Introduction

In plants, MADS-box genes control the development of distinct organs, such as flower, ovule, fruit, leaf, and root1,2,3. Plant MADS-box genes can be classified into types I and II genes on the basis of phylogenetic analysis4. The best studied plant type II MADS-box transcription factors are those involved in floral organ identity determination. The determination of floral organ primordia by genes of the A, B, C, D and E classes led to the ABCDE model5,6,7,8,9,10. Furthermore, most of plant type II MADS-box proteins share a conserved structure consisting of four domains: MADS (M), intervening (I), keratin-like (K), and C-terminal (C)11,12. The DNA binding partner specificity is mediated to a large extent by the I domain, and the K domain likely promotes protein dimerization as well as tetramerization13,14,15. Proteins of floral MADS-box genes participating in floral organ identity interact with each other to control downstream genes16,17,18,19. Most of the protein–protein interactions necessary for the constitution of quaternary complexes, as recommended by the “quartet model”, are conserved20. Proliferating ovule primordia is specified by specific ovule identity factors, such as the MADS-box family members SEEDSTICK (STK), SHATTERPROOF1 (SHP1), SHP2, SEPALLATA (SEP) and AGAMOUS21,22,23. Moreover, Bsister genes are “marker genes” for the development of (inner) integument structures, phylogenetically the “oldest” structures surrounding the female gametophyte of seed plants24.

In dicotyledons, the first Bsister gene was isolated from Arabidopsis thaliana and was named ARABIDOPSIS BSISTER (ABS/TT16)25,26. ABS/TT16 is expressed mainly in the innermost integument layer, the endothelium; a closely related paralog is GORDITA (GOA, formerly known as AGL63). Besides study of eudicots, Bsister MADS-box genes have been investigated in the monocot Oryza sativa. ABS/TT16 and GOA were found not functionally redundant in ovule development24,26,27,28. ABS/TT16 is required for the proper differentiation of the inner integument, and GOA is necessary for at least the early development of the outer integument28. The single mutants abs/tt16 and goa still produce seeds, which germinate properly24,26,27. In addition, silencing of OsMADS29, a Bsister gene in rice, led to severe phenotypes with degeneration of the pericarp, ovular vascular trace, integuments, nucellar epidermis and nucellar projection29.

Orchids, constituting approximately 10% of all seed plant species, have enormous value for commercial horticulture and are of specific scientific interest because of their extraordinary diversity of floral morphology, ecological adaptations, and unique reproductive strategies30. The unique reproductive strategies include mature pollen grains packaged as pollinia, pollination-regulated ovary/ovule development, synchronized timing of micro- and mega-gametogenesis for effective fertilization, and release of thousands or millions of immature embryos (seeds without endosperm) in mature pods31.

In most flowering plants, the ovules are mature, and the egg cells are ready for fertilization at anthesis. In contrast, in orchids, ovule development is triggered by pollination. In most orchids such as Cattleya, Sophronitis, Epidendron, Laelia, Phalaenopsis, Dendrobium and Doritis, ovules are completely absent in unpollinated ovaries, and the development of ovule is triggered only after pollination32. The long-term progressive process of ovule development in orchids as compared with other flowering plants is an attractive system for investigating ovule initiation and subsequent development.

With high economic value, Phalaenopsis orchids are beautiful ornamental plants and very popular worldwide. The genome of P. equestris was recently sequenced33 and the information provides a great opportunity to identify and characterize the genes involved in regulating orchid ovule development34. In this study, we identified and functionally characterized PeMADS28, a Bsister MADS-box gene, in P. equestris. Our results indicate that the function of PeMADS28 plays an important role in ovule integument development in orchid and reveals the functional conservation of Bsister genes between monocots and dicots.

Results

Identification of PeMADS28 MADS-box gene in P. equestris

Only one Bsister MADS-box gene, PeMADS28 (predicted proteome gene ID Peq004141), exists in the P. equestris genome33. The sequence of PeMADS28 was retrieved from OrchidBase35,36. The ORF including 723 bp encodes a protein of 240 amino acids. Multiple sequence alignment with other Bsister proteins from gymnosperm, dicots and monocots demonstrated that PeMADS28 has a typical MIKC-type domain structure (Supplementary Fig. S1). Bsister proteins also contain a conserved PI Motif-Derived sequence in their C-terminal regions that are also representative of B-class MADS-box proteins (Supplementary Fig. S1)25.

