP subfamily as an outgroup (Fig. 2). TeAP1-1 and TeAP1-2 were co-orthologous to Arabidopsis AP1 and Antirrhinum SQUA. TeFUL1 and TeFUL3 were phylogenetically close to euFUL, and TeFUL2 was co-orthologs of the Antirrhinum protein DEFH28. Remarkably, TeFUL1 and HaFUL (Helianthus annuus) shared 78.39% amino-acid identity, and TeAP1-1 and HAM75 (H. annuus) were more closely related to each other with over 97.45% amino-acid identity. The high homology might indicate the functional similarity between them.

Expression analysis of TeAP1/FUL-like genes in marigold
The expression patterns for the ve AP1/FUL genes in different vegetable tissues, oral organs, and different development phases of oral buds were examined by qRT-PCR. TeAP1-1 was mainly expressed in leaves, receptacles, bracts, sepals of ray orets, petals of disk orets, and ovules, but not expressed in oral buds and roots (Fig. 3, S1). Compared to TeAP1-1, TeAP1-2 was weakly expressed in different development phases of oral buds, but was highly expressed in receptacles, sepals of ray and disk orets, petals of disk orets, and ovules (Fig. 3, S1). TeFUL1 and TeFUL3 shared a similar expression pattern, and they were widely expressed in vegetative and productive tissues (Fig. 3, S1). The difference in expression was detected between TeFUL1 and TeFUL3. For example, TeFUL1 was expressed mainly in petals of disk orets, stamens, ovules and sepals of ray orets, while TeFUL3 was highly expressed in all oral organs of two-type orets and receptacles (Fig. 3, S1). Contrary to TeFUL1 and TeFUL3, TeFUL2 was highly expressed in vegetative tissues, ower buds, receptacles, and bracts, and weakly expressed in oral organs (Fig. 3, S1).

Interactions between TeAP1/FUL proteins and other MADS-box proteins in marigold
In order to investigate whether difference in expression pattern of these ve TeAP1/FUL genes caused their difference in protein interactions, a yeast twohybrid analysis was performed. The ability of pairwise interaction between TeAP1/FUL and other MADS-box proteins in marigold was determined by yeast two hybrid method. The marigold proteins were individually fused to the binding domain and the activation domain, and were combined in both directions. No autoactivation was observed among these ve proteins ( Supplementary Fig. S2a). As shown in Table 1 and Supplementary Fig. S1b-e, TeAP1/FUL proteins only interacted with SEPATELLA (SEP) proteins, but they did not interact with class B (TePI, TeAP3-1, TeAP3-2, TeTM6-1 and TeTM6-2), class C (TeAG1 and TeAG2), and class D (TeAGL11-1 and TeAGL11-2) proteins. TeAP1-1 and TeAP1-2 exhibited a same protein interaction manner, both of which interacted with class E proteins TeSEP3-2 and TeSEP3-3 to form heterodimers (Table 1, Supplementary Fig. S2b, e). Contrary to TeAP1 proteins, TeFUL proteins displayed a different protein interaction manner. TeFUL1 only interacted with TeAGL6. TeFUL2 interacted with TeSEP3-2, TeAGL6, and itself. TeFUL3 exhibited a variety of protein interaction manners, which could form heterodimers with TeSEP1, TeSEP3-1, TeSEP3-2, TeSEP-3, and TeSEP4 (Table 1, Supplementary Fig. S2b, e). Table 1 Interactions of Marigold AP1/FUL proteins detected by yeast two-hybrid assays.

