Role of MADS-box Gene in the Development of Flower Organs in Pineapple of a Typical Collective Fruit

Background: MADS-box genes play crucial roles in plant vegetative and reproductive growth, especially in inorescences, ower, and fruit. Pineapple is a typical collective fruit, and a comprehensive analysis of the MADS-box gene family in the development of oral organs of pineapple is still lacking. Results: In this study, the whole-genome survey and expression proling of the MADS-box family in pineapple were introduced. Forty-four AcMADS genes were identied in pineapple, 39 of them were located on 18 chromosomes and ve genes were distributed in ve scaffolds. Twenty-two AcMADS genes were dened as 15 pairs of segmental duplication events. Syntenic analysis showed that pineapple is closely related to monocotyledon plants. Most members of the type II subfamily of AcMADS genes had higher expression levels in oral organs compared with type I subfamily, thereby suggesting that AcMADS of type II may play more crucial roles in the development of oral organs of pineapple. Six AcMADS genes have signicant tissue-specicity expression, thereby suggesting that they may participate in the formation of one or more oral organs. Conclusions: Our ndings not only benet to reveal the functional characterization of MADS-box genes in the oral organ development of pineapple but also provide additional information for further understanding the formation and development collective fruit.


Background
The MADS-box transcription factors are one of the important families in higher plant [1] that play fundamental roles during plant development and oral organ differentiation [2]. The prominent feature of the MADS-box proteins is its MADS domain that consists of 56-58 amino acids [3]. The MADS domain can recognize the CArG-box with similar 10-bp A/T-rich DNA sequences [4]. In plants, MADS-box genes can be classi ed into two distinct groups, namely, types I and II, on the basis of the evolutionary relationships: type I members are SRF-like in plant, which only have a MADS (M) domain; type II MADSbox proteins are MEF2-like in plant, animal, and yeast, which contain a highly conservative DNA-binding domain (M), an intervening (I) domain, a semiconservative K domain, and a C-terminal region [5][6]. The type I proteins were further divided into Mα, Mβ, and Mγ subfamilies, and the type II group also were de ned as MIKC-type proteins, which comprise MIKC C -and MIKC*-type proteins [7][8].
Flowering is a complex process that requires the cooperation and interaction of numerous genes.
Previous reports have shown that the MADS-box genes can regulate the characteristics of oral meristems [9][10]. The ABCDE model completely explained the individual development of plant owers and the determination of the identity of oral organs [11][12]. A genes determined the sepal development, petal development required A and B genes, stamen development needed B and C genes to work together, carpel development was ascertained by C genes, and ovule development was identi ed by C and D genes [12]. Class E genes need to assist other genes to participate in the determination of all ower organs and meristems [13]. In Arabidopsis, almost every gene from this model, such as A (APETALA1) [14], B (PISTILLATA, AP3) [15], C (AGAMOUS) [16], D (AGAMOUS-LIKE 11) [17], and E ( SEPALLATA 1,2,3,4) [18], belong to the type II subfamily, thereby suggesting that type II MADS-box genes play a vital role in the control of oral organ development.
Pineapple is one of important tropical fruits with great economic and research value [19][20][21]. This fruit is a typical collective fruit, and each pistil forms a separate little fruit, which is gathered on the enlarged torus. At present, few studies have been conducted on the morphological and physiological basis of collective fruits, especially the molecular mechanism. The physiological basis and molecular mechanism of the formation of collective fruits in pineapple should be understood. Previous studies had shown that the MADS-box genes are involved in various physiological processes, especially with the identi cation of oral organs. Thus, the exploration of the gene function of MADS-box and the molecular mechanism of pineapple ower organ development has received considerable interest.
In the present study, the comprehensive analysis, including the chromosomal localization, synteny analysis, and gene duplication, was investigated on the basis of the pineapple genome. Global expression analysis of MADS-box genes in different tissues and oral organs has been conducted by using RNA sequencing (RNA-Seq) and quantitative real-time polymerase chain reaction (qRT-PCR) to identify the speci c MADS-box genes involved in the different biological processes. Six AcMADS genes demonstrated signi cant tissue speci city in different oral organs. These initiatives provided a reference for the functions of MADS-box genes in pineapple.

Characteristics of Pineapple Flower
After the in orescence of pineapple emerges, its in orescence comes out from among leaf clumps and looks like a pine cone. The entire in orescence development to bloom is approximately 1 month. The capitulum of pineapple is composed of 50 to 200 orets. The blossoming order of the capitulum is from the bottom to the top (Fig. 1A) [22]. The pineapple owers are hermaphrodite and consist of three sepals, three petals, ve stamens, and one pistil (Fig. 1B). The edible part of pineapple involves the eshy axis of in orescence and the ovary of orets.

Identi cation and Classi cation of MADS-box Genes in Pineapple
The MADS-box protein sequences were used to Hidden Markov Model (HMM) search, and 54 candidate genes were originally obtained. The sequence analysis indicated that ten candidate MADS-box genes containing incomplete MADS-box domains were removed. Forty-four MADS-box genes were selected and annotated in pineapple (Supplementary Table S1). Two maximum likelihood trees (ML) were further constructed on the basis of the full-length sequence alignment of all AcMADS genes together with grape and Arabidopsis to provide a reference for the evolutionary relationship of the MADS-box family in pineapple (Fig. 2). The 44 AcMADS genes of the pineapple can be divided into two categories, namely, type I (16) and type II (28). In the rst ML tree, type I AcMADS genes were further classi ed into three subclasses: Mα, Mβ, and Mγ ( Fig. 2A). One MIKC*-type and 27 MIKC C -type genes were showed in the second ML tree, and the MIKC C -type genes were further classed into 11 major groups (Fig. 2B). These groups were named as follows: SVP (SHORT VEGETATIVE PHASE), AGL12, SEP, AGL6, AP1, FLC (FLOWERING LOCUS C), SOC1 (SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1), AG, PI/AP3, TT16 (TRANSPARENT TESTA16), and ANR.

Chromosomal Location and Duplication Analysis of Pineapple MADS-box Genes
Forty-four MADS-box genes were unevenly mapped to the 18 linkage groups (LGs) and ve scaffolds and named from AcMADS1 to AcMADS44 according to their order on the LGs. The largest number of six genes (13.64%) was found in LG01, and the other LGs contained less than three AcMADS genes. Among the 44 AcMADS genes, 16 genes of type I were distributed on 10 chromosomes and three scaffolds, and 28 genes of type II were mapped to 16 chromosomes and two scaffolds.
Tandem duplication event refers to a region of chromosomes within 200 kb containing two or more genes [23]. The gene replication events of the MADS-box family found that one pair of genes (AcMADS17/AcMADS18) underwent tandem repeat events within the AcMADS gene family on LG7 (Fig.  3). These results showed randomness and nonuniform distribution of the AcMADS family in pineapple.

Syntenic Analysis of Pineapple AcMADS Genes
The segmental and tandem duplication events of the AcMADS genes family were identi ed to test the duplication effect in pineapple. In addition to the above-mentioned two tandem duplication events, twenty-two AcMADS genes were clustered into 15 segmental duplication events by BLASTP and MCScanX methods (Fig. 4, Supplementary Table S2). We found many copies of the segmental duplicated gene pairs from the same group, such as AcMADS08/15.AcMADS08/34 and AcMADS14/27 were from the ANR and SVP groups, respectively. AcMADS3/12 and AcMADS3/35 were from the AG group, AcMADS1/16 were from the AP1 group, and AcMADS38/39 were from SOC1. These AcMADS genes showed highly orthologous relationship, thereby indicating that they were obtained by gene duplication and may be from a common ancestor.
The comparative syntenic maps between pineapple with other ve representative species (Arabidopsis, grape, banana, rice, and maize) were performed to further derive the origin and evolutionary mechanisms of pineapple MADS family (Fig. 5). Twenty-four AcMADS genes displayed syntenic relationship with those in banana, followed by maize (23, 52%), rice (22, 50%), grape (10, 32%), and Arabidopsis (6, 13%) through the whole genome-wide comparative analysis. The numbers of orthologous pairs between pineapple and other ve species (banana, rice, maize, grape, and Arabidopsis) were 42, 42, 40, 14, and 12, respectively (Supplementary Table S3). Many collinear gene pairs were only found in monocots but not in dicots, thereby suggesting evolutionary difference between dicotyledonous and monocotyledonous plants. The mutual collinear pairs involving three AcMADS genes were identi ed between pineapple and all ve other species. This nding indicates that these orthologous pairs may be derived from the same ancestor, and duplication occurred before species divergence. These pairs may participate in the evolution of MADS family.

Tissue-speci c Expression Patterns of AcMADS Genes in Pineapple
MADS-box genes were reported to participate in plant organ development, especially oral organ speci cation [2]. The expression patterns of 44 pineapple AcMADS genes in different tissues were obtained from the transcriptome data. The results showed that the transcriptional abundance of MADS genes in pineapple greatly varied in all detected samples (Fig. 6). The genes with high expression were mainly concentrated in type II subfamily. Twenty-six AcMADS genes (59%) were highly expressed in the ower of pineapple. Meanwhile, 14 AcMADS genes (32%) were highly expressed in the fruit of pineapple. Sixteen genes (36%) were expressed at low levels or not expressed in pineapple root, bud, leaf, ower, and fruit (Supplementary Table S4). Multiple of AcMADS genes were speci cally expressed in ower and fruit. For example, ve AcMADS genes (AcMADS5/10/16/21/24) showed high transcript abundance in ower and fruit, two genes (AcMADS06/13) only demonstrated high transcript abundance in owers, and two genes (AcMADS36/4) only presented high expression in the roots (Fig. 6A). These implied MADS-box genes might participate in the ower development of pineapple, which was in line with previous reports [24].
The expressions of all 44 AcMADS genes in different oral organs were investigated to further verify the potential functions of the AcMADS genes in the formation of oral organs (Fig. 6B). Thirty-eight AcMADS genes exhibited relatively high expression in one or more ower organs of pineapple, and fteen of them were highly expressed. Most AcMADS genes with high expression in oral organs originated from the type II subfamily, thereby suggesting that AcMADS genes of type II may play crucial roles in the development of oral organs of pineapple contrast with type I (Fig. 6B). Twenty-eight type II AcMADS genes were further performed by qRT-PCR to validate the RNA-seq results (Fig. 7). The qRT-PCR results showed that the expression pro les of most AcMADS genes were consistent with RNA-Seq. Many AcMADS genes showed obvious tissue speci city. For example, AcMADS12 only expressed in pistil, AcMADS03, AcMADS31, AcMADS33, and AcMADS35 had speci cally high expression in the ovary, and AcMADS08 only expressed in the stamen (Fig. 8).

Co-expression network of key AcMADS genes
Plant owering is a complex physiological process that requires multiple genes to work together.
Studying gene co-expression networks contributes to the exploration of the potential functions of genes. The qRT-PCR and RNA-seq results indicated that six AcMADS genes with speci c expressions in the oral organs of pineapple were selected to construct an interaction network through the String Protein Interaction Database (https://string-db.org/). A total of 129 protein pairs with interactions were detected. These interaction proteins were mainly involved in the oral organ developmental genes and oral induction, including AP1, CO, WUS, SEU, AP2, UFO, and LFY (Fig. 9). In the co-expression network diagram, AcMADS3 interacted with 10 known proteins with the largest number of interacting proteins. AcMADS35, AcMADS12, AcMADS31, and AcMADS33 interacted with 2, 4, 6, and 8 known proteins, respectively.
AcMADS8 only interacted with one known protein. These results will be bene cial to future research and verify its biological function on the basis of relevant experiments.

Discussion
Pineapple is a typical collective fruit, and research about its formation mechanism is still lacking. The MADS-box genes play crucial roles in the development of oral organs and had been investigated in multiple species [25][26][27][28][29]. However, previous studies only analyzed some functions of AcMADS genes in the CAM photosynthesis of pineapple [30]. In this study, 44 genes with typical MADS domains were de ned as MADS-box family genes, which is inconsistent with previous research [30]. The number of MADS-box in pineapple (44) is much lower than that of Arabidopsis (106) [26], rice (75) [27], poplar (105) [28], apple (147) [29], and grape (54) [30]. The lack of pan-grass ρ whole-genome duplication (WGD) event and only having the σ-WGD event during pineapple evolution may affect the amount of AcMADS genes [31].
The AcMADS family of pineapple is classi ed into two categories, namely, types I and II; the type II genes are further divided into 11 subfamilies (Fig. 2B). However, the classi cation and number of type II AcMADS are slightly different among different species. The genome-wide duplication events are common in angiosperm evolution and generally result in the expansion of gene families [32]. Type II MADS genes originate from the whole genome replication, while type I genes are mainly duplicated by small-scale and recent duplications [33]. In our studies, 22 of the 44 MADS-box genes are associated with segmental duplication events, which is higher than that of the Arabidopsis [26]. AcMADS genes had more collinear gene pairs in monocotyledonous plants than dicotyledonous plants, thereby indicating that pineapple is closely related to monocotyledonous plants. The exploration of the genes related to ower development and owering remains to be a major research topic. MADS-box genes play a major role in determining the identity of oral organs [34], and most type II genes are extensively expressed in the reproductive organs and lowly expressed in vegetative organs [35]. In this study, 38 AcMADS genes are expressed in one or more tissues, and 13 of them are specially expressed in owers. The type II of AcMADS genes are highly expressed compared with type I in ower organs (Fig. 6B), thereby indicating that the type II MADS genes might play more important roles in the ower development of pineapple. This nding is consistent with previous reports [34]. The ABCDE model is a classical model of plant ower development [36,37]. In Arabidopsis, AP1 act as a gene for ower meristem and organ morphology to promote the development of petals and sepals [13,38]. In pineapple, two homologous genes of class A genes (AcMADS1 and AcMADS16) are identi ed. AcMADS1 and AcMADS16 are highly expressed in petals and sepals, respectively (Fig. 7), which is consistent with previous reports. The main function of class B genes (AP3 and PI) is to determine the development of the second round of petals and the third round of stamens in Arabidopsis [36]. Two AP3-like (AcMADS41 and AcMADS22) and one PI-like (AcMADS13) genes from the B class genes are identi ed in pineapple and had similar expression patterns (Fig. 7). The class B genes participate in the second and third rounds of oral organ formation of pineapple consistent with Arabidopsis. These genes might also participate in the formation of the fourth round of pistils in pineapple.
AG was a typical class C gene and essential for the identi cation of stamens and carpels [39]. Four AcMADS genes of class-C were detected in pineapple (Fig. 2B). AcMADS3, AcMADS12, and AcMADS35 were speci cally highly expressed in the pistil, and AcMADS24 showed high expression in the stamen and pistil (Fig. 7). AcMADS3, AcMADS12, AcMADS3, and AcMADS35 were also clustered into the segment replication events (Fig. 4). These results showed that these paralogous genes of class C genes could derive from gene duplication and have similar functions in the development of stamen and pistil of pineapple. The homologous genes of AG were involved in the development of stamen and carpel in pineapple, which was consistent with previous reports [40]. The SEP genes, which are functionally important "adhesives" for A, B, C, and D genes, form higher order MADS-box protein complex multimers with other genes and play an effective role in protein-protein interactions [41,42]. Genetic and molecular studies had shown that class E genes (SEP1/2/3/4) had an obvious redundant function in ower development and were necessary to determine all four whorls of the ower organs [41,42]. In pineapple, two SEP-like genes (AcMADS21 and AcMADS5) were identi ed (Fig. 2B), and they were highly expressed throughout the oral meristem (Fig. 6B).
Six genes had signi cantly speci c expression in one or two oral organs of pineapple (Fig. 8). AcMADS3, AcMADS12, and AcMADS35 belong to the AG subgroup, while AcMADS8, AcMADS31, and AcMADS33 belong to the ANR, TT16, and FLC subgroups (Fig. 2B). AcMADS8 was speci cally expressed in the stamens; however, homologous gene AtAGL16, which negatively regulated owering transition through FLOWERING LOCUS T (FT), was not found in stamens [43]. This nding indicated that AcMADS8 played a novel and key role in the development of pineapple stamens. TT16 (AT5G23260.2) was involved in the developmental regulation of the endothelium, which is essential for ovule development [44]. AcMADS31, a homologous gene of TT16, was speci cally expressed in the ovary of pineapple, manifested that AcMADS31 was involved in the development of ovary, and was essential for the formation of female gametophyte of pineapple. AcMADS33, which was homologous with FLC that is owering repressor in Arabidopsis [45], was also speci cally expressed in the ovary of pineapple. However, the involvement of this gene in pistil development is yet to be reported. These AcMADS genes could directly or indirectly determine the formation of pineapple oral organs and provide a reference for further exploring the molecular mechanism of pineapple ower formation.

Plant Materials and Treatments
The pineapple plants (Ananas comosus L. cv. Comte de Paris) used in this study were grown in South Subtropical Crop Research Institute, Zhanjiang, China (21°10′2″N; 110°16′34″E). The different tissues of pineapple, including the bud, ower, fruit, leaf, and root, were collected, immediately frozen in liquid nitrogen, and stored at −80 °C until further use. The oral organs of the pineapple ower, including petals, ovary, stamens, sepals, and stylet, were collected. All treated tissue samples were frozen in liquid nitrogen as quickly as possible and stored at −80 °C.
Total RNA Isolation and qRT-PCR The total RNA was extracted from the pineapple tissues with the RNA extraction kit (Huayueyang, China) according to the manufacturer's instructions. The concentration and quality of all puri ed RNA were checked on a 1% agarose gel and Bio Photometer Plus (Eppendorf, Germany). RNA (5 μg) was reverse transcribed to cDNA with the Revert Aid First-Strand cDNA Synthesis Kit (Thermo Fisher Scienti c, USA).
The quantitative RT-PCR assays were conducted in the Light Cycler 480 II (Roche, Switzerland) by using SYBR Green qPCR Master Mixes (Thermo Fisher Scienti c, USA). AcActin gene was used as the internal control of pineapple. The reaction mixture included 5 μL of 2× SYBR Green PCR Master Mix (Applied Biosystems), a diluted cDNA template of 1 μL, and 1 μL of each primer in a nal volume of 10 μL. The PCR conditions were as follows: 50 °C for 2 min, 95 °C for 2 min, 45 cycles of 15 s at 95 °C, 56°C for 15 s, and 72°C for 40 s. The 2 −ΔΔCt [46] method was used to calculate the relative expression levels of each gene. All primers were designed by the Primer Premier 5.0 [47] and listed in Additional le 4.

Database Search and MADS-box Gene Family Identi cation in Pineapple
The nucleotide and protein sequences of AtMADS genes were searched and obtained from TAIR (http://www.arabidopsis.org/) databases. We downloaded the pineapple genome and proteome sequences from the Pineapple Genomics Database (http://pineapple.angiosperms. org/pineapple/html/index.html) [21]. The genome data of grape, banana, rice, and maize were downloaded from Ensemble plants database (http://plants.ensembl.org/index.html). We downloaded the HMM le keeping with the MADS domain (PF00319) from the Pfam protein database (http://pfam.xfam.org, Pfam 31.0) and searched for the MADS-box genes in the pineapple genome database through HMMER 3.0. The e-value lower than 0.01 and the default parameters were selected. The MADS-box core sequences were con rmed by using the SMART database and the NCBI CDD web server (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). Sequences without MADS-box domain will be deleted. The length of sequences, molecular weight, and isoelectric point (PI) of the MADS proteins were obtained by using the compute pI/Mw tool in the ExPASy server (http://web.expasy.org/protparam/). The subcellular localization of the identi ed MADS proteins was predicted by the cello web server (http://cello.life.nctu.edu.tw/).

Phylogenetic Tree Construction and Classi cation in Pineapple MADS-box Genes
MADS-box genes of Arabidopsis and grape were used as a reference to classify the MADS-box genes of pineapple. A single alignment of pineapple MADS domain by using the Clustal W program was built in MEGA6.0 software; a phylogenetic tree was then constructed by using maximum likelihood (ML) method [48] with the following parameters: 1000 bootstrap replications, partial deletion, and Jones-Taylor-Thornton (JTT) + gamma distributed (G) model.

Chromosomal Location, Gene Duplication, and Syntenic Analysis
The physical positions of the AcMADS genes on chromosomes were identi ed with TBtools [49] according to the gene location in the pineapple genome. The tandem duplication events were de ned as the single chromosomal region contiguous homologous genes with the original repeat, while the duplicate of the whole blocks of genes between different chromosomes was de ned as segmental duplication [50]. Gene duplication events were drafted with Multiple Collinearity Scan tool kit (MCScanX) [51]. In the syntenic analysis, the genome data of ve representative species were downloaded from Ensemble plants database, and the diagrams were visualized using the TBtools with Dual Systeny Plotter.

Conclusions
In this study, the evolution and functional differentiation of the MADS-box genes of pineapple were comprehensively analyzed, and the expression pro le of the AcMADS genes in the oral organ were proposed. The type II AcMADS genes played crucial roles in the development of oral organs of pineapple. AcMADS3\12\35 of the AG subfamily and AcMADS31 of the TTI6 subfamily were highly related to ovary development and pistil formation. AcMADS8 of the ANR subfamily controlled the formation of stamens. Thus, these genes can be identi ed as candidate genes for vector construction and further functional analysis. These genes provided resources for exploring the regulation network of pineapple owering and references for the genetic improvement of transgenic crops and traditional breeding.  Phylogenetic analysis of type I (A) and type II (B) MADS-box genes in Arabidopsis, grape, and pineapple.
The phylogenetic trees were constructed using the ML method. The blue, dark green, and green triangles represent the MADS-box proteins from the grape, pineapple, and Arabidopsis, respectively. MADS-box proteins from the grape with the pre x "GSVIVT" indicate "VvMADS" and "At" means "AtMADS" in Arabidopsis.     Six tissue-speci c AcMADS genes related to pineapple oral organ development. The blue module represents relatively low expression in oral organs. The red module represents relatively high expression in oral organs.