Phylogenetic relationship of PeMADS28 and other MADS-box genes

To determine the phylogenetic relationships of PeMADS28 and other Bsister genes, we constructed a phylogenetic tree by using the amino acid sequences of PeMADS28 with other known gymnosperm and angiosperm Bsister sequences and the AGL63-like sequence from Brassicaceae. Amino acid sequences of Bsister proteins and AGL63-like proteins were retrieved from the National Center for Biotechnology Information (NCBI). This phylogeny has two supported major clades, one containing monocot Bsister proteins and the other dicot Bsister proteins (Fig. 1). Moreover, the Bsister and AGL63-like proteins were divided into two groups in dicots clade. PeMADS28 is close to the orchid Erycina pusilla Bsister protein EpMADS24 (Fig. 1). These results strongly suggest that PeMADS28 belongs to the Bsister gene family.

Figure 1
figure 1

Phylogenetic analysis of Bsister proteins. GGM13 from Gnetum gnemon and GbMADS10 from Ginkgo biloba were outgroup representatives. Bootstrap values from 1000 replicates are indicated on most major nodes. PeMADS28 is highlighted by the asterisks.

Full size image

Spatial and temporal expression of PeMADS28 in P. equestris

RT-PCR and quantitative real-time RT-PCR were used to survey the spatial and temporal expression patterns of PeMADS28. Because Bsister MADS-box genes are involved in ovule development and pollination is a key regulatory event in orchid ovule initiation, we determined the temporal mRNA expression patterns of PeMADS28 in developing ovules triggered by pollination. During ovule development, PeMADS28 transcript level was highest from 32 to 48 days after pollination (DAP) (Fig. 2a,b), then decreased from 56 to 100 DAP (Fig. 2a,b). However, PeMADS28 expression was barely observed in flower buds and was absent from vegetative tissues (Supplementary Fig. S2). Previous research showed that ovule development between 32 and 48 DAP is associated with inner and outer integument development37. These results suggest PeMADS28 has functions in ovule integument development.

Figure 2
figure 2

Expression patterns of PeMADS28 at various developing ovule stages in Phalaenopsis equestris (a) RT-PCR analysis of PeMADS28. Expression of Phalaenopsis actin was an internal control. (b) Quantitative real-time RT-PCR analysis of PeMADS28. DAP: days after pollination.

Full size image

In situ hybridization of PeMADS28 transcripts

We further examined the detailed spatial and temporal expression patterns of PeMADS28 during ovule development by in situ hybridization with antisense RNA probes. During the early stage of ovule development, when the final branches of placental protuberances differentiate ovular primordia, PeMADS28 transcript expression was detected in all ovule primordia at their initiation (Fig. 3a,c,d). In the later stage, the expression was more concentrated in developing ovules (Fig. 3e). At 48 DAP, PeMADS28 mRNA was detected in the whole ovule including nucellus and integument (Fig. 3f). PeMADS28 transcript expression was not detected in 56-DAP ovules (Fig. 3h). The negative control was sense RNA used as a probe (Fig. 3b,g,i). These results supported that PeMADS28 might be involved in orchid ovule initiation and integument development.

Figure 3
figure 3

In situ hybridization of PeMADS28 in developing ovules of P. equestris. (a,b) Placenta with ovule primordium at 4 DAP; (c) placenta with developing ovule at 32 DAP; (d) enlarged region of the dark arrow in (c); (e) placenta with developing ovule at 40 DAP; (f,g) developing ovules at 48 DAP; (h,i) developing ovule at 56 DAP. In (a), (c), (d), (e), (f) and (h), antisense probes were used to detect PeMADS28 transcripts. In (b), (g) and (i), hybridization involved sense probes (negative controls). Bars, 0.1 mm. p, placenta; op, ovule primordium; do, developing ovule; ii, inner integument; oi, outer integument; nu, nucellus; me, megaspores; DAP: days after pollination. The nucellus is highlighted by the red dash line.

Full size image

Subcellular localization of PeMADS28-GFP fusion protein

As a member of MADS-box transcription factors, PeMADS28 was expected to localize in the nucleus. To test the subcellular localization of PeMADS28, a PeMADS28-GFP fusion protein was generated with a GFP reporter gene fused in-frame to the PeMADS28 coding region under control of the 35S promoter. Transient expression of PeMADS28-GFP fusion protein was analyzed in Phalaenopsis petal protoplasts. PeMADS28-GFP fusion protein signals were observed in both the nucleus and cytoplasm along with GFP signals (Fig. 4). Thus, PeMADS28 might need to interact with other proteins to exclusively localize in the nucleus.