Expression Analysis Of Endogenous Genes In Transgenic Plants
To reveal the mechanism underlying phenotypic changes of transgenic lines 35S:TeAP1-2 and 35S:TeFUL2, the expression levels of endogenous genes regulated by AP1 were analyzed when the T3 transgenic and wild-type seedlings were 10 days old. As shown in Fig. 5, TeAP1-2 and TeFUL2 displayed a similar function in regulating the expression level of AP1 downstream genes. For example, the expressions of AP1, FT, LFY, SOC1, SPE3, and TFL in 35S:TeAP1-2 and 35S:TeFUL2 transgenic lines were obviously higher than those in wild-type plants. The expression level of AGL24 showed no remarkable changes in both 35S:AP1-2 and 35S:FUL2 transgenic lines. However, it should be noted that genes TeAP1-2 and TeFUL2 speci cally regulated some downstream genes. For instance, SPL9 was signi cantly upregulated in 35S:TeAP1-2 transgenic lines, but it exhibited no change in 35S:FUL2 transgenic lines. In contrast to SPL9, SVP was signi cantly highly expressed in 35S:FUL2 transgenic lines, but no signi cant change in 35S:AP1-2 transgenic lines.

Discussion
The study of many AP1/FUL-like genes from various species has demonstrated that AP1/FUL genes play key roles in owering time, ower and fruit development. Like genes APETALLA3 (AP3, B class gene) and AGAMOUS (AG, C class gene), the AP1/FUL genes undergo several duplication events, resulting in the occurrence of euAP1 and euFUL clade in core eudicots [13,14]. In this study, ve AP1/FUL-like genes were obtained from marigold. Sequence alignment analysis indicated that all these 5 AP1/FUL-like proteins are typical MIKC proteins, and they contained conserved motif at their C terminal domain (Fig. 1).
TeAP1-1 and TeAP1-2 were clustered into euAP1 clade proteins harboring an acidic domain and a farnesylation motif (Fig. 1), and the TeFUL1, TeFUL2 and TeFUL3 possessed a conserved FUL motif ( Fig. 1) that was demonstrated to be all members of FUL clade proteins [13,14]. Such changes in amino acid sequence have been explained by a frameshift mutation in an ancestral AP1/FUL-like gene [13,43] and are responsible for gene-speci c functions.
Our phylogenetic analysis indicated that TeAP1-1 and TeAP1-2 were members of AP1 clade, and seemed to be homologous to antirrhinum SQUA which was previously reported to be involved in regulating the oral meristem development and specifying the sepal and petal identity [44]. TeFUL1, TeFUL2 and TeFUL3 were clustered into the FUL clade, and TeFUL1 and TeFUL3 proteins were closer to the euFUL group. TeFUL2 belonged to the euFULII group (Fig. 2). The TeFUL2 was orthologous to the antirrhinum gene DEFH28 which was also clustered into euFULII group, and this gene participated in regulation of oral meristem development, fruit development, and owering time [34]. Base on the expression pattern analysis, the TeFUL2 was mainly expressed at the early stage of in orescence development (Fig. 3, S1), which was similar to the expression pattern of the early function genes represented by Arabidopsis FUL [21] and petunia PFG [36], implying a role of TeFUL2 in meristem identity. However, TeFUL1 and TeFUL3 were expressed in vegetative tissues, different stages of oral buds, and oral organs (Fig. 3, S1). Based on these ndings, it could be speculated that TeFUL2 and TeFUL1(or TeFUL3) might arise due to gene duplication, and that this duplication event might cause the change in their expression patterns. Many previous studies reveal that functional divergence is caused by gene duplication which further drives evolution [10,45]. Therefore, we speculated that the duplication events and the modi cation of transcript pattern of TeFUL genes might imply the diversi cation of their functions in marigold.

