Identification and Characterization of RcMADS1, an AGL24 Ortholog from the Holoparasitic Plant Rafflesia cantleyi Solms-Laubach (Rafflesiaceae)

Rafflesia, a holoparasitic genus that produces the largest flower in the world is characterized by the absence of leaves, stem and other macroscopic organs. To better understand the molecular regulation of flower development in this genus we isolated and characterized a floral MADS-box gene, namely, RcMADS1 from Rafflesia cantleyi. Heterologous expression analysis in Arabidopsis was chosen because Rafflesia is not amenable to genetic manipulations. RcMADS1 shares sequence similarity with AGAMOUS-LIKE 24 (AGL24) and SHORT VEGETATIVE PHASE (SVP) of Arabidopsis. Ectopic expression of RcMADS1 in Arabidopsis caused early flowering and conversion of sepals and petals into leaf-like structures, and carpels into inflorescences. In 35S::RcMADS1 plants SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1), a downstream target gene of AGL24, was upregulated. 35S::RcMADS1 plants exhibit early flowering and conversion of the floral meristem into inflorescence meristem, as in 35S::AGL24 plants. Similar to AGL24, RcMADS1 could rescue the late flowering phenotypes of agl24-1 and FRIGIDA, but not the early flowering of svp-41. Based on these results, we propose that RcMADS1 is a functional ortholog of Arabidopsis AGL24.