Figure 4
figure 4

Localization patterns of PeMADS28-GFP fusions in Phalaenopsis protoplasts. Images show fluorescence and bright-field confocal microscopy images and merged images of flower protoplast. (a) Empty vector was no green fluorescence in the cytoplasm and nucleus. (b) Cell in (a) stained with propidium iodide (PI) represented in red to confirm the nucleus. (c) Cell in (a) and (b) by bright-field confocal microscopy. (d) Merged image of (a), (b) and (c) to confirm green fluorescence in the cytoplasm of a flower cell. (e) PeMADS28-GFP green fluorescence in nucleus and cytoplasm. (f) Cell in (e) stained with PI represented in red to confirm the nucleus. (g) Cell in (e) and (f) by bright-field confocal microscopy. (h) Merged image of (e), (f) and (g) to confirm green fluorescence in cytoplasm. Bars: 20 µm.

Full size image

Interaction behavior of PeMADS28 analyzed by bimolecular fluorescence complementation (BiFC) assay

A number of previous studies demonstrated that MADS-box transcription factors form dimers or higher-order complexes for their functions in flower and ovule development17,20,21,24,38,39,40. To investigate the ability of homodimer formation of PeMADS28 and nuclear localization of this self-association, we used BiFC assay. A BiFC vector pair with PeMADS28 fused to N- or C-terminal halves of YFP (PeMADS28:YFPn and PeMADS28:YFPc) was prepared and used to co-transfect Phalaenopsis petal protoplasts. Fluorescence YFP signal clearly indicated an interaction between the two PeMADS28 monomers. The formed homodimer was exclusively localized in the nucleus (Fig. 5a). These results demonstrate that dimerization of PeMADS28 monomers plays an important role in retaining PeMADS28 in the nucleus. Previous reports indicated that PeMADS1 (C-class), PeMADS7 (D-class), and PeSEP3 (E-class) MADS-box genes are involved in orchid ovule development37,41. To gain more insight into the interaction of MADS-box proteins involved in orchid ovule development, interaction behaviors among Bs and C-class, D-class, and E-class proteins were further investigated by BiFC assay. Interaction fluorescence signals were observed in the combination of PeMADS28 and PeSEP3 (Fig. 5b,c) as well as PeMADS28 and PeMADS7 (Fig. 5d,e), which suggests that the orchid Bsister protein can interact with E- and D-class MADS-box proteins. In addition, the signals were localized in the nucleus, as indicated by use of the nuclear dye propidium iodide (PI). However, interaction was not observed with the combination of PeMADS28 and PeMADS1 (Fig. 5f,g). Therefore, PeMADS28 may not form heterodimers with C-class MADS-box proteins. No fluorescence was detected with the empty vector control (Fig. 5h).

Figure 5
figure 5

Analysis of protein–protein interactions among Bsister PeMADS28, C-class PeMADS1, D-class PeMADS7 and E-class PeSEP3 proteins by BiFC method. Fusion proteins were expressed in Phlaenopsis petal protoplasts. (a) PeMADS28:YFPc + PeMADS28:YFPn. (b) PeMADS28:YFPc + PeSEP3:YFPn. (c) PeSEP3:YFPc + PeMADS28:YFPn. (d) PeMADS28:YFPc + PeMADS7:YFPn. (e) PeMADS7:YFPc + PeMADS28:YFPn. (f) PeMADS28:YFPc + PeMADS1:YFPn. (g) PeMADS1:YFPc + PeMADS28:YFPn. (h) YFPc + YFPn as a negative control. BF, bright field; Fl, fluorescence image; M, Merged image; PI, propidium iodide. Bars, 20 µm.

Full size image

Functional analysis of PeMADS28 gene by ectopic expression and complementation in Arabidopsis thaliana

For functional characterization of PeMADS28, we constructed transgenic Arabidopsis plants expressing PeMADS28 under control of the cauliflower mosaic virus (CaMV) 35S promoter via Agrobacterium-mediated transformation. A total of 20 independent overexpressed PeMADS28 transgenic lines were obtained based on kanamycin selection and a similar phenotype. Among twenty transgenic lines, nine showed a 3:1 segregating kanamycin resistance phenotype. As compared with wild-type plants, six independent PeMADS28 overexpressed lines shows the early flowering phenotype (Fig. 6a,c [wild-type plant]; 6b,d [transgenic plant]) and fewer flower bud production (Fig. 6i [wild-type plant]; 6j [transgenic plant]). The rosette and cauline leaves of transgenic plants had upwardly curled profiles and were smaller than those of wild-type plants (Fig. 6a,k [wild type plant]; 6b,l [transgenic plant]). No homeotic conversion of floral organs was observed in transgenic plants. Moreover, transgenic plants had smaller flowers with cracked sepals than wild-type plants (Fig. 6e,g [wild-type plant]; 6f,h [transgenic plant]). The length of siliques was shorter in transgenic plants (Fig. 6n and Table 1). Transgenic plants also showed more undeveloped seeds in siliques than did wild-type plants (Fig. 6o and Table 1). In addition, transgenic seeds were larger and heavier (Fig. 6m and Table 1). Previously, GORDITA and ABS/TT16 are the paralogs in Arabidopsis. Consistently, over-expressed the GORDITA or ABS/TT16 in Arabidopsis caused that the plant size shorter, and all organs are smaller than those in the wild-type24,26,28. Both two over-expressed plants were affected the fruit development, and ABS/TT16 led to the rosette leaves curled24,26,28. It is similar to 35::PeMADS28 phenotype. These results appeared that the PeMADS28 may play a role in fruit development.

Figure 6
figure 6

Phenotype analysis of transgenic Arabidopsis overexpressing PeMADS28. (a) 20-day-old wild-type plant; (b) 20-day-old 35S::PeMADS28 transgenic plant; (c) 31-day-old wild-type plant; (d) 31-day-old 35S::PeMADS28 transgenic plant. Bars (a–d), 5 mm. (e) Side-view of wild-type flower; (f) side-view of 35S::PeMADS28 transgenic flower; (g) top-view of wild-type flower; (h) top-view of 35S::PeMADS28 transgenic flower; (i) Wild-type floral inflorescence; (j) 35S::PeMADS28 transgenic floral inflorescence; (k) cauline leaf of wild-type plant; (l) cauline leaf of 35S::PeMADS28 transgenic plant. Bars (e–l) 1 mm. (m) Wild-type seed (left); 35S::PeMADS28 transgenic seed (right); (n) silique of wild-type (upper); silique of 35S::PeMADS28 transgenic plant (lower); (o) silique of wild-type without one valve (upper); silique of 35S::PeMADS28 transgenic plant without one valve (lower). Bars (m–o) 1 mm.

Full size image
Table 1 Silique length and seeds in OXPeMADS28 transgenic plants and wild-type (WT) plants.
Full size table

To further validate the function of PeMADS28, we used complementation testing with the tt16-1 mutant and examined the seed pigmentation and development of the endothelium. A total of 8 transgenic lines were obtained and 4 showed a 3:1 segregating kanamycin-resistance phenotype. All of the T2 line seeds showed restoration of pigmentation to a brown color from the straw color of the tt16-1 mutant (Fig. 7a–c). In addition, PeMADS28 could rescue the development of endothelium in tt16-1 plants. Endothelial cells in immature wild-type seeds were small, almost rectangular in shape and regularly spaced (Fig. 7d–f). In abs/tt16 immature seeds, endothelium cells seemed to be flatter and more irregularly shaped than wild-type cells, resembled parenchymatic cells, and often seemed to collapse (Fig. 7f)26. All of the T2 line (35S::PeMADS28 transgenic tt16-1) seed coats showed the normal endothelium of the wild-type seed coat (Fig. 7e), so PeMADS28 was sufficient to complete the function of Arabidopsis ABS/TT16.

Figure 7
figure 7

Phenotypes of seed pigmentation and structure of the seed coat. Seed pigmentation of mature seeds from wild-type (Col) (a), a tt16-1 mutant (b), and transgenic 35S::PeMADS28 in tt16-1 mutant plant (c). The development of seed coat from cleaned seed in wild-type (Col) (a), tt16-1 mutant (b), and transgenic 35S::PeMADS28 in tt16-1 mutant plant (c). The dark arrows are indicated the endothelium. cb, chalazal bulb; ii, inner integument; mi, micropyle; oi, outer integument.

Full size image

Discussion

In this work, we identified a Bsister-like gene, PeMADS28, from the P. equestris genome and characterized its function by sequence comparison, expression profile analysis, protein–protein interaction behavior, ectopic expression and complementation experiments in Arabidopsis. Protein sequence alignment showed that PeMADS28 is a typical Bsister protein with respect to its protein sequence because it contains a conserved sub-terminal “PI motif-derived sequence,” which is also representative of B-class MADS-box proteins25. Phylogenetic analysis with use of a deduced amino acid sequence revealed that PeMADS28 belongs to the monocot Bsister subclade. Both analyses suggested that PeMADS28 is a putative orchid ortholog of Bsister genes like ABS/TT16 from Arabidopsis.