Conserved function of AP1/FUL genes in owering performance
Function analysis of the AP/FUL-like genes in core eudicots and non-core eudicots revealed that AP1/FUL-like genes displayed conserved roles in regulating the owering time. For example, overexpression of AP1 or FUL in Arabidopsis both leads to early owering [25,46]. Furthermore, the similar phenomena were also observed in the case of ectopic overexpression of AP1-like or FUL-like genes from Asteraceae species, such as C. morifolium (CDM111) [39], C. lavandulifolium (ClM8) [38] and G. hybrida (GSUQA2 ) [37]. In this study, heterologous expression of TeAP1-2 and TeFUL2 into Arabidopsis resulted in early owering without affecting oral organ identity (Fig. 4d, f, h). Additionally, ectopic expression of TeAP1-2 also led to the curl of rosette leaf and cauline leaf (Fig. 4e, g), which was similar to the function of the AP1/FUL-like gene MBP20 [47]. The MADS-box transcription factors possess a DNA-binding domain to regulate their downstream gene expression [45]. Therefore, we speculated that the early owering phenotypes observed in 35S:TeAP1-2 and 35S:TeFUL2 transgenic lines might be related to the change in endogenous gene expression level. In this study, AP1, FT, LFY, SOC1, SPE3, and TFL were signi cantly upregulated in 10-day-old seedlings of transgenic lines containing 35S:TeAP1-2 and 35S:TeFUL fusion vectors (Fig. 5), suggesting TeAP1-2 and TeFUL2 might share overlapping regulation network of a series of downstream genes in Arabidopsis. In Arabidopsis, AP1 directly represses SVP, AGL24, and SOC1 to partially specify oral meristem identities [48]. However, in our study, no remarkable change in the expression level of AGL24 was observed in both 35S:TeAP1-2 and 35S:TeFUL2 transgenic lines (Fig. 5). Additionally, the expression level of the owering repressor gene SVP was signi cantly activated in 35S:TeFUL2 transgenic lines, but not in 35S:AP1-2 transgenic lines (Fig. 5). In contrast to SVP, SPL9 was signi cantly upregulated in 35S:TeAP1-2 transgenic lines, but not in 35S:FUL2 transgenic lines (Fig. 5). These results revealed that TeAP1-2 and TeFUL2 had a divergent function in regulating downstream genes, which was further supported by their difference in protein interaction manners (Table 1, Supplementary Fig. S2b-e).
Potential redundant function of TeAP1-1 and TeAP1-2 as class A genes In Arabidopsis, AP1 is an early-acting gene and functions as an class A gene to specify sepal and petal identity [20,49]. AP1 is expressed in oral meristems and developing sepal and petal primordia [20,21,24,50]. However, in other core eudicots, the AP1-like genes can be also expressed in bracts and reproductive organs [27,39,51]. Similarly, TeAP1-1 and TeAP1-2 were both highly expressed in sepals of two-type orets and petals of disk orets as well as in bracts, receptables, and ovules (Fig. 3, S1). In Arabidopsis, AP1 only interacted with SEP3 to form heterodimer. Furthermore, in Asteraceae species, the AP1-like proteins C. morifolium CDM111 [39,40], G. hybrida GSQUA1, and GSQUA3 [37] also had a limited protein interaction manner. In other words, they only formed heterodimers with SEP3 proteins. In this study, TeAP1-1 and TeAP1-2 shared a similar protein interaction pattern to form heterodimers with TeSEP3-2 and TeSEP3-3 (Table 1, Supplementary Fig. S2b, e), suggesting that euAP1-like proteins shared a conserved protein interaction manner. Taken together, TeAP1-1 and TeAP1-2 may play a redundant role as class A genes.

