Leaf margin phenotype-specific restriction-site-associated DNA-derived markers for pineapple (Ananas comosus L.).

To explore genome-wide DNA polymorphisms and identify DNA markers for leaf margin phenotypes, a restriction-site-associated DNA sequencing analysis was employed to analyze three bulked DNAs of F1 progeny from a cross between a 'piping-leaf-type' cultivar, 'Yugafu', and a 'spiny-tip-leaf-type' variety, 'Yonekura'. The parents were both Ananas comosus var. comosus. From the analysis, piping-leaf and spiny-tip-leaf gene-specific restriction-site-associated DNA sequencing tags were obtained and designated as PLSTs and STLSTs, respectively. The five PLSTs and two STSLTs were successfully converted to cleaved amplified polymorphic sequence (CAPS) or simple sequence repeat (SSR) markers using the sequence differences between alleles. Based on the genotyping of the F1 with two SSR and three CAPS markers, the five PLST markers were mapped in the vicinity of the P locus, with the closest marker, PLST1_SSR, being located 1.5 cM from the P locus. The two CAPS markers from STLST1 and STLST3 perfectly assessed the 'spiny-leaf type' as homozygotes of the recessive s allele of the S gene. The recombination value between the S locus and STLST loci was 2.4, and STLSTs were located 2.2 cM from the S locus. SSR and CAPS markers are applicable to marker-assisted selection of leaf margin phenotypes in pineapple breeding.

Introduction

Pineapple (Ananas comosus) is an important crop in tropical and subtropical countries, as well as in some mild regions. Pineapple is an allogamous plant species and conventionally propagated with propagules, such as suckers, slips, and crowns (Nakasone and Paull 1998). The leaf margins of pineapple can be classified into three phenotypes (Supplemental Fig. 1). The completely spineless leaf type and entirely spiny leaf type are designated as ‘piping-leaf type’ and ‘spiny-leaf type’, respectively. The third, ‘spiny-tip-leaf type’, shows spines at the tip of the leaf and sometimes has spines in the region below the tip separated by a non-spiny leaf margin. The term ‘piping’ originates from the peculiar structure of the leaf margin, which shows a folded narrow portion of the edge of the leaf. On the other hand, spiny-tip leaf and spiny leaf have a sharp leaf edge (Collins and Kerns 1946). Leaf margin phenotype is controlled by two genes, P (piping) and S (spiny tip), which correspond to piping leaf and spiny-tip leaf, respectively. P and S loci are independent of each other, and P is epistatic to S (Collins and Kerns 1946). Therefore, plants with a P locus show piping leaf regardless of the genotype of the S locus. Genotypes of ‘spiny-tip-leaf type’ and ‘spiny-leaf type’ are ppSS/ppSs and ppss, respectively (Collins and Kerns 1946). Because cultivars classified as ‘piping-leaf type’, which are easy to handle, are preferred by producers as well as consumers, in our breeding program, ‘piping-leaf-type’ F1 seedlings are selected for planting. However, at the growth stage suitable for planting, it is difficult to distinguish ‘piping-leaf-type’ plants from ‘spiny-tip-leaf-type’ plants. Thus, the development of a rapid technique for the identification of leaf margin phenotypes is necessary. In pineapple, molecular markers, sequenced characterized amplified regions (SCARs), cleaved amplified polymorphic sequences (CAPSs), simple sequence repeats (SSRs), and expressed sequence tag (EST)-SSRs have been developed (Carlier , Rodríguez , Shoda , Sousa , Wöhrmann and Weising 2011). Furthermore, Carlier et al. (2012) and Sousa have constructed genetic maps of pineapple and mapped the P locus. Their plant materials for mapping analysis were F2 and F1 progeny from a cross of genetically distantly related plants, A. comosus var. comosus × A. comosus var. bracteatus.

Recently, restriction-site-associated DNA sequencing (RAD-seq) markers have been introduced (Baird , Matsumura , Pegadaraju , Yang ). RAD-seq is based on sequencing short fragments from defined positions, flanking restriction enzyme recognition sites in the genome, and counting their frequency. DNA polymorphisms among cultivars or segregating individuals are represented by the presence or absence of these short sequences (tags). RAD-seq using next-generation sequencing technology is applicable to genetically closely related plant resources. Furthermore, alleles of tags linked to agronomically important traits are convertible to co-dominant DNA markers suitable for plant breeding. For instance, bitter gourd (Momordica charantia) is a monoecious plant of the Cucurbitaceae family that has both male and female unisexual flowers. Its gynoecious line is essential as a maternal parent in the breeding of F1 cultivars. A RAD-seq analysis was applied to the bulked genomic DNAs from monoecious or gynoecious F2 plants. From the analysis, GTFL-1 was identified by its linkage to the putative gynoecious locus and converted to a conventional DNA marker using invader assay technology, which is applicable to the marker-assisted selection of gynoecy in bitter gourd breeding (Matsumura ).

In this paper, to develop practical molecular markers linked to leaf margin phenotypes of pineapple, we applied a RAD-seq analysis to an F1 population obtained from a cross between closely related parents, A. comosus var. comosus cultivar ‘Yugafu’ and breeding line ‘Yonekura’. We report RAD-derived markers specific to leaf margin phenotypes and their conversion into SSR or CAPS markers, allowing for the rapid identification of leaf margin phenotypes in pineapple. We further show a linkage analysis of markers in F1 progeny and genotypes of markers in our pineapple accessions.