Introduction
Rafflesia is a parasitic plant in a distinctive flowering plant genus from Rafflesiaceae that develops the largest flower in the world [1]. The members of this genus are unique among the flowering plants owing to their highly reduced vegetative morphology, prominent and large floral structures, and physiology. Rafflesia species are holoparasitic endophytes of Tetrastigma (Vitaceae). They lack visible leaves, stems and roots and only appear as flowers for sexual reproduction [2]. Their vegetative body is reduced to mycelium-like structure, which grows completely embedded within the host plants. Hence, they are fully dependent on their host for nutrition. Flowers of Rafflesia cantleyi are particularly distinctive compared to those of other species. The large fleshy flowers can reach up to one meter in diameter (Figure 1), and produce the smell of rotting flesh that attracts carrion flies for pollination [2]. Rafflesia flowers have some unusual structures, such as a modified perianth (perigone) enclosed by a diaphragm; a central column with an apical disk; and the presence of ramenta on the interior surface of the perigone tube and diaphragm ( Figure  S1) [3].
The habitat of Rafflesia is confined to the Indo-Malayan region [3]. In recent years, several new species have been discovered in the Philippines such as Rafflesia speciosa [4], Rafflesia mira [5], Rafflesia baletei [6], Rafflesia lobata [7] and Rafflesia banahawensis [8]. Rafflesia has not been well studied, with only a few reports published in the past (for example, [1,9,10]). The paucity of work on this genus is partly due to its rarity and inaccessibility of its habitats. Holoparasitic plants like Rafflesia have undergone many physiological and morphological adaptations during their evolution and have lost most of the vegetative plant structures. Based on the sequence of the mitochondrial gene matR, Rafflesia was phylogenetically placed under the order Malpighiales [1]. Rafflesia was later found to be nested in the Euphorbiaceae in a more restricted study of the Malpighiales using five mitochondrial genes (ccmB, cob, matR, nad6 and rps3) and a chloroplast gene (matK) [11]. It is interesting to note that Rafflesia evolved from a family with very small flowers. It is worthwhile to study the molecular nature of floral development of Rafflesia because the flower is the only macroscopic structure of the plant, and the flowers are highly unusual ( Figure 1). Such studies can shed light on the developmental and evolutionary processes that Rafflesia has undergone.
Many key processes in growth and development are regulated by transcription factors, which can be classified into gene families according to the conserved DNA-binding domain present. In plants, there are about 60 different transcription factor gene families [12,13,14]. The MADS-box gene family is one of the major transcription factor gene families and it is particularly important in controlling floral transition, floral meristem identity, floral organ specification, and fruit and ovule development [15,16,17].The MADS-box encodes a DNA-binding domain comprising of approximately 60 amino acids, and is highly conserved across plants, fungi, and animals [18]. MADS-box genes are classified as type I (comprising of the subfamilies Ma, Mb and Mc), and type II (consisting of subfamilies Md/MIKC* and MIKC C ) [19,20,21]. Almost all plant MADS-box genes that have been characterized so far belong to MIKC C subfamily, with a modular structure comprising of four domains, namely, the MADS (M), intervening (I), keratin-like (K), and C-terminal (C) domains [18,19,22]. Additionally, based on their functions and expression patterns the MADS-box genes involved in regulating floral development are grouped under classes A, B, C, D, E, F, G, Bs (B-sister) and T [23].
Flowering is controlled by an intricate network of signaling pathways, which are regulated by environmental and developmental cues [24]. Five different pathways, namely, vernalization, photoperiod, gibberellin, autonomous and aging pathways are involved in this process [25,26]. Many genes characterized in these pathways were MADS-box genes, which include floral organidentity (eg. AGAMOUS and SEPALLATA 1, 2, 3 and 4) and flowering-time genes, such as AGL24 [27,28,29,30] and SVP [31,32]. AGL24 and SVP are members of the StMADS11 clade [22,31], and are involved in the contrasting functions of promotion and repression of flowering, respectively. AGL24 promotes flowering by inducing expression of SOC1 by directly binding to its promoter [30]. In contrast, SVP suppresses the expression of SOC1 by directly binding to its promoter [33]. These observations clearly show that AGL24 and SVP are key integrators of flowering signals, along with other floral transition signals [30].
Overexpression of AGL24 in Arabidopsis results in early flowering and floral abnormalities such as, secondary flowers in the axils of leaf-like sepals of primary flowers, elongation of the base of ovaries into inflorescence stem-like structures, and production of ectopic inflorescences from swollen ovaries [27,32]. In contrast, overexpression of SVP results in late flowering and loss of carpels as well as the conversion of flowers into shoot-like structures with chimaeric characteristics of vegetative shoots and flowers [32]. Homologues of AGL24 and SVP have been isolated from Antirrhinum [34], Brassica [35], barley [36], rice [37,38,39], tomato [40] and Withania [41]. When they are ectopically expressed in Arabidopsis, phenotypes are similar to either 35S::AGL24 [28] or 35S::SVP [32] plants with altered flowering time and floral organ defects. This shows that AGL24 and SVP are likely to have a conserved function in specifying floral meristem development among most of the flowering plants.
Here we report the isolation, molecular and functional characterization of a MADS-box gene named as RcMADS1 from Rafflesia cantleyi. The full-length cDNA was cloned using a reverse transcription-PCR approach. This cDNA shows high sequence similarity to several MADS-box genes, in particular, to AGL24 and  RcMADS1, an AGL24 Ortholog from Rafflesia PLOS ONE | www.plosone.org rescue and SOC1 expression analysis suggest that RcMADS1 is a functional ortholog of AGL24. Also, despite the highly specialized evolutionary reduction of vegetative parts in Rafflesia the molecular regulation of flowering may be conserved between Arabidopsis and Rafflesia.

Plant Materials
Flower buds of various sizes of Rafflesia cantleyi Solms-Laubach were collected from several localities in Pulau Tioman, Pahang, Malaysia (Permission from the Economic Planning Unit, Prime Minister's Office, Malaysia -project reference no. UPE:/40/200/ 19 SJ. 1200, and research pass no.1163). The buds were surfacesterilized using a 10% (v/v) CloroxH (1% sodium hypochlorite) for 5-10 min, followed by rinsing with sterile water. Tissues were cut and weighed, then flash-frozen in liquid nitrogen. All samples were stored at -80uC.
Transgenic and mutant Arabidopsis thaliana plants used in the experiments were of the same genetic background, i.e., Columbia ecotype. Arabidopsis thaliana seeds were sown on soil (Flora Fleur) and stratified for 3-4 days at 4uC, before being transferred to a growth chamber maintained at 2262uC under 16 h of light/8 h of dark photoperiod.