Divergent functions among TeFULs genes
The functions of FUL-like genes in the transition from vegetative meristems to reproductive meristems and in fruit development were well-known in many core eudicots and non-core eudicots. In model plant Arabidopsis, FUL regulates the cell differentiation during fruit development [31,33,52] and participates in specifying oral meristem identity with AP1 and CAL [21]. In basal eudicots, the Aquilegia coerulea FUL-like genes regulate leaf morphogenesis and in orescence development [8]. Additionally, in monocots, the Oryza sativa homologues genes OsMADS14 and OsMADS15 are involved in specifying the meristem identity, palea and lodicule identity [7]. In contrast to the AP1-like genes, the FUL-like genes are widely expressed in vegetative and productive tissues [6,31,46].
In our study, TeFUL1 and TeFUL3 were expressed in stems and leaves as well as in productive tissues (Fig. 3, S1), which was consistent with the typical FULlike expression pattern [6,31,46], implying that TeFUL1 and TeFUL3 might play a role as FUL genes. Furthermore, ectopic expression of TeFUL1 or TeFUL3 into Arabidopsis led to no visible phenotype changes. In Arabidopsis, FUL functions redundantly with CAL and AP1 to specify the oral meristem identity, and single ful mutation has no ability to affect oral organ identity [21]. In general, we speculated that TeFUL1 and TeFUL3 might function redundantly in regulating the oral meristem identity, or that TeFUL1 and TeFUL3 need to work together with AP1-like genes to regulate the oral meristem development. However, the striking difference in protein interaction manner was observed between TeFUL1 and TeFUL3 ( Table 1, Supplementary Fig. S2b-e). TeFUL1 only interacted with TeAGL6, while TeFUL3 interacted with TeSEP1, TeSEP3-1, TeSEP3-2, TeSEP3-3, and TeSEP4 to form heterodimers (Table 1, Fig. S2b-e).
Different protein interaction patterns might be related to their different conserved regions at C domains (Fig. 1). The above results suggested that TeFUL1 and TeFUL3 might be partially functionally redundant, but they might have their own speci c functions in regulating oral organ identity.
In contrast to TeFUL1 and TeFUL3, TeFUL2 was highly expressed in oral buds and vegetative tissues, and weakly expressed or unexpressed in oral organs and ovules (Fig. 3, S1). Additionally, TeFUL2 could form homodimer, and heterodimers with TeAGL6 and TeSEP3-2 ( Table 1, Supplementary Fig. S2b, e). Ectopic expression of TeFUL2 into Arabidopsis also led to early owering with less number rosette leaves (Fig. 4h), which was consistent with phenotype of the overexpressed euFULII (DEFH28) clade genes from core eudicots and non-core eudicots [28,37]. The above results suggested that FUL1 and FUL3 might lose some functions, but these functions might have been retained in FUL2. Overexpression of Antirrhinum DEFH28 (euFULII clade genes) into Arabidopsis resulted in early owering, two to four carpel formation, and failure to silique dehiscence [28]. However, ectopic expression of TeFUL2 into Arabidopsis did not affect oral organ identity and silique dehiscence (Fig. 4), which was in line with the study results of Gerbera GSQUA2. In general, TeFUL2 might retain a conserved role in regulating the meristem transition rather than fruit ripping.

Conclusion
In conclusion, marigold has ve AP1/FUL-like genes, two of which are clustered into euAP1 clade, and three of which FUL-like clade. Based on the results of expression pattern and protein interaction manner, TeAP1-1 and TeAP1-2 are likely to play a redundant role as class A genes in regulating sepal and petal identity. Additionally, ectopic expression of TeAP1-2 resulted in early owering, implying that TeAP1-2 might be involved in the regulation of meristem transition. However, three FUL-like genes display divergent functions. TeFUL1 and TeFUL3 are more functionally close to euFUL genes, whereas TeFUL2 is more functionally close to antirrhinum DEFH28 belonging to euFULII gene. Our results will provide a theoretical basis for the study of class A genes in Asteraceae. Considering the great difference in orescence structure between marigold and Arabidopsis, the function analysis of these ve AP1/FUL-like genes has been insu cient. Therefore, we need to further explore the potential functions of these ve genes by transforming them into marigold.