Materials and Methods

Plant materials

To develop DNA markers associated with leaf margin phenotype, 169 plants of F1 progeny obtained from a cross between cultivar ‘Yugafu’ and breeding line ‘Yonekura’ were used (Table 1). The maternal parent ‘Yugafu’ has a piping-leaf phenotype, and the estimated genotype is PpSs based on knowledge acquired from our breeding program. The paternal parent ‘Yonekura’ has a spiny-tip-leaf phenotype, and the estimated genotype is ppSs. In the F1, 87 plants were ‘piping-leaf type’, and 57 and 25 plants were ‘spiny-tip-leaf type’ and ‘spiny-leaf type’, respectively. Additionally, 31 pineapple accessions, excluding the parents of the F1, six newly released cultivars from the Okinawa Prefectural Agricultural Research Center Nago Branch (OPARC-Nago, Nago 905-0012, Japan), 10 breeding lines from OPARC-Nago, 14 accessions introduced from the USA, Brazil, Taiwan, and Australia, and one related species, Ananas ananassoides, were also used (Supplemental Table 5). All materials were maintained at OPARC-Nago. The leaf margin phenotypes of plant materials used in this study were determined at their mature stage.

Table 1

Summary of pineapple plant materials and phenotype-specific RAD-seq analysis

		Yugafu, maternal plant	Yonekura, paternal plant	F1 progenyb			Chi-test
				P-bulk	St-bulk	S-bulk
F1 progeny used in this study	Number of plants			87	57	25	0.490a
	Leaf margin phenotype	Piping	Spiny-tip	Piping	Spiny-tip	Spiny
	Leaf margin genotype (Expected segregation ratio)	PpSs	ppSs	PpSS : PpSs : Ppss (1 : 2 : 1)	ppSS : ppSs (1 : 2)	ppss (1)
Summary of RAD-seq analysis	No. of unique RAD-tags	1,651,063	1,724,232	1,982,054	1,899,722	1,950,272
	No. of parent-specific unique RAD-tags	313,393	386,562
	No. of total RAD-tags	98,029,077	104,561,020	150,432,871	91,728,445	116,290,663

P-value. Chi-test was carried out with the expected segregation ratio 4 : 3 : 1 (Piping leaf : Spiny-tip leaf : Spiny leaf).

P-bulk: ‘piping-leaf-type’ plants, St-bulk: ‘spiny-tip-leaf-type’ plants, and S-bulk: ‘spiny-leaf-type’ plants.

DNA extraction and preparation of RAD-seq library

Genomic DNAs were isolated from young leaves using a DNeasy Plant Mini Kit (Qiagen, Hilden, Germany). Based on the leaf margin phenotype, DNAs from F1 plants were bulked into three samples. Bulks of ‘piping-leaf type’, ‘spiny-tip-leaf type’, and ‘spiny-leaf type’ were named P-bulk, St-bulk, and S-bulk, respectively, and used for RAD-seq analysis. Aliquots of 1 μg of DNAs from ‘Yugafu’, ‘Yonekura’, and the three bulked samples were independently digested with the six-base-cutting restriction enzyme AseI (New England Biolabs, Beverly, MA, USA) and purified using a QIAquick PCR Purification Kit (Qiagen). AseI-digested DNAs were then independently digested with the four-base cutter NlaIII (New England Biolabs) and purified again. To prepare the library for Hiseq2500 (Illumina Inc., San Diego, CA, USA), double-digested DNAs were applied to the step ‘End Repair’ in the TruSeq DNA LT Sample Prep Kit (Illumina Inc.) protocol. Five different indexed adaptors (A002, A004, A005, A006, and A007) were used for each sample. Adaptor-ligated DNAs ranging in size from 300 bp to 1000 bp were selected using the agarose gel electrophoresis method described in the kit. The five size-selected DNAs formed a RAD-seq library.

Sequencing and RAD-tag analysis

The five RAD-seq libraries (‘Yugafu’, ‘Yonekura’, and the three F1 bulked samples) were pooled. The pooled sample was sequenced by HiSeq2500 (Illumina Inc.) with a paired-end flow cell. The paired-end read length was 100 bp. The resulting data (bcl files) were converted to fastq files for each of the five libraries by configureBclToFastq.pl. Thereafter, sequence reads were assessed for their quality using a FASTX toolkit (http://hannonlab.cshl.edu/fastx_toolkit/commandline.html). Based on quality scores, 30 bases from 3′ ends in each read were trimmed. The filtered reads were designated as RAD-tags and numbers of identical RAD-tags were counted in each sample. RAD-tag sequences and their corresponding RAD-tag counts presented in each sample were listed in a table. The table was deposited in DDBJ and was named Master Table in this study (accession numbers for analyses: DRZ003524, DRZ003525, DRZ006806, DRZ006807, DRZ006808). The Master Table excludes data of RAD-tags showing less than 50 total RAD-tag counts for all samples. To explore RAD-tags linked to the P and S loci, RAD-tags in the Master Table were sorted using ‘Excel’ software. In exploration for P locus, RAD-tags presented in both ‘Yugafu’ and P-bulk, and absented in ‘Yonekura’, St-bulk and S-bulk, were sorted. In S locus, RAD-tags showing null counts in S-bulk were sorted. Sequencing reads (fastq files) and RAD-tag sequences and their counts can be downloaded from the DDBJ Sequenced Read Archive. Accession numbers for submissions are DRA002741, DRA002746, DRA002754, DRA002755, and DRA002756.