RNA Extraction
Total RNA from the Rafflesia cantleyi flower buds was isolated using a modified RNeasyH Plant Mini Kit (QIAGEN) method [42]. The modification involves an initial CTAB extraction [43]. The extract was then applied to an RNeasyH column and purified following the manufacturer's instructions.
Total RNA from Arabidopsis thaliana plant tissues was isolated using the RNeasyH Plant Mini Kit (QIAGEN) following the manufacturer's instructions.

Reverse Transcription
First strand cDNA synthesis was carried out from 2 mg of total RNA using MaximaH First Strand cDNA Synthesis Kit (Fermentas, Cat.No.K1641) as per the manufacturer's protocol.

PCR Amplification
PCR amplification of MADS-box genes from Rafflesia cantleyi cDNA was performed using degenerate primers and an oligo (dT) 15 primer. These degenerate primers were designed based on the conserved MADS box of MADS box genes. The primers used were forward: 59-GGGGTACCAAYMGICARGTIACITAYT-CIA AGMGIMG-39, reverse: PY1: GAGAGAGAGAGAAC-TAGTCTCGAGTTTTTTT TTTTTTTT. PCR reactions were performed using step-up conditions with the following cycling parameters: an initial denaturation at 95uC for 1 min; 10 cycles of denaturation at 95uC for 30 s, annealing at 35uC for 1 min, and extension at 72uC for 1 min; 25 cycles of denaturation at 95uC for 30 s, annealing at 40uC for 1 min, and extension at 72uC for 1 min; and a final extension at 72uC for 10 min. 1 mg of cDNA template was added to a reaction mixture consisting of 16 DyNAzyme PCR buffer, 0.2 mM dNTP mix, and 2 pmole each of forward and reverse primers and DyNAzyme polymerase. PCR reactions were visualized by performing gel electrophoresis in a 1.2% agarose gel. Amplified fragments over 400 bp in size were selected for cloning and sequencing.

Cloning, DNA Sequencing and Sequence Analysis of PCR Products
The PCR products were purified using the QIAquickH PCR purification kit (QIAGEN) following the manufacturer's instructions. The purified PCR product was cloned into the pGEMH-T Easy Vector (Promega). Selected clones were sequenced via an automated sequencing method using ABI PRISM TM Big Dye TM Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems). Sequences obtained after automated sequencing were collated and compared with published sequences in the GenBank database using the Basic Local Alignment Search Tool (BLAST) program on the National Center for Biotechnology Information (NCBI) website. The algorithms used were blastn (to search the nucleotide database using a nucleotide query) and tblastx (to

RACE
Complete cDNAs were obtained using BD SMART TM RACE cDNA Amplification Kit (Clontech) following the manufacturer's instructions. The full length cDNA sequence of RcMADS1 was

Construction of RcMADS1 Ectopic Expression Transgenic Arabidopsis thaliana Lines
The full open reading frame of the RcMADS1 cDNA from Rafflesia cantleyi was amplified using the following primers containing restriction enzyme sites RcMADS1-F-BamHI 59-GGATCCATGGCTCGAGAAAAGATCAA-39, and RcMADS1-R-SpeI 59-ACTAGTGCTTGAGAAGGA-CAATCCC-39. The PCR products were purified using the QIAquickH PCR purification kit (QIAGEN) following the manufacturer's instructions. The purified PCR product was cloned into pGEMH-T Easy Vector (Promega). Resultant plasmids were digested for 2 h with BamHI and SpeI restriction enzymes to release the RcMADS1 fragment, which was then inserted in between the CaMV 35S promoter and terminator in a sense direction into the pGreen0229 vector [45]. This ectopic expression construct was named 35S::RcMADS1. To generate AGL24::Rc-MADS1 construct ,3.2 kb genomic fragment of AGL24 promoter region was cloned, which includes the 1 st exon and intron (it was reported that 1 st intron of AGL24 has two CArG-box cis-elements, which is important for its regulation by upstream transcription factors) [32], Transformation of Arabidopsis thaliana plants was carried out using the floral dipping method as described by Clough and Bent, 1998 [46].  three technical replicates were used and relative expression levels were calculated as previously described [29].