Plant materials and growth conditions
Marigold (T.erecta, M525B-1) is an inbred line with more than 10 generations of continuous self-crossing, and this marigold has only one whorl of ray orets outside capitulum [53]. The plants were grown in the experimental eld of Huazhong Agricultural University (lat. 30°28'36.5" N, long, 114°21'59.4" E) under natural conditions. To investigate AP1/FUL-like genes expression patterns, the samples of vegetative tissues, different sizes of ower buds, and oral organs in the blooming period were collected as described by Ai et al [54], and were frozen immediately in liquid nitrogen and stored at -80 ℃.  [54] were named TeAP1-1, TeAP1-2, TeFUL1, TeFUL2, and TeFUL3, respectively. To verify the accuracy of these ve gene sequences, the speci c primers of TeAP1-1-full-F/R, TeAP1-2-full-F/R, TeFUL1-full-F/R, TeFUL2-full-F/R, and TeFUL3-full-F/R were designed in the 3' and 5' terminal region by Primer Premier 5 (Supplementary Table S1) and used to clone full length of gene sequences. The PCR ampli ed fragments were puri ed and then cloned into pMD18-T vector (Takara, Dalian, China). Positive clones were veri ed by PCR using M13-F/R universal primers, and 3-5 of them were selected and sequenced in the Sangon company in Shanghai.
The Open Reading Frame (ORF) of these ve AP1/FUL-like genes were predicted on line (https://www.ncbi.nlm.nih.gov/or nder/) and were blasted against the NCBI to search for homologous sequences. To identify the conserved motifs of AP1/FUL amino acids, the multiple sequence alignment was performed by using the DNAMAN (v.6.0) software (https://www.lynnon.com) and BoxShade (https://embnet.vital-it.ch/software/BOX_form.html). A total of 38 AP1/FUL-like genes and a SEP-like gene OsMADS1 derived from Oryza sativa were downloaded from the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov) for phylogenetic analysis. Gene accession number was listed in supplementary Table S2. The O. sativa OsMADS1 belonging to SEP subfamily was used as outgroup. The construction of phylogenetic tree was based on amino acid alignment using the default settings of MUSCLE in MEGA (v. 7.0). A phylogenetic tree was constructed by the neighbor-joining (NJ) method with a bootstrap con dence values of 1000 replicates, and distances were calculated with Poisson corrections for multiple substitutions.

Gene expression analysis
The analysis of expression patterns of AP1/FUL-like genes in different tissues and different oral bud development phases was performed by quantitative real-time PCR (qRT-PCR). The speci c primers were designed within the non-conservative C-terminal region by the Primer Premier 5.0 and were listed in Supplementary Table S1. The speci c and unique PCR products for each primer pair were con rmed by 1.2% agarose gel electrophoresis. QRT-PCR was carried out in an optical 384-well plate in the QuantStudio 6 Flex real-time PCR system (Applied Biosystems) with SYBR Primix Ex Taq kit (TaKaRa, Dalian, China) according to manufacturer's instructions. QRT-PCR data were analyzed in the ABI 7500 Detection System (Applied Biosystems, USA). The qRT-PCR products were ampli ed in 10 μl reaction solution containing 1 μl template of the reaction mixture, 5 μl 2 × SYBR Green Master Mix, 0.2 μl forward primer and reverse primer (10 μmol/μl for primers), and double-distilled water to supply nal volume of 10 μl. The PCR was performed as follows: 95°C for 2 min and 40 cycles of 95°C for 10 s and 60°C for 20 s. The expression level of each gene was summarized from three replicates for each sample. The house-keeping gene beta-actin was used as an internal control for qRT-PCR and the relative expression levels were calculated using the 2 -ΔΔCt method.
Yeast two-hybrid assay The full-length coding sequences of TeAP1-1, TeAP1-2, TeFUL1, TeFUL2, and TeFUL3 were ampli ed using primers with speci c restriction sites and cloned into the activation domain plasmid pGBKT7 (Clontech, Palo Alto, CA, USA) and binding domain plasmid pGADT7 (Clontech, Palo Alto, CA, USA), respectively. All constructs were con rmed by sequencing analyses. The primers were supplied in Supplementary Table S1. The bait and prey constructs of ve class B genes (TeAP3-1, TeAP3-2, TePI, TeTM6-1, and TeTM6-2 ) and class C+D genes (C: TeAG1 and TeAG2, D: TeAGL11-1 and TeAGL11-2) were previously described by Ai et al [54] and Zhang et al [55], respectively. The full-length sequences, six class E genes (TeSEP1, TeSEP3-1, TeSEP3-2, TeSEP3-3, TeSEP4 and TeAGL6) were downloaded from NCBI, and bait and prey recombinants of their proteins were also constructed, respectively. Both bait and prey constructs were transformed into yeast cell strain AH109 using LiAc method (Clontech) following the Frozen-EZ Yeast Transformation II Kit protocols (Zymo Research Corp, Irvine, CA, USA). Interaction results between bait proteins and empty AD, prey proteins and empty BD, and empty BD and empty AD were used as negative controls. The interaction products between pGBKT7-53 and pGADT7-T vectors were used as positive control. Yeast double transformants were plated on SD medium without tryptophane (Trp) and leucine (Leu) (Sigma, St. Louis, MO, USA, A8056), and medium was incubated at 30 °C for 3-5 days. Positive clones were con rmed by PCR with general primers AD-R/F or BD-F/R (Listed in supplementary Table S1). Three randomly selected positive yeast cells were spotted on the X-α-gal-supplemented selection medium without Leu, Trp, histidine (His), and adenine (Ade). The interaction between the tested proteins was analyzed after incubating the positive yeast cell for 3-5-day at 30 °C.