Allele exploration of RAD-tags and conversion of RAD-tags to CAPS or SSR markers

P and S gene-specific RAD-tags were designated as PLSTs and STLSTs, respectively. According to the genotypes of the parents of F1, alleles of PLSTs and STLSTs associated with recessive alleles of P and S loci were expected to be in the parents and three bulk samples. Therefore, putative alleles of the PLSTs and STLSTs were extracted from the Master Table using the BLAST program (Altschul ) with the PLST and STLST sequences as queries. RAD-tags were present in the parents and three bulk samples and showed two or less single nucleotide differences, or differences of repetition in SSR obtained by BLAST search were presumed as putative alleles of their corresponding PLSTs and STLSTs. The sequences flanking RAD-tags were extracted from the scaffold sequences of pineapple (cultivar ‘N67-10’) using the BLAST program. The scaffold sequences were constructed by ‘CLC Genomics Workbench version 6. 5. 1’ (CLC-bio, Aarhus, Denmark) using reads from the GS FLX+ system (454 Life Sciences, Roche Diagnostics Corp., USA) and Hiseq2500 (Illumina Inc.). Total length of the scaffold sequences and N50 were 376,903,941 bp and 6,370 bp, respectively. The scaffold sequences in this study are available upon request to Naoya Urasaki. The SSR markers converted from PLST1 and PLST5 were detected by an ABI 3130 DNA sequencer (Shoda ). The size standard used for SSR markers was GeneScanTM-600LIZ (Applied Biosystems, Foster City, CA, USA). The SNPs were converted to CAPS markers, and the CAPS markers were detected using agarose gel electrophoresis (Shudo ). The primer sequences for the amplification of the regions flanking the RAD-tags are listed in Supplemental Table 6. PCR amplification for CAPS and SSR markers was performed in a 25-μL reaction mixture containing 0.25 μL of KOD FX, 12.5 μL of 2× PCR Buffer for KOD FX, 5 μL of 2 mM dNTPs (TOYOBO, Japan), 40 pmol each of forward primer and reverse primer, and 10 ng of genomic DNA. DNA was amplified in 40 cycles at 94°C for 10 sec, 55°C for 10 sec, and 68°C for 1 min, and a final extension of 5 min at 68°C. The sequences of the regions flanking the RAD-tags were sequenced on an ABI 3130 DNA sequencer and deposited in DDBJ (Accession numbers: LC009454-LC009468). The restriction enzymes for the CAPS markers converted from PLST10, PLST16, PLST21, STLST1, and STLST3 were MwoI, BstUI, BsaHI, MfeI, and BslI, respectively (Supplemental Table 6). The PCR products digested by the corresponding enzymes were purified by a QIAquick PCR Purification Kit (Qiagen). The purified products were detected by agarose gel electrophoresis. For the CAPS markers from PLST10, a 3% agarose gel was used. For the other four CAPS markers, a 2% agarose gel was used.

Genotyping of PLST and STLST loci and map construction

To determine the genotypes of the five PLSTs and two STLSTs in the 169 F1 plants as well as the 33 pineapple accessions, including the F1 parents, the corresponding CAPS and SSR markers were used. The PLST and STLST locus genotype data and the leaf margin phenotype data in the F1 were analyzed using JoinMap version 4.0 (van Ooijen 2006) to construct their linkage maps. For map construction of P locus, genotype data and the leaf margin phenotype data in the 169 F1 plants were used. However, for S locus, data on the 57 and 25 F1 plants having spiny-tip and spiny leaves, respectively, were used, because from the leaf margin phenotypes in the F1, the genotype of the S gene in ‘piping-leaf type’ and ‘spiny-tip-leaf type’ of F1 plants could not be estimated by the genotypes of the parents. Therefore, for map construction, the phenotypes ‘spiny-tip leaf’ and ‘spiny leaf’ were converted to ‘D’ and ‘B’, respectively, and the genotypes of STLSTs were converted to ‘A’, ‘H’, and ‘B’ for JoinMap version 4.0. Furthermore, recombination values between the S gene and the STLSTs were also calculated using F1 progeny showing spiny-tipped and spiny leaves. As a control of our mapping analysis, the previously reported marker, SSR_AJ845056 (Carlier , Sousa ), linked to the P locus, was also mapped. The primer sequences for SSR_AJ845056 are listed in Supplemental Table 6. PCR products were labeled with FAM using the post-labeling method (Schuelke 2000) and were also detected using the capillary sequencer.

Annotation of PLSTs and STLSTs

To explore PLSTs and STLSTs encoding genes, BLASTN and BLASTX searches were carried out using the sequences of the RAD-tags as queries. For BLASTN searching, the genome sequence of rice (IRGSP-1.0_genome.fasta), a model plant for monocots including pineapple, was downloaded from the rice annotation project database (http://rapdb.dna.affrc.go.jp/download/irgsp1.html) and used as a database. The RAD-tags showing significant homology to the sequences in the rice genome were listed and applied to a BLASTX search against the non-redundant protein sequences (nr) in GenBank.