Bsister genes are closely related to B-class genes but express predominantly in the female reproductive organ. Previous studies showed that Bsister genes are expressed in the ovule and envelope in gymnosperms and in the ovule and integuments of angiosperms. In the gymnosperm Gnetum gnemon, expression of Bsister GGM13 is specifically strong at the adaxial base of the cupules, where ovules subsequently develop42. When ovules appear, GGM13 expression is limited to the developing nucellus and inner envelopes42. In dicots, the Arabidopsis Bsister gene ABS/TT16 is expressed mainly in endothelium43. In petunia, FBP24 is expressed in young ovule primordia, nucellus and integument. Later, the expression is confined to the endothelium in mature ovules38. In snapdragon, DEFH21 expression was found in only a few inner cell layers of the inner integuments of the ovules25. In monocots, wheat WBsis mRNA was detected in the developing inner integument at the late floral organ developmental stage44. In rice, OsMADS29 transcripts are localized in the ovule, including integuments and nucellus throughout ovule development29.

These results reveal a similarity of expression of Bsister genes suggesting conservation of the gene expression pattern over at least 300 million years27. In our study, temporal expression analysis revealed significant PeMADS28 transcript expression between 32 and 48 DAP (Fig. 2a,b). In addition, in situ hybridization signals of PeMADS28 transcripts were concentrated in the developing ovules (Fig. 3c–f). Hence, Bsister genes may have conserved expression patterns in seed plants. Interestingly, although Arabidopsis genome contains two and rice has three Bsister genes, these homologous genes have been occurred diversified expression and functional differentiation. ABS/TT16 is involvement in endothelial cell specification and control of flavonoid biosynthesis in Arabidopsis seed coat26. The GOA is a young paralog of ABS/TT16 and play a role in fruit longitudinal growth27. The rice OsMADS29 was identified as a key regulator of early rice seed development by regulating the programmed cell death of maternal tissues29. OsMADS30 does not have a canonical ‘Bsister function’, and revealed neo-function in shoot size and architecture45. In fact, the development of orchid ovule is the typical monosporic Polygonum type in which the functional megaspore passes through three mitotic divisions producing a seven celled embryo sac consisting of three antipodal cells, one central cell formed by two polar nuclei, two synergid cells, and the egg cell. In addition, the embryo sac is enclosed by inner and outer integuments. The orchid ovule structure and development is highly similar to that of Arabidopsis and cereal except that the inner integument gradually degenerated during the early stages of embryo proper formation and ovule initiation and development is precisely triggered by pollination. Our data considered that the Bsister gene PeMADS28 might involve in the typical ovule development including integument morphogenesis.

Previously, it has been shown that the antagonistic development of nucellus and endosperm in Arabidopsis46. The endosperm delivers the signal for the differentiation of seed coat and then both of tissues orchestrates seed growth. However, the endosperm could also initiate nucellus degeneration via vacuolar cell death and necrosis46. It also has been demonstrated that TT16/ABS can regulate proanthocyanidins synthesis in the seed coat and conversely TT16/ABS expression in the seed coat is sufficient to activate the nucellus degeneration46. In Phalaenopsis orchids, double fertilization could be observed. However, the triple fusion nucleus of the endosperm initial is amorphous in shape and apparently begins to degenerate immediately, consequently forming no endosperm47,48. We speculated that the signal generated by fertilization of the central cell triggers its degeneration through activation of the PeMADS28 expression. Because endosperm initial lives shortly, the signal might not spread to the seed coat. In fact, the Phalaenopsis seed coat do not accumulate proanthocyanidins49. However, whether signal generated by fertilization of the central cell could reach to the nucellus and initiates the nucellus cell death should be necessary for further study.

As transcriptional regulatory proteins, a number of MADS-box proteins have been shown to localize in the nucleus. However, some MADS-box proteins are unable to translocate into the nucleus by themselves, but their dimers deposit in the nucleus; examples are AP3-PI50 and UNSHAVEN-FLORAL BINDING PROTEIN 9 (FBP9)51. In this study, we detected PeMADS28-GFP fusion proteins in the nucleus and cytoplasm (Fig. 4h). However, BiFC results showed the PeMADS28 homodimer specifically retained in the nucleus. The results suggest that homodimerization of PeMADS28 drives a conformational change to bring it into a nuclear-retaining structure. This kind of behavior of orchid Bsister PeMADS28 is similar to that of rice OsMADS2952. Our data suggest the probability of PeMADS28 being regulated at the post-translational level via its interactions, which may affect its function by regulating entry into the nucleus and regulation of its targets.