Vector construction and plant transformation
The full-length coding sequences of TeAP1-1, TeAP1-2, TeFUL1, TeFUL2, and TeFUL3 were ampli ed by using primer pairs with speci c restriction sites (Supplementary Table S1), and the ampli cation products were ligated to the plasmid under the control of CaMV35S promoter. The recombinant plasmids were named 35S:TeAP1-1, 35S:TeAP1-2, 35S:TeFUL1, 35S:TeFUL2, and 35S:TeAFUL3, respectively. All the recombinant plasmids were introduced into Escherichia coliDH5a and tested by sequencing. These plasmids were separately transformed into chemically competent Agrobacterium tumefaciens strain GV3101. Wild-type Arabidopsis ecotype Columbia plants were transformed by oral dip method [56]. Transgenic plants of T1 and T2 generations were selected in kanamycin and veri ed by PCR with general forward primer of 35S-F and gene-speci c reverse primers 35S-TeAP1-1-R, 35S-TeAP1-2-R, 35S-TeFUL1-R, 35S-TeFUL2-R, and 35S-TeFUL3-R (listed Supplementary Table S1), respectively. The genomic DNA was isolated from the transgenic plants and wild-type Arabidopsis, respectively. The transcript levels of TeAP1-1, TeAP1-2, TeFUL1, TeFUL2, and TeFUL3 were analyzed by semi-quantitative PCR (Semi-PCR). The total RNA of blooming owers from T1 transgenic lines and wild type plants was isolated and reverse-transcribed with the above-mentioned reagent kit. The Arabidopsis house-keeping gene EF1α (AtEF1α, AT5G60390) was used as a control. Phenotype changes of transgenic plants of both T1 and T2 generations were analyzed. The main morphological traits of the control and transgenic plants of T2 generations were recorded. The transcript levels of some endogenous genes of the T3 generations were analyzed

Expression Analysis of Endogenous Genes in transgenic plants
In order to investigate the functional conservation of AP1/FUL-like genes in marigold and to reveal the mechanism underlying phenotypic changes of 35S:TeAP1-2 and 35S:TeFUL2 transgenic lines, the transcript levels of some endogenous genes regulated by AP1 (including LFY, FT, SEP3, SOC1, SVP, TFL1, AGL24, and SPL9) were analyzed by qRT-PCR [57]. Total RNA was isolated from 35S:TeAP1-2, 35S:TeFUL2 transgenic (T3) and wild-type Arabidopsis seedlings 10 days old. Reverse transcription and qRT-PCR were performed in the same way as described above. The gene-speci c primers were listed in Supplementary table S1.  Phylogenetic tree based on the amino-acid alignment of TeAP1/FUL proteins. The tree was generated with the MEGA v7.0 software, using the neighbor-joining (NJ) method and 1000 bootstrap replicates. The TeAP1-1, TeAP12, TeFUL1, TeFUL2 and TeFUL3 are marked with black stars.   qRT-PCR analysis of endogenous owering related genes in 10-D-old seedlings of Arabidopsis wild-type and 35S:TeAP1-2 and 35S:TeFUL2 transgenic lines. WT1: wild-type line 1; WT2: wild-type line 2; * expression level of endogenous genes in transgenic plants was 2 times higher or 1/2 lower than that in wild-type plants.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.