Results

RAD-seq analysis of the parents and F1 progeny

To develop DNA markers linked to the leaf margin phenotypes of pineapple, RAD-seq analysis was employed. The RAD-seq libraries from the maternal plant cultivar ‘Yugafu’, the paternal line ‘Yonekura’, and their three F1 bulked samples were sequenced by Hiseq2500 (Illumina Inc.). The NlaIII recognition sites used in the preparation of the RAD-seq library were observed at the 5′ ends in the RAD-tag sequences (Supplemental Fig. 3). A filtered and trimmed read, as described in the Materials and Methods, of 70 bp was designated as a RAD-tag. The RAD-tag counts were calculated for each sample. In total, 98,029,077 and 104,561,020 RAD-tags were obtained from ‘Yugafu’ and ‘Yonekura’, respectively (Table 1). In the F1, 150,432,871 RAD-tags from P-bulk (‘piping-leaf-type’ plants), 91,728,445 RAD-tags from St-bulk (‘spiny-tip-leaf-type’ plants), and 116,290,663 RAD-tags from S-bulk (‘spiny-leaf-type’ plants) were obtained (Table 1, Supplemental Fig. 2). The sums (total RAD-tag count of the five samples) of the 10 most abundant RAD-tags ranged from 535,522 to 201,213 (Supplemental Table 1). To develop DNA markers linked to the P locus, RAD-tags specific to ‘Yugafu’ and the bulked ‘piping-leaf-type’ plants were analyzed and are listed in Supplemental Table 2. These RAD-tags were designated PLSTs. PLSTs having a sum of less than 50 were eliminated. In total, 3,252 unique PLSTs were obtained. PLSTs were expected to be linked to the P locus. RAD-tags showing zero counts in the bulked ‘spiny-leaf-type’ plants were named STLSTs and are listed in Supplemental Table 3. STLSTs having a sum of less than 50 were also eliminated. In total, 37,223 unique STLSTs were obtained. STLSTs were expected to be linked to the dominant S locus.

Allele exploration and conversion of RAD-tags into CAPS and SSR markers

To explore putative alleles and flanking sequences of PLSTs and STLSTs, a BLASTN search was performed as described in the Materials and Methods. Eight RAD-tags were selected and seven of eight were successfully converted to practical DNA markers (Table 2). Among the PLSTs, PLST1 was converted to an SSR marker, designated as PLST1_SSR, using the SSR observed in the PLST1 flanking sequences (Fig. 1, Supplemental Fig. 3). There were three alleles at the PLST1 locus in our plant accessions. PLST1_SSR_407 is an SSR marker and an allele associated with the P locus (Fig. 1). PLST5 was also converted to an SSR marker, designated as PLST5_SSR, using the SSR observed in PLST5 and the allele (Fig. 1, Table 2, Supplemental Fig. 3). There were two alleles at the PLST5 locus. PLST5_SSR_188 is an SSR marker and an allele associated with the P locus. Although a recognition site of the restriction enzyme MwoI was observed in the sequence of PLST10, the allele contained two sites. Using this difference, PLST10 was converted to a CAPS marker, PLST10_CAPS (Fig. 1, Table 2, Supplemental Fig. 3). PLST10_CAPS_106 is a CAPS marker and an allele associated with the P locus. The remaining PLSTs and STLSTs in Table 2 were also converted to CAPS markers, PLST16_CAPS, PLST21_CAPS, STLST1_CAPS, and STLST3_CAPS, using the presence/absence of the recognition sites of the corresponding enzymes (Fig. 1, Table 2, Supplemental Fig. 3). PLST16_CAPS_472 and PLST21_CAPS_510 are CAPS markers and alleles associated with the P locus. STLST1_CAPS_95_212 and STLST3_CAPS_194_251 are CAPS markers and alleles associated with the S locus.