In Arabidopsis, previous study suggested a specific interaction of ABS with STK, SEP3, SHP1 and (much weaker) SHP2 but not AG24,38. OsMADS29 could interact with OsMADS3 (C-class proteins) and all five E-class proteins of rice52. Our results indicate that PeMADS28 can form a homodimer in the nucleus. In addition, it could interact with D-class (PeMADS7) and E-class (PeMADS8) MADS-box proteins. However, PeMADS28 and PeMADS1 may not form heterodimers directly. These results suggest that protein interaction behaviors among Bsister, D- and E-class proteins are conserved in angiosperms. Furthermore, in previous study, protein–protein interaction analyses revealed that PeSEP3 could bridge the interaction between PeMADS1 and PeMADS7 involved in Phalaenopsis gynostemium and ovule development37. Thus, a higher-order protein complex formed by C-E-D-Bsister genes (PeMADS1-PeMADS8-PeMADS-PeMADS28) might have an important role in regulation of orchid ovule development.

Functional analysis of ABS has shown abnormal characteristics in vegetative and reproductive organs of ABS-overexpressing Arabidopsis, including curled rosette leaves, late flowering, small flowers and shrunken siliques with few developed seeds26. Overexpression of GOA, the paralog of ABS, showed similar phenotypes as ABS-overexpressed plants, except that GOA-overexpressing plants displayed early flowering27. Overexpression of the Ginkgo Bsister gene GBM10 in tobacco resulted in reduced size of transgenic seedlings, small and curled leaves, small flowers, small fruit with wrinkled surface and massive abortion of undeveloped ovules42. Similar to these phenotypes, our PeMADS28-overexpression Arabidopsis showed curled and small rosette leaves, early flowering, small flowers, short siliques and few developed seeds. Overexpressing PeMADS28 in wild-type Arabidopsis demonstrated that PeMADS28 has functions similar to those of Bsister genes in regulating ovule development. Moreover, overexpression of PeMADS28 could restore the development of endothelial cells in the tt16 mutant. Conserved functions of orchid Bsister genes for specifying integument development could occur in developing seeds of Arabidopsis, which indicates that a competent endothelium is needed for PeMADS28 function to specify integument development.

In most orchids, ovary and ovule development is precisely triggered by pollination32. Previous studies showed that pollination inhibits PeMADS6 (B-PI MADS-box gene) expression in the ovary via the auxin signaling pathway to promote Phalaenopsis ovary/ovule development32,53. In addition, expression of C-class PeMADS1 and D-class PeMADS7 was significantly induced by pollination37. Furthermore, the TCP gene PeCIN8 showed a parallel expression pattern in the developing ovules of Phalaenopsis to that of PeMADS2834. Understanding the interaction as well as regulation networks of these genes, then stimulating pollination will help in further exploring the molecular mechanism of orchid ovule development. Moreover, the availability of several whole-genome sequences of orchids, including P. equestris, Dendrobium catenatum, and Apostasia shenzhenica33,54,55,56, can lead to promising exploration of more genes involved in the orchid ovule development.

Materials and methods

Plant materials and growth conditions

The plants of wild-type P. equestris (S82–159) were grown in greenhouses under natural light and controlled temperature from 23 to 27 °C48. A. thaliana ecotype Columbia was used in transformation experiments. Seeds were surface-sterilized in 10% (v/v) bleach for 15 min, then rinsed 3–4 times with sterile water. Sterilized seeds were grown on half-strength Murashige and Skoog medium (INVITROGEN, CARLSBAD CA, USA) in the presence of 1% (w/v) sucrose and 0.8% (w/v) agar. Plated seeds were incubated at 4 °C for 48 h, then maintained in a fully automated growth chamber (CHIN HSIN, Taiwan) under a 16-h light/8-h dark photoperiod at 22 °C for 10 days before being transplanted to soil37.

Sequence alignment and phylogenetic analysis

Sequence alignment involved use of CLUSTALW and phylogenetic analysis MEGA 6 by the neighbor-joining method. Bootstrap analysis was with 1000 replicates. The GeneBank accession numbers for amino acid sequences are AtsMADS29 (XP_020188803), TaWM25 (CAM59071), HvBsister (BAK06913), BdMADS29 (NP_001288325), OsMADS29 (XP_015624837), SbBsister (XP_002453370), ZMM17 (NP_001105130), OsMADS30 (Q655V4), OsMADS31 (Q84NC2), MaMADS29 (XP_018678849), PdAP3-like (XP_00880798), AcAP3-like (XP_020109780), PeMADS28 (KT865880), EpMADS24 (AHM92100), VvFBP24 (RVW42148), VrFBP24-like (XP_034698718), ABS (Q8RYD9), RsTT16-like (XP_018481949), AmDEFH21 (CAC85225), FBP24 (AAK21255), SlFBP24-like (XP_019066630), AmtGGM13 (XP_006829168.2), CcBsister1 (ADD25185), GGM13 (CAB44459), GbMADS10 (BAD93174), GORDITA (NP_174399.2), CrAGL63 (XP_006306362.2), LcAGL63 (APB93359).

RNA extraction

We collected unpollinated ovaries; ovaries and ovules at 1, 2, 4, 8 day after pollination (DAP); ovules at 16, 32, 40, 48, 56, 64 DAP; and developing seeds at 80 and 100 DAP from P. equestris37. Samples were immersed in liquid nitrogen, and stored at – 80 °C until the RNA was extracted. Total RNA was isolated with use of TRIZOL reagent (SIGMA-ALDRICH). Briefly, frozen tissue (0.5–1 g) was ground with liquid nitrogen with a pestle and mortar and homogenized in TRIZOL reagent. Then the dissolved RNA was extracted with chloroform. After centrifugation in 13,000 rpm to remove insoluble material, total RNA was precipitated with isopropanol and 0.8 M sodium citrate was added to dissolve polysaccharides at − 20 °C overnight; then samples were precipitated again with 4 M LiCl, pelleted, washed, and the final RNA precipitate was dissolved in a suitable volume of sterilized DEPC-treated water. Before cDNA synthesis, RNA was treated with RNase-free DNase I (INVITROGEN) to remove DNA contamination.

RT-PCR and quantitative real-time PCR

RNA was used as a template for cDNA synthesis with reverse transcriptase and the SuperScript II kit (INVITROGEN). Transcripts of PeMADS28 were detected by RT-PCR with gene-specific primers (Supplementary Table S1) for 25–30 cycles. The RT-PCR program was 95 °C for 7 min for denaturation of DNA and activation of polymerase, then amplification at 95 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s and extension at 72 °C for 10 min as described previously37. The amplified products were analyzed on 1% agarose gels. Quantitative real-time PCR involved using the ABI Prism 7000 sequence detection system (APPLIED BIOSYSTEMS) with 2X SYBR green PCR master mix (APPLIED BIOSYSTEMS)34. Reaction involved incubation at 50 °C for 2 min, then 95 °C for 10 min, and thermal cycling for 40 cycles (95 °C for 15 s and 60 °C for 1 min). The relative quantification was calculated according to the manufacturer’s instructions (APPLIED BIOSYSTEMS)34. The expression of PeActin4 (PACT4, AY134752) was used for normalization34. Primers used for amplification are in Supplementary Table S1.

In situ hybridization

Developing ovules and developing seeds of P. equestris were fixed in 4% (v/v) paraformaldehyde and 0.5% (v/v) glutaraldehyde for 24 h at 4 °C, dehydrated through an ethanol series, embedded in Histoplast and longitudinal sectioned at 6–8 μm with use of a rotary microtome. Tissue sections were deparaffinized with xylene, rehydrated through an ethanol series, pre-treated with proteinase K (2 μg ml−1) in 1 × phosphate-buffered saline (PBS) at 37 °C for 60 min, acetylated with 0.5% acetic anhydride for 10 min, and dehydrated with an ethanol series. The resulting PCR fragments were used as templates for synthesis of both antisense and sense riboprobes with digoxigenin-labeled UTP-DIG (ROCHE APPLIED SCIENCE) and the T7/SP6 Riboprobe in vitro Transcription System (PROMEGA) following the manufacturer’s instructions. For quality control, hybridization probes were tested by using dot blot to analyze the sensitivity before in situ hybridization. Hybridization and immunological detection of signals with alkaline phosphatase were performed as described48.