Isolation and Sequence Analysis of a MADS-box Gene from Rafflesia cantleyi
We succeeded in cloning a MADS-box gene by reversetranscription PCR using degenerate primers with cDNA synthesized from Rafflesia cantleyi floral buds. One fragment of 750 bp from the initial round of PCR contained a MADS-box gene that was named RcMADS1 (for Rafflesia cantleyi MADS-box1). Using 59 Rapid Amplification of cDNA Ends (59 RACE) followed by amplification with gene-specific primer and oligo-dT primer, we obtained an 899 bp long sequence. This cDNA sequence contains a 687 bp long open reading frame (ORF) including a stop codon, which encodes a polypeptide of 228 amino acids, as well as 59 and 39 untranslated regions ( Figure S2). Results from a tblastx search using the online BLAST program revealed high sequence similarities to MADS-box proteins such as PtMADS1 from Populus tomentosa, JOINTLESS from Solanum lycopersicum, MPF2 from Physalis pubescens, MPP3 from Physalis peruviana, IbMADS4 from  Ipomoea batatas, StMADS11 and StMADS16 from Solanum tuberosum, and SVP and AGL24 from Arabidopsis thaliana, all from the StMADS11 clade of MADS-box genes. An alignment of these proteins with RcMADS1 showed that the MADS-box and 'K' domain are highly conserved amongst these members of the StMADS11 clade, with the 'I' domain somewhat well conserved as well. The MADS domain (69 amino acids) was identified using the NCBI conserved domain search, while comparison with known MADS-box genes allowed the identification of the K-box consisting of 79 amino acid residues ( Figure S3).

Phylogenetic Analysis of RcMADS1
Phylogenetic analysis of amino acid sequences derived from members of the StMADS11 clade showed that RcMADS1 is nested within the StMADS11 clade, and fell into a clade with AGL24, to the exclusion of SVP, with 83% bootstrap support (Figure 2A). When we used RcMADS1 with only StMADS11, AGL24 and SVP amino acid sequences, it still forms a clade with AGL24 distinct from SVP ( Figure 2B). Phylogenetic analysis of derived amino acid sequences from the conserved MADS-box and K-box also showed that RcMADS1 groups with AGL24 ( Figure  S4A and B). When a more extensive phylogenetic tree was constructed using the conserved MADS-box domain of the StMADS11 clade and representative members of other MADSbox gene clades, RcMADS1 remained in the StMADS11 clade, which shows that despite high sequence similarities, RcMADS1 is closer to AGL24 than to SVP ( Figure S5).

Ectopic Expression of RcMADS1 in Arabidopsis Displayed Abnormal Floral Organs and Early Flowering
In order to understand whether RcMADS1 is functionally related to AGL24 or SVP, we generated transgenic Arabidopsis thaliana lines. Since Rafflesia is a holoparasite that lacks vegetative parts, the use of the heterologous system is the best approach for functional analyses. A total of 23 independent transgenic Arabidopsis lines harboring 35S::RcMADS1 were generated using the floral dip method [46].
The 35S::RcMADS1 plants showed three distinct phenotypes. Five out of the 23 transgenic lines had a 'strong' phenotype where plants produced more inflorescence axes compared to wild-type plants ( Figure 3A and B) and exhibited sepals and petals into leaflike structures bearing conspicuous trichomes ( Figure 3C and D). Additionally, instead of carpels at the center, elongated inflorescence-like structures developed ( Figure 3E and F). The late-formed flowers produced fertile siliques, but containing only a few viable seeds ( Figure S6A, B and C). Four lines had a 'weak' phenotype where only leaf-like structures were formed with trichomes where the sepals should be, but the carpels developed as siliques and set viable seeds in the weak phenotype plants ( Figure S7). The remaining transgenic lines exhibited phenotypes similar to that of wild-type plants. RcMADS1 expression analysis in one line from each phenotype pool by qRT-PCR showed a direct correlation between the level of RcMADS1 expression and phenotypic severities. Expression level was high in plants exhibiting the strong phenotype, and proportionately lower in lines with weak or no obvious changes in phenotype ( Figure 3G and H). Even though RcMADS1 sequence had high similarities to both AGL24 and SVP, the 'strong' phenotype exhibited by plants ectopically expressing RcMADS1 -was similar to that of AGL24 ectopic expression phenotype [27,28] and distinct from SVP ectopic expression [32] in Arabidopsis thaliana ( Figure 3I).
Ectopic expression lines (ELs) also showed an altered flowering time compared to wild-type (WT) plants as determined by rosette leaf numbers at the time of bolting. The ELs displayed an early flowering phenotype and a gradual decrease in the number of rosette leaves according to the phenotype severities. Three of the representative transgenic lines showing weak phenotype (EL2), no change (EL4) or with strong phenotype (EL8) were chosen for further analyses. Even though EL4 plants were similar to the WT they flowered earlier than the WT. The EL2 line flowered earlier than EL4, and EL8 flowered earlier than EL2 plants. The flowering time of ELs were also correlated to the RcMADS1 expression levels and the phenotypic severities ( Figure 4A and B). Thus, the flowering time of EL8 plants resembled the early flowering 35S::AGL24 plants rather than the late flowering 35S::SVP plants ( Figure 4C). The ectopic expression floral phenotypes and flowering time strongly suggested that RcMADS1 is functionally similar to AGL24.