Table 2

PLSTs and STLSTs and their alleles used for conversion to CAPS or SSR markers

ID	RAD-tag sequencea	RAD-tag counts					Sumb
		Yugafu	Yonekura	F1 progenyc
				P-bulk	St-bulk	S-bulk
PLST1	ATCGGTGGTGAAAAGGAGAGTGTGCTGTAGCTCCTGTGGAGAGAGTAAGAGGAGGGGAGAGACGAGGAGT	158	0	322	0	0	480
	ATCGGTGGTGAAAAGGAGCGTGTGCTGTAGCTCCTGTGGAGAGAGTAAGAGGAGGGGAGAGACGAGGAGT	73	168	188	162	231	822
PLST5	CTCCTCTGCAGCAACACCTCCACCGCCACCGCCACCGCCGCTTCGTCCCCTCTCTCCTTCAAACCCAACT	122	0	231	0	0	353
	CTCCTCTGCAGCAACACCTCCACCGCCACCGCCGCTTCGTCCCCTCTCTCCTTCAAACCCAACTCTTCCA	142	294	267	234	305	1242
PLST10	AATACTATGATCGCTTCTTATGCACAGAGCGGGAATTTCATTGATGCACTGAGGATATTGAGCCAAATGA	118	0	173	0	0	291
	AATACTATGATCGCTTCTTATGCACAGAGCGGGAATTGCATTGATGCACTGAGGATATTGAGCCAAATGA	124	219	194	190	283	1010
PLST16	TATAGTGTTGCGTTCTGTCATCCGATTAAAGAGTTGCTCAGCTAGTTGAATTCTTCCCAGACTACCAAAT	129	0	151	0	0	280
	TATAGTGTCGCGTTCTGTCATCCGATTAAAGAGTTGCTCAGCTAGTTGAATTCTTCCCAGACTACCAAAT	16	113	93	40	92	354
PLST21	CATTGAGGCCAGCGATGACACCAGCGTCCTTGGTGGCCTGGCGCTGAGAGTCGTTGAAGTATGCCGGGAC	100	0	145	0	0	245
	CATTGAGGCCAGCGATGACGCCAGCGTCCTTGGTGGCCTGGCGCTGAGAGTCGTTGAAGTAGGCCGGGAC	99	207	183	162	218	869
STLST1	AGGCAGAGAGAACAGGTTTCTGCAAAGCAAAATGTGAGGGACAATTGTAGTAAAATTAGAGCAACTAGAG	144	136	253	128	0	661
	AGGCAGAGAGTACAGGTTTCTGCAAAGCAAAATGTGAGGGACAAATGTAGTAAAATTAGAGCAACTAGAG	59	159	238	81	348	885
STLST3	ACACGACACACCATAAAAAGGGGAATAGAACATCAAAAATGCAAGCTTAATCAAGACTTCAGAAATGTGA	133	130	239	139	0	641
	ACACGACACACCATAAAAAGAGGAATAGAACATCAAAAATGCAAGCTTAATCAAGACTTCAGAAATGTGA	128	123	243	98	305	897

Underlines indicate the polymorphisms between alleles.

Total RAD-tag counts of five samples.

P-bulk: ‘piping-leaf-type’ plants, St-bulk: ‘spiny-tip-leaf-type’ plants, and S-bulk: ‘spiny-leaf-type’ plants.

Fig. 1

Pineapple CAPS and SSR markers converted from piping-leaf gene- and spiny-tip-leaf gene-specific RAD-tags, PLSTs and STLSTs, respectively. PLST1 and PLST5 were converted into the SSR markers PLST1_SSR and PLST5_SSR, respectively. PLST1_SSR and PLST5_SSR markers were detected using an ABI 3130 DNA sequencer. Other RAD-tags were converted into CAPS markers. Agarose gel electropherograms show the CAPS markers. * indicates the markers associated with the P and S loci, and are designated as follows: PLST1_SSR_407; PLST5_SSR_188; PLST10_CAPS_106; PLST16_CAPS_472; PLST21_CAPS_510; STLST1_CAPS_95_212; and STLST3_CAPS_194_251. § indicates the markers associated with recessive alleles of the P and S loci, and are designated as follows: PLST1_SSR_406 or _421; PLST5_SSR_182; PLST10_CAPS_90; PLST16_CAPS_274_198; PLST21_CAPS_265_245; STLST1, STLST1_CAPS_307; and STLST3_CAPS_445. M is a 2-log DNA Ladder Marker (New England Biolabs). Yu and Yo indicate ‘Yugafu’ and ‘Yonekura’, respectively. 1 and 2 are No. 6 and No. 4 in F1 progeny, respectively.

Genotyping of PLST and STLST loci in the F1 population and pineapple accessions

Genotypes of PLSTs and STLSTs in 169 F1 plants and 33 pineapple accessions, including the F1 parents, were scored using the SSR and CAPS markers developed (Supplemental Tables 4, 5). In the PLSTs analyses, the recombination value between the P locus and each marker from a PLST was calculated, and the genetic distance (cM) was also estimated from the F1 genotyping results. As a control for the mapping analysis of the P locus, genotypes of SSR_AJ845056 in the 169 F1 plants were also scored (Supplemental Table 4). Consequently, a genetic map surrounding the P locus could be generated (Fig. 2). The five PLSTs were distributed in the vicinity of the P locus, and the closest marker, PLST1, was located 1.5 cM from the P locus (Fig. 2). SSR_AJ845056 was located 12.2 cM from the P locus. The locus had only three recombinants, cultivars ‘Soft Touch’, ‘Okinawa No. 2’, and ‘Okinawa No. 19’; therefore, PLST1_SSR showed the ability to identify leaf margin phenotypes in the tested pineapple resources (Supplemental Table 5).

Fig. 2

Genetic map of the putative P (left) and S (right) loci in pineapple. In the P locus, the genetic distances and locations between each SSR and CAPS marker and the P locus were calculated from the genotypes of 169 F1 plants from a cross between a ‘piping-leaf-type’ cultivar, ‘Yugafu’, and a ‘spiny-tip-leaf-type’ breeding line, ‘Yonekura’. In the S locus, the genetic distances and locations between each CAPS marker and the S locus were calculated from the genotypes of 57 and 25 F1 plants having spiny-tip and spiny leaves, respectively. The genotypes of these markers are shown in Supplemental Table 4.

Twenty-five ‘spiny-leaf-type’ F1 plants were assessed as homozygotes of STLST1_CAPS_307 and STLST3_CAPS_445 associated with the recessive s allele of the S locus (Supplemental Table 4). Although the genotypes of STLSTs in the ‘piping-leaf-type’ and ‘spiny-tip-leaf-type’ F1 plants could be obtained, their genotypes of the S locus could not be estimated from their phenotypes or the parental genotypes. Therefore, a genetic map surrounding S locus was generated using 57 and 25 F1 plants having spiny-tip and spiny leaves, respectively. Both STLST1 and STLST3 were located 2.2 cM from the S locus. The recombination value was 2.4. In two F1 plants, No. 90 and No. 132, recombination between the S gene and both STLST1 and STLST3 was observed (Supplemental Table 4). In the pineapple accessions tested, four ‘spiny-leaf-type’ cultivars, excluding Ananas ananasoides, were assessed as homozygotes of STLST1_CAPS_307 and STLST3_CAPS_445. In the 13 ‘spiny-tip-leaf-type’ pineapple accessions, the cultivar ‘Yellow Mauritius’ was genotyped as a homozygote of STLST1_CAPS_307 and STLST3_CAPS_445 even though it has a ‘spiny-tip-leaf-type’ phenotype (Supplemental Table 5).

To explore PLSTs and STLSTs encoding genes, BLASTN and BLASTX searches were carried out. The 900 most abundant PLSTs and 500 most abundant STLSTs were applied (Supplemental Tables 2, 3) and 25 RAD-tags were obtained from each PLST and STLST (Table 3). The rate of RAD-tags hitting the rice genome was 2.8% in PLSTs and 5.0% in STLSTs. These RAD-tags were expected to be the partial sequences of genes. Therefore, the RAD-tags were applied to a BLASTX search. From these analyses, for 24 of the 25 PLSTs, deduced proteins could be determined. PLSTs were sparsely mapped on the rice chromosomes, excluding chromosomes 8 and 12. For 19 of the 25 STLSTs, deduced proteins could be obtained. Interestingly, 19 of the 25 STLSTs were specifically mapped over three locations, including locations listed in Table 3, to rice chromosome 9. From this result, the STLST mapped region in rice chromosome 9 was considered to be the repetitive sequence. The 19 STLSTs on chromosome 9 were concentrated in the short pineapple contig_1083 (Supplemental Fig. 4). Furthermore, several STLSTs were overlapped in contig_1083, and the recognition sites of the restriction enzymes used were not observed around overlapped STLSTs.

Table 3

PLSTs and STLSTs encoding genes

ID	Annotation
	Chr.a	Location in chr.	e-valueb	Deduced protein (accession number)	e-valuec
PLST21	chr3	9371444-9371377	8E-17	Hsp70 (AFZ61867.1)	2E-07
PLST22	chr11	1692455-1692400	3E-19	Calcium-dependent protein kinase (BAE98496.1)	3E-07
PLST42	chr4	32137876-32137813	2E-14	Organic cation/carnitine transporter 4-like (XP_006653799.1)	8E-04
PLST96	chr1	36141123-36141069	4E-09	Ribonuclease (EXC30979.1)	4E-07
PLST107	chr4	22083687-22083746	5E-12	Endoglucanase (XP_004499712.1)	3E-05
PLST124	chr3	10152585-10152534	1E-09	GDSL/SGNH-like Acyl-Esterase (EEE52601.1)	6E-08
PLST184	chr2	28760247-28760203	7E-08	Guanosine-3′,5′-bis 3′-pyrophosphohydrolase (XP_004151644.1)	8E-05
PLST271	chr5	22068623-25465220	3E-10	Lysosomal beta glucosidase (XP_003568338.1)	2E-01
PLST274	chr11	28806764-28806833	3E-13	HSP70 (AAL69383.1)	3E-05
PLST328	chr9	37404-37347d	3E-25	Hypothetical protein (WP_009088948.1)	2E-04
	chr2	28721421-28721364	3E-25
PLST353	chr1	35886692-35886754	1E-06	Ankyrin repeat-containing protein (EMT18375.1)	1E-04
PLST406	chr3	34449478-34449541	3E-07	Hsp70 (CAC16169.1)	9E-06
PLST424	chr9	31318-31249d	6E-30	Hypothetical protein (XP_002489102.1)	2E-06
PLST465	chr9	37432-37364d	9E-32	RRNA intron-encoded homing endonuclease (XP_003614389.1)	6E-05
	chr2	28721449-28721381	9E-32
PLST468	chr11	28807895-28807831	5E-15	Hsp70 (AFS65092.1)	1E-05
	chr1	36042675-36042611	5E-15
PLST491	chr9	31054-31123d	6E-30	Hypothetical protein (XP_002489033.1)	2E-06
PLST542	chr1	35539191-35539254	3E-07	Protein TRICHOME BIREFRINGENCE-LIKE 30 (XP_007008936.1)	4E-05
PLST601	chr9	35384-35321d	2E-26	No hits found.
	chr2	28719403-28719340	2E-26
PLST637	chr7	15991271-15991202	7E-11	Transmembrane protein (BAK05888.1)	2E-04
PLST701	chr3	27467999-27467935	3E-10	Calcium dependent protein kinase (AEW09346.1)	7e-03
PLST779	chr6	12816000-12815944	7E-08	Polyprotein (EMT06479.1)	9E-06
PLST806	chr6	21040915-21040981	9E-26	Ethylene-responsive transcription factor (XP_003516798.1)	8E-07
	chr2	7365464-7365530	9E-26
PLST842	chr3	22283654-22283597	2E-08	Uridylyltransferase (EMT11618.1)	1E-04
PLST876	chr9	36614-36545d	2E-08	Hypothetical protein (WP_002300193.1)	2E-02
	chr2	28720633-28720564	2E-08
PLST895	chr9	31506-31575d	3E-25	Expressed protein (XP_002884233.1)	2E-05
STLST6	chr7	13745379-13745315	7E-08	Cellulose synthase (XP_003598145.1)	1E-05
STLST99	chr6	1398633-1398577	2E-17	Receptor-like protein kinase (BAC24825.1)	8E-04
STLST117	chr7	13746324-13746393	7E-11	No hits found.
STLST228	chr9	29916-29975d	5E-24	RRNA intron-encoded homing endonuclease (BAD07869.1)	2E-05
	chr2	28721862-28721921	5E-24
STLST242	chr6	12852135-12852204	4E-06	Thioredoxin reductase 2 (EMT02702.1)	8E-06
STLST265	chr9	36951-37020d	2E-32	RRNA intron-encoded homing endonuclease (XP_004509500)	2E-07
	chr2	28720970-28721039	2E-32
STLST284	chr9	32770-32839d	1E-27	RRNA intron-encoded homing endonuclease (XP_003614385.1)	1.2E-01
	chr2	28716570-28716639	1E-27
STLST287	chr9	30962-31031d	1E-27	ATP-synthase (XP_003627732.1)	9E-05
STLST293	chr9	37472-37406d	9E-26	RRNA intron-encoded homing endonuclease (XP_006380094.1)	8E-06
	chr2	28721489-28721423	9E-26
STLST299	chr9	37462-37393d	1E-27	RRNA intron-encoded homing endonuclease (XP_006380094.1)	9E-08
	chr2	28721479-28721410	1E-27
STLST324	chr9	31315-31383d	2E-17	No hits found.
STLST333	chr9	31389-31459d	2E-11	No hits found.
STLST338	chr9	35609-35678d	6E-30	Hypothetical protein (EEC70730.1)	7E-05
STLST346	chr11	12110635-12110683	3E-10	No hit found.
		12107304-12107352	3E-10
STLST383	chr9	31037-31106d	2E-32	Uncharacterized protein (XP_004154316.1)	1E-06
STLST384	chr9	31388-31447d	1E-09	No hit found.
STLST400	chr9	37472-37407d	3E-25	RRNA intron-encoded homing endonuclease (XP_006405913.1)	3E-06
	chr2	28721489-28721424	3E-25
STLST415	chr9	31601-31668d	5E-24	Hypothetical protein (ELU18102.1)	1E-06
	chr2	28715401-28715468	5E-24
STLST418	chr9	31709-31640d	6E-30	Hypothetical protein (ELU18102.1)	1E-06
	chr2	28715509-28715440	6E-30
STLST434	chr9	29983-30048d	1E-027	Hypothetical protein (XP_002489117.1)	1E-05
STLST439	chr9	30025-30094d	2E-20	Hypothetical protein (ELU05451.1)	2E-05
STLST449	chr9	31413-31473d	5E-15	No hit found.
STLST483	chr9	32779-32710d	2E-32	Senescence-associated protein (ACA30301.1)	4E-08
	chr2	28716579-28716510	2E-32
STLST484	chr9	32042-31973d	2E-32	Senescence-associated protein (XP_003064992.1)	3E-07
	chr2	28715842-28715773	2E-32
STLST490	chr9	32286-32344d	5E-21	Chytochrome P-450 monooxgenase (T02955)	5E-03
	chr2	28716086-28716144	5E-21

Rice chromosome.

e-value obtained from BLASTN search.

e-value obtained from BLASTX search.

Corresponding PLSTs and STLSTs were mapped over three locations, including locations listed, with the identical e-value.

Discussion

In this study, an F1 population obtained by crossing the ‘piping-leaf-type’ cultivar ‘Yugafu’ and the ‘spiny-tip-leaf-type’ breeding line ‘Yonekura’ was used. In F1 progeny, the segregation of leaf margin phenotypes was observed. From the number of plants with each phenotype (Table 1), we confirmed that the estimated genotypes of ‘Yugafu’ and ‘Yonekura’ were correct and the P and S genes, as proposed by Collins and Kerns (1946), control the leaf margin phenotypes in pineapple. Our main objective was to explore DNA polymorphisms linked to leaf margin phenotypes. Thus, a bulked segregant analysis was performed (Giovannoni , Matsumura , Michelmore ). The RAD-seq analysis was applied to the independent bulked DNAs from the ‘piping-leaf’, ‘spiny-tip-leaf’, and ‘spiny-leaf’ types of F1 plants. In this analysis, the choice of an appropriate restriction enzyme was important. In the original RAD-seq analysis, SbfI was used for stickleback (Baird ). For plant RAD-seq analyses, Matsumura surveyed the recognition sites in rice, Arabidopsis, papaya, and cucumber genomes and used PacI or AseI based on the survey. Additionally, the physical DNA shearing procedure was replaced by NlaIII digestion. Using these enzymes, Matsumura identified the RAD-tags specific to gynoecy in bitter gourd (M. charantia). Therefore, AseI and NlaIII were also used in this analysis. From the analysis, over 91 million RAD-tags were obtained from each of the five samples (Table 1). As expected, the NlaIII recognition sites were observed at the 5′ end of the RAD-tag in all the PLST and STLST flanking sequences (Supplemental Fig. 3). The sums (total RAD-tag counts of the five samples) of the 10 most abundant RAD-tags ranged from 535,522 to 201,213 (Supplemental Table 1) and eight were partial sequences of ribosomal RNAs. In general, ribosomal RNA genes (rDNA) in higher plants are known to exist as long tandem repeating units in genomes (Rogers and Bendich 1987). These results suggest that ribosomal RNAs also exist as repeating units in the pineapple genome. In contrast, the sums of PLST1 and STLST1 were 480 and 661, respectively (Table 2). If our sequencing scale had been an order of magnitude smaller, then PLST1 and STLST1 may have been eliminated during the in silico analysis because RAD-tags showing less than 50 RAD-tag counts were removed. These results suggest that our experimental design, bulk analysis, restriction enzyme choices, library preparation, and sequencing scale were appropriate for identifying polymorphisms linked to leaf margin phenotypes.