Subcellular localization of PeMADS28-GFP fusion protein

Template-specific primers were designed by the addition of an attB1 adapter primer (5′-GGG GAC AAG TTT GTA CAA AAA AGC AGG CTG G-3′) to the 5′ end of the first 18–25 nt of the open reading frame (ORF) and attB2 adapter primer (5′- GGG GAC CAC TTT GTA CAA GAA AGC TGG GTT-3′) to the 3′ end of the first 18–25 nt of the ORF, which generated the full-length attB1 and attB2 sites flanking the ORF (Supplementary Table S1). Gateway -compatible amplified ORFs were recombined into the pDONR 221 vector (INVITROGEN) by BP cloning: 1 µl (15–150 ng) PCR products, 2 µl BP clonase II Enzyme Mix (INVITROGEN), 150 ng pDONR vector plasmid and TE buffer (pH 8.0) were incubated at 25 °C for 1 h. Entry clones were used directly for transformation of E. coli DH5α cells, and bacteria were plated on LB medium containing 50 µg/ml of kanamycin. These entry clones were for recombination of target genes into the destination vector p2GWF7, C-terminal fusions57 in a reaction mixture containing 2 µl LR clonase II Enzyme Mix (INVITROGEN), 150 ng p2GWF7 vector, and TE buffer (pH 8.0), and incubation at 25 °C for 1 h. The LR reactions were used for transformation, then transformants were selected in plates containing 50 µg/ml ampicillin. The plasmids were transfected into Phalaenopsis protoplasts by PEG transformation. After culturing for 16 h, signals were visualized under a confocal laser microscope (CARL ZEISS LSM780, Instrument Development Center, NCKU). Separate bright field and fluorescence images were overlaid by using Axio Vision 4 Rel.4.8.

Bimolecular fluorescence complementation assay (BiFC)

To construct the interaction vectors, we used gene-specific primers with an additional attB1 adapter primer (5′-GGG GAC AAG TTT GTA CAA AAA AGC AGG CTG G-3′) added to the 5′ end of the first 18–25 nt of the ORF and attB2 adapter primer (5′- GGG GAC CAC TTT GTA CAA GAA AGC TGG GTT-3′) to the 3′ end of the first 18–25 nt of the ORF by using Pfu DNA polymerase. Primers for amplification are in Supplementary Table S1. Gateway compatible amplified ORFs were recombined into the pDONR 221 vector (INVITROGEN) by BP cloning described previously34. Gateway LR clonase enzyme mix was used for cloning the entry clones into the BiFC destination vectors pSAT4-DEST-nEYFP-C1 (pE3136) and pSAT5(A)-DEST-cEYFP-N1 (pE3132). After the LR reactions, plasmids were transformed into DH5α cells and transfected into Phalaenopsis protoplasts by PEG transformation34. Signals were visualized by confocal laser microscopy (CARL ZEISS LSM780, Instrument Development Center, NCKU).

Arabidopsis transformation

cDNA fragments containing the coding regions of PeMADS28 were cloned into the pBI121 vector (primers are in Supplementary Table S1). Constructs were then introduced into Agrobacterium tumefaciens (strain GV3101). GV3101 was inoculated drop-by-drop into closed floral buds by using a micropipette. Arabidopsis transformation was modified by the addition of 0.05% (v/v) Silwet L-77 (LEHLE SEEDS, ROUND ROCK, TX, USA) in the transformation media. To select transformed Arabidopsis, seeds (T0) were screened on media supplemented with 50 μg/ml kanamycin (SIGMA- ALDRICH). After 2 weeks of selection, the kanamycin-resistant seedlings (T1) were transferred to soil and grown under the conditions described above. Kanamycin segregation in the T1 generation was analyzed by chi-square test. The homozygous, kanamycin-resistant T2 generation was used to confirm the integration fragment by PCR for each construct. Transformed lines with segregation ratio 3:1 were collected for further analysis. The seeds of 35S::PeMADS28 transgenic Arabidopsis plants were grown in the same environment as described previously37.

Complementation assay

The tt16-1 mutant was obtained from Dr. L. Lepiniec (Institut Jean-Pierre Bourgin, France26). The pBI121-PeMADS28 construct was transformed into the tt16-1 mutant and screened on media supplemented with 50 μg/ml kanamycin. The seeds of 35S::PeMADS28 tt16-1 transgenic Arabidopsis plants were used.

Differential interference contrast (DIC) microscopy

Immature seeds were removed from different developmental stages of siliques and soaked overnight in clear solution (chloral hydrate:water:glycerol, 8:2:1 [w/v/v]). The Cleared seeds were examined by using a microscope equipped with Nomarski optics.