RcMADS1 can Functionally Complement the agl24-1 Mutant and Rescue the FRI Late Flowering Phenotypes
To further verify RcMADS1 function, we made use of the late flowering agl24-1, FRI-containing Col line and the early flowering svp-41 mutants. We generated transgenic plants expressing RcMADS1 under the 35S promoter or AGL24 promoter in these three genotypes by floral dipping. At least three independent lines showing the phenotype were considered for all the complementation analyses. Plants ectopically expressing RcMADS1 in the agl24-1 background flowered earlier than the agl24-1 mutant ( Figure 5A and B). When RcMADS1 is driven by AGL24 promoter, regulation of RcMADS1 expression should be similar to that of native AGL24. Transgenic agl24-1 plants expressing AGL24::Rc-MADS1 also flowered earlier than the agl24-1 mutant ( Figure 5C and D), indicating that RcMADS1 is able to rescue the late flowering phenotype of agl24-1 with both the promoters used. On the other hand, the svp-41 early flowering phenotype was not affected by RcMADS1 ectopic expression and transgenic plants exhibited floral abnormalities such as conversion of sepals and petals into leaf-like structures ( Figure 6A and B). In Arabidopsis, natural allelic variation at the FRI locus is one of the major determinants for flowering time and the dominant alleles of FRI confer late flowering, which can be reversed to early flowering by vernalization [25,47]. The late flowering phenotype of FRI locus also can be rescued by AGL24 overexpression in Arabidopsis [28]. Hence, we wanted to study the functional similarities of RcMADS1 to AGL24 by generating transgenic plants ectopically expressing RcMADS1 in the late flowering FRI-containing Col line. These transgenic plants displayed an early flowering phenotype when compared with the FRI-containing Col line and WT ( Figure 6A and B). Therefore, RcMADS1 could rescue the late flowering phenotype of the FRI-containing Col line, similar to AGL24. These observations further indicate that RcMADS1 functions like AGL24 in Arabidopsis plants.

RcMADS1 can Induce the Expression of SOC1, a Direct Downstream Target of AGL24
Functional complementation analyses in a heterologous system still pose the question of whether the molecular regulation of flowering time by the transgene is similar to that of AGL24. SOC1 is known to be one of the direct downstream targets of both AGL24 and SVP, but AGL24 promotes SOC1 expression while SVP represses it. Furthermore, according to earlier reports changes in SOC1 expression can be detected during floral transition. Therefore, in order to gain insight into the molecular function of RcMADS1, we tested the expression of SOC1 in the 35S::RcMADS1 transgenic line showing the strong phenotype along with WT, 35S::AGL24, and 35S::SVP seedlings as control.
Expression of SOC1 was analyzed by qRT-PCR using RNA from various seedling stages (5 to 11 days after germination, on alternate days). Our qRT-PCR results revealed that SOC1 expression was significantly elevated in 35S::AGL24 and 35S::RcMADS1 plants compared to those in 35S::SVP and WT seedlings 9 days after germination (Figure 7), which corresponds to the floral transition stage of development. This observation further showed that the molecular function may be conserved between RcMADS1 and AGL24 with respect to regulation of floral transition.

Discussion
Rafflesia species are well known for their conspicuous flowers and holoparasitic endophytic nature that lack vegetative body parts such as leaves, roots, and stems [2]. Hence, flowers are the only plant parts amenable for analyses in this plant species. In this study we chose to study MADS-box genes, which are known to be involved in regulating floral organ identity and floral transition [15,16,17,22,28]. Using degenerate primers designed based on the conserved MADS-box region sequences, we could clone a full length MADS-box gene from Rafflesia cantleyi and named it RcMADS1. It showed high level of sequence similarity to the Arabidopsis MADS-box genes AGL24 and SVP. Phylogenetic analysis of RcMADS1 showed that it might be closer to AGL24 than to SVP (Figure 2, Figure S4 and S5). In Arabidopsis, AGL24 is one of the integrators of flowering signals leading to a precise regulation of floral meristem specification [48], which occurs in a dosage-dependant manner [27]. RcMADS1, a putative ortholog of AGL24 from Rafflesia may have a similar function. Similarly, ectopic expression of StMADS11 clade genes from other species in Arabidopsis caused alteration in flowering time and floral morphology, e.g., BM1 from Hordeum vulgare [36] and INCO from Antirrhinum majus [34] in Arabidopsis caused the conversion of sepals and petals to leaf-like organs and the production of secondary inflorescence from axils of carpel. Likewise OsMADS22, Os-MADS47 and OsMADS55 from Oryza sativa ectopically expressed in Arabidopsis showed floral abnormalities [37,39] and WAS206 (MPF2-like MADS-box gene related to AGL24) from Withania ectopic expression displayed early flowering in Arabidopsis [41].
Owing to the highly reduced nature of Rafflesia, we chose to use the heterologous model plant Arabidopsis thaliana for our studies. Transgenic Arabidopsis plants expressing RcMADS1 allowed us to perform functional analyses that would be otherwise impossible to conduct with holoparasitic species. Phenotypic characterization of five independent transgenic lines expressing RcMADS1 showed that plants were early flowering and exhibited conversion of the floral meristem into inflorescence meristems as seen in 35S::AGL24 plants [27,28]. This is distinct from 35S::SVP lines, which are characterized by late flowering and conversion of the inflorescences into chimeric shoot-like structures [48]. Moreover, the effect of RcMADS1 ectopic expression is dosage-dependant, as is the case for AGL24 [27], which was confirmed by the observed range of phenotypic severity seen in the various lines with differing levels of RcMADS1 expression. These observations collectively suggest that RcMADS1 might be a functional ortholog of AGL24 in Rafflesia cantleyi. To further verify this notion, we performed complementation analysis using agl24-1 and svp-41 mutants. Our results showed that RcMADS1 expression could functionally complement the loss of AGL24, but not SVP. Furthermore, we showed that RcMADS1 is able to rescue the late flowering phenotype of FRI containing Col line, similar to the activity of AGL24 [28]. Finally, regulation of expression of a direct downstream target of AGL24, namely SOC1 by RcMADS1 further supports the argument that RcMADS1 is an ortholog of AGL24 from Rafflesia cantleyi. Expression levels of SOC1 were previously shown to be antagonistically regulated by AGL24 and SVP [33,48] and our results showed that RcMADS1 has the same effect on SOC1 expression as AGL24. Taken together, our data show that RcMADS1 is a functional ortholog of Arabidopsis AGL24. The expression of SOC1 in seedlings undergoing floral transition implies that signals necessary for activation of the upstream regulator, namely, AGL24, and the functional ortholog RcMADS1 in this case, are perceived and processed by the vegetative organs of the plant. Despite the fact that in Rafflesia these vegetative organs are absent, the function of RcMADS1 appears to be conserved. It is tempting to speculate that the flowering signals that usually originate from the vegetative parts of the plant e.g., florigen [49,50,51] may be provided by the host plant in holoparasitic species such as Rafflesia. Also, it should be noted that structurally similar proteins may behave differently in heterologous systems. Likewise RcMADS1 may function differently in Rafflesia than in Arabidopsis, which is yet to be studied. Although the rest of the mechanism of regulation of the process is not understood at this time, our study suggests that the molecular regulation of flowering in this species may be well conserved.