Five PLSTs were successfully converted to SSR or CAPS markers. Genotyping five PLST loci in the 169 F1 plants with converted markers confirmed their linkage to the P locus. According to the mapping analysis of PLSTs, the closest marker, PLST1, and the most distant marker, PLST5, are located 1.5 cM and 2.7 cM from the P locus, respectively (Fig. 2). The control, SSR_AJ845056, was located 12.2 cM from the P locus. According to previous reports (Carlier , Sousa ), SSR_AJ845056 is located 10.4 cM (Carlier ) and 11.1 cM (Sousa ) from the P locus. Based on these results, our mapping analysis was considered to be conducted appropriately. If the P locus is mapped between PLSTs, then a map-based cloning technique could be used to clone the P gene. Unfortunately, the five markers found here mapped to one side of the P locus. It is unclear whether this result is unexpected. In contrast, Carlier and Sousa constructed genetic maps and mapped the P locus using populations originating from A. comosus var. comosus (cv. Rondon, cloneBR 50) × A. comosus var. bracteatus (Branco do mato, clone BR 20). Although there are closer markers than the PLSTs, PLST1 was considered to be applicable in practical breeding because only three recombinants, ‘Soft Touch’, ‘Okinawa No. 19’, and ‘Okinawa No. 2’, of the PLST1 locus existed in the 33 pineapple accessions (Supplemental Table 5).

Two STLSTs linked to the S locus were also successfully converted to CAPS markers and located 2.2 cM from the S locus. In the genotyping of the 169 F1 plants, only one F1 plant, No. 95, was assessed differently by STLST1_CAPS and STLST3_CAPS (Supplemental Table 4). The genotype of No. 95 was estimated as Ss and SS by STLST1_CAPS and STLST3_CAPS, respectively. The phenotype of No. 95 is piping leaf; therefore, the two genotypes, Ss and SS, are both consistent. Based on this result, although we could not estimate which marker was closest to the S locus, STLST1 and STLST3 were determined to be close to each other. Furthermore, all 25 ‘spiny-leaf types’ of the F1 were assessed as homozygotes of the recessive s allele by both markers, as expected. In the genotyping of pineapple accessions, all four foreign ‘spiny-leaf-type’ cultivars, ‘Bogor’, ‘McgregerST-1’, ‘Tainung No. 9’, and ‘Red Spanish’, were also shown to be the expected genotypes, homozygotes of the recessive s allele of the S locus, by both markers. In the analysis of 13 ‘spiny-tip-leaf-type’ accessions, although ‘Yellow Mauritius’ was assessed as a homozygote of the recessive s allele of the S locus by both markers, the 12 other accessions showed the expected genotypes using both markers. Based on these results, STLST1 and STLST3 were considered to be applicable in practical breeding.

In the annotation, the PLSTs were sparsely mapped across the rice chromosomes, while STLSTs were concentrated in the repetitive region of chromosome 9 (Table 3). Because the 19 STLSTs were concentrated in the short pineapple contig_1083 (Supplemental Fig. 4) and the recognition sites were not observed around STLSTs, contig_1083 was considered to be an immature region of repetitive sequence as in rice chromosome 9. Therefore, the sequences surrounding the S locus in pineapple were considered to resemble the corresponding region in rice chromosome 9. This is significant and useful information for cloning the S gene in pineapple without the complete pineapple genome sequence. Using the BLASTX search with RAD-tag sequences as queries, deduced proteins were successfully obtained (Table 3, Supplemental Tables 2, 3). Two particularly interesting proteins, Protein TRICHOME BIREFRINGENCE-LIKE 30 from PLST542 and cellulose synthase from STLST6, were identified. In Arabidopsis, the homolog of the trichome birefringence-like protein gene is transcriptionally coordinated with the cellulose synthase gene and is important to the formation of crystalline cellulose in trichomes (Bischoff ). Although there is no evidence that Protein TRICHOME BIREFRINGENCE-LIKE 30 from PLST542 and cellulose synthase from STLST6 contribute to leaf margin phenotypes in pineapple, the annotation of RAD-tags produced this significant information to elucidate the mechanism of spine formation in pineapple.

In this report, we identified RAD-tags specific to leaf margin phenotypes and converted them into conventional SSR and CAPS markers. SSR and CAPS marker development is applicable to marker-assisted selection of the desirable leaf margin phenotype of pineapple in breeding.