Related polymorphic F-box protein genes between haplotypes clustering in the BAC contig sequences around the S-RNase of Japanese pear.

Most fruit trees in the Rosaceae exhibit self-incompatibility, which is controlled by the pistil S gene, encoding a ribonuclease (S-RNase), and the pollen S gene at the S-locus. The pollen S in Prunus is an F-box protein gene (SLF/SFB) located near the S-RNase, but it has not been identified in Pyrus and Malus. In the Japanese pear, various F-box protein genes (PpSFBB(-α-γ)) linked to the S-RNase are proposed as the pollen S candidate. Two bacterial artificial chromosome (BAC) contigs around the S-RNase genes of Japanese pear were constructed, and 649 kb around S(4)-RNase and 378 kb around S(2)-RNase were sequenced. Six and 10 pollen-specific F-box protein genes (designated as PpSFBB(4-u1-u4, 4-d1-d2) and PpSFBB(2-u1-u5,) (2-d1-d5), respectively) were found, but PpSFBB(4-α-γ) and PpSFBB(2-γ) were absent. The PpSFBB(4) genes showed 66.2-93.1% amino acid identity with the PpSFBB(2) genes, which indicated clustering of related polymorphic F-box protein genes between haplotypes near the S-RNase of the Japanese pear. Phylogenetic analysis classified 36 F-box protein genes of Pyrus and Malus into two major groups (I and II), and also generated gene pairs of PpSFBB genes and PpSFBB/Malus F-box protein genes. Group I consisted of gene pairs with 76.3-94.9% identity, while group II consisted of gene pairs with higher identities (>92%) than group I. This grouping suggests that less polymorphic PpSFBB genes in group II are non-S pollen genes and that the pollen S candidates are included in the group I PpSFBB genes.

Introduction

Self-incompatibility (SI) is a genetic system that prevents self-fertilization in flowering plants by the recognition and rejection of self-pollen (de Nettancourt, 2001). In the Rosaceae, Solanaceae, and Plantaginaceae families, SI is classified as gametophytic SI (GSI), and is controlled by a single S-locus with multiple S-haplotypes. Each S-haplotype contains two genetically linked genes, the pistil S gene and the pollen S gene, which determine the S-haplotype specificity of the pistil and pollen, respectively (McCubbin and Kao, 2000). The pistil S encodes a ribonuclease known as S-RNase (McClure ; Ishimizu ; Xue ). The RNase activity of S-RNases is essential for rejection of self-pollen, and the degradation of rRNA by S-RNases inside the self-pollen tube results in inhibition of pollen growth (McClure ; Huang ). Thus, it is thought that the self S-RNase inhibits growth of the self-pollen tube via degradation of pollen rRNAs. On the other hand, the identity and function of the pollen S remained unknown for a long time. Recently, F-box protein genes were identified as the pollen S genes by sequence analyses of cosmid and bacterial artificial chromosome (BAC) contigs around S-RNase in Prunus species of the Rosaceae, in Petunia inflata of the Solanaceae, and in Antirrhinum hispanicum of the Plantaginaceae. These F-box protein genes were termed SLF (S-locus F-box) or SFB (S-haplotype-specific F-box protein) (Lai ; Entani ; Ushijima ; Sijacic ). Transformation experiments in P. inflata and analyses of pollen-part self-compatible (SC) mutants in Prunus species provided evidence that SLF/SFB genes are the pollen S genes (Sijacic ; Ushijima ; Sonneveld ; Hauck ; Tsukamoto ; Vilanova ). Generally, F-box proteins function as one of the four major subunits (CUL1, SKP1, RBX1, and F-box) that make up the SCF complex, which regulates protein stability through the ubiquitin–proteasome system (Lechner ). The model for S-RNase degradation proposes that the non-self-interaction between S-RNase and SLF/SFB leads to S-RNase ubiquitylation and degradation by the 26S proteasome (McClure and Franklin-Tong, 2006).

In Rosaceae, the pollen S has been identified only in Prunus (almond, apricots, and cherry), but not in Pyrus (pear) and Malus (apple). The Rosaceae comprises three subfamilies: Rosoideae, Dryadoideae, and Spiraeoideae. Prunus, Pyrus, and Malus are all included in Spiraeoideae (Potter ). Therefore, it is likely that the pollen S genes in Pyrus and Malus are also F-box protein genes. Recently, S-locus-linked and pollen-specific polymorphic F-box protein genes were isolated from apple (Malus×domestica) and Japanese pear (Pyrus pyrifolia), and these have been proposed as good candidates for the pollen S genes. Cheng cloned two S-locus-linked F-box protein genes (MdSLF and MdSLF) from apple by reverse transcription-PCR (RT-PCR) with degenerate primers designed from the conserved SLF/SFB sequences. Sassa found several pollen-specific polymorphic F-box protein genes termed SFBB (S locus F-box brothers) in BAC contig sequences around apple S-RNase genes. These SFBB genes include MdSFBB3-α and MdSFBB3-β around S-RNase, and MdSFBB9-α and MdSFBB9-β around S. Using RT-PCR, they also cloned various PpSFBB genes (PpSFBB-α, PpSFBB-β, and PpSFBB-γ) that are linked to S-RNase genes of the Japanese pear; PpSFBB4-α, PpSFBB4-β, and PpSFBB4-γ are linked to S-RNase, and PpSFBB5-α, PpSFBB5-β, and PpSFBB5-γ are linked to S-RNase. PpSFBB-γ genes that are linked to another eight S-RNase genes have been cloned. They show high amino acid sequence identities (97.5–99.7%) among the 10 S-haplotypes (Kakui ). However, it is not clear whether PpSFBB genes are located near the S-RNase, like MdSFBB genes, or whether they are the pollen S genes. To identify the pollen S genes in the Japanese pear, a previously constructed BAC library from an S homozygote was used and a BAC contig of ∼570 kb around S was assembled. Sequence analysis of the 240 kb spanning 51 kb upstream to 189 kb downstream of S revealed a pollen-specific F-box protein gene (S; S-haplotype F-box protein gene) that differed from PpSFBB4-α–γ. S is located 127 kb downstream of S (Okada ). The SC cultivar ‘Osa Nijisseiki’ (S) is a natural mutant derived from ‘Nijisseiki’ (S). The S-haplotype of ‘Osa Nijisseiki’ lacks the pistil S function but retains the pollen S function, and is termed the S-haplotype, where ‘sm’ stands for ‘stylar-part mutant’ (Sato, 1993). The S-haplotype has a 236 kb deletion, which includes S and S, suggesting that the pollen S allele is located outside of the region spanning 48 kb upstream to 188 kb downstream of S—that is, outside the region that is deleted in the S-haplotype (Okada ).

In this study, the sequence outside of the deleted region in S was analysed, and the 649 kb region from 290 kb upstream to 359 kb downstream of S was determined; six PpSFBB4 genes were found. To evaluate the S-haplotype polymorphism of PpSFBB4 genes, a BAC library was constructed from the Japanese pear cultivar ‘Choujuuro’ (S) to assemble a BAC contig around S. A 378 kb region from 166 kb upstream to 212 kb downstream of S was sequenced, and 10 PpSFBB2 genes were found. Relationships among 36 F-box protein genes of Pyrus and Malus were analysed by comparing their amino acid sequences and by phylogenetic clustering.

Materials and methods

Plant materials

One cultivar and three S homozygotes of the Japanese pear were used: ‘Choujuuro’ (S), and S, S, and S homozygotes. The S and S homozygotes were selected from bud-selfed progeny of ‘Choujuuro’ (S) (Terai ). The S homozygote was segregated from bud-selfed progeny of ‘Nijisseiki’ (S) (Okada ). The leaves, mature pollen, and pistils were frozen in liquid nitrogen, and stored at –80 °C until use.

Construction and characterization of an S BAC library

An S BAC library was constructed and characterized according to the method of Okada . High molecular weight DNA was isolated from leaf tissue (3 g) of ‘Choujuuro’ (S), partially digested with HindIII, and size-selected twice by pulsed field gel electrophoresis (PFGE). In the first size selection, an agarose slice containing DNA fragments of 60–210 kb was excised and embedded into a new 1% SeaPlaque GTG agarose gel (Cambrex, http://www.cambrex.com/). In the second size selection, two size fractions (145–185 kb and 185–205 kb) were recovered by digestion of agarose slices with β-agarase I. DNA from each fraction was separately ligated into HindIII-digested CopyControl pCC1BAC Cloning-Ready vector (EPICENTRE, http://www.epibio.com/) and transformed into Escherichia coli strain TransforMax EPI300 (EPICENTRE). Equal numbers of transformed cells were picked from each fraction and a total of 61 440 colonies were pooled in 64 individual 96-well plates with 12 columns and eight rows (10 colonies per well) and stored at –80 °C. The BAC plasmid was extracted from the randomly chosen BAC clones by the standard alkaline lysis method, digested with NotI, and separated by PFGE. Insert size was estimated by comparison with a PFGE lambda ladder (New England Biolabs, http://www.neb.com/).

Chromosome walking

Chromosome walking in the region around S-RNase was performed by PCR screening of the S BAC library and the previously constructed S BAC library (Okada ). The PCR screening was performed in three consecutive steps as described by Okada . Chromosome walking around the S was initiated by PCR screening of the S BAC library with an S-RNase-specific primer pair, ‘FTQQYQ’ and ‘anti-IIWPNV’ (Ishimizu ).

BAC plasmids were isolated from positive BAC clones using the Plasmid Midi Kit (Qiagen, http://www1.qiagen.com/). Both ends of the BACs (∼600 bp) of the positive clones were sequenced using T7 and RP vector primers, and a primer pair was designed from each BAC-end sequence (Supplementary Table S1 available at JXB online). For chromosome walking, non-repetitive primer pairs were selected from the BAC-end primer pairs located at the outer ends of the contig by PCR amplification of plate pool templates, which were prepared by mixing all 960 BAC clones in each plate. Furthermore, S-specific primer pairs were identified from among the non-repetitive primer pairs by PCR, using genomic DNA of the S and S homozygotes as templates. These S-specific primer pairs were used for PCR screening of the BAC library (Supplementary Table S1). For PCR, genomic DNA was isolated from leaves of the S and S homozygotes by the modified cetyltrimethylammonium bromide (CTAB) method (Castillo ).

To estimate insert sizes and compare restriction patterns, BAC plasmids were digested with NotI and subjected to PFGE. Based on overlapping of the BAC clones, their insert sizes, and restriction patterns, the physical distance was calculated to construct a BAC contig.

BAC subcloning

The plasmids of BAC clones were completely digested with HindIII or EcoRI, separated on a 0.7% agarose gel, and purified from the gels using GENECLEAN Kit III (Qbiogene, http://www.qbiogene.com/). Each fragment was ligated into pBluescriptII SK (+) and transformed into E. coli strain TOP10F’ (Invitrogen, http://www.invitrogen.com/). Inserts from subclones that were smaller than 7 kb were sequenced by primer walking, and those that were larger than 7 kb were sequenced after subcloning using other restriction enzymes. A primer was designed from each insert-end sequence. Using these primers, the regions outside of the subclones in the BAC plasmids were sequenced. The sequences from subclones and the outside sequences were assembled to construct contigs for each BAC clone. Gap regions, for which no sequence data were obtained, were amplified from BAC plasmids by PCR with the Expand High Fidelity PCR system (Roche Diagnostics, http://www.roche-diagnostics.jp/), and directly sequenced.

Nucleotide and amino acid sequence analysis

Nucleotide sequences were determined with the BigDye Terminator v3.1 Cycle Sequencing Kit using an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems, http://www3.appliedbiosystems.com/). The sequence data were analysed using GENETYX-MAC Ver. 13 and ATGC Ver. 4 software packages (Genetyx, http://www.sdc.co.jp/genetyx/). Protein-coding sequences were predicted using the GENSCAN program (Burge and Karlin, 1997). Homology searches were carried out using the BLASTX program (Altschul ). Deduced amino acid sequences were analysed by Pfam (http://pfam.janelia.org/) to search for protein motifs. Amino acid sequences were aligned using ClustalW (http://clustalw.ddbj.nig.ac.jp/top-j.html) and manually optimized. A phylogenetic tree was constructed by the Neighbor–Joining method (Saitou and Nei, 1987). A Harr plot analysis was performed using GENETYX-MAC Ver. 13 software.

RT-PCR

Total RNAs extracted from pollen, pistils, and leaves were subjected to first-strand cDNA synthesis using ReverTra Ace α (TOYOBO, http://www.toyobo.co.jp/). Using the Expand High Fidelity PCR system (Roche Diagnostics), PCR was then carried out with gene-specific primer pairs (Supplementary Table S2 at JXB online). PCR products were separated by electrophoresis on 1.5% or 0.8% agarose gels and visualized by ethidium bromide staining. The PCR products were purified from the gels, and sequenced to confirm gene specificity.

Sequence data

The 649 kb and 378 kb sequences around S and S genes have been deposited in the EMBL/GenBank Data Libraries under accession nos AB545981 and AB545982, respectively.

Results

Sequence analysis of 649 kb around the S gene

Previously, an S BAC contig spanning 202 kb upstream (18F9 T7-end) to 359 kb downstream (33G4 T7-end) of S was constructed. In addition, the complete sequences of two BAC clones (17C7 and 5D3) and the partial sequences of two other BAC clones (32D11 and 4D9) were analysed to determine a 240 001 bp sequence spanning 51 kb upstream to 189 kb downstream of S (Okada ). In this study, 32D11 and 4D9 were sequenced in their entirety. To extend the S BAC contig upstream, chromosome walking was resumed using a non-repetitive BAC-end primer pair (33H11-T7). PCR screening of the S BAC library yielded five BAC clones: 12B8, 15C1, 31C7, 31F1, and 36B6. As a result, chromosome walking from S produced a set of overlapping BAC clones (36B6, 18F9, 32D11, 17C7, 5D3, 4D9, and 33G4) covering ∼649 kb, spanning 290 kb upstream to 359 kb downstream of S (Fig. 1A). Three BAC clones (36B6, 18F9, and 33G4) were subcloned and completely sequenced. The sequence assembly of the seven BAC clones (36B6, 18F9, 32D11, 17C7, 5D3, 4D9, and 33G4) yielded a 648 516 bp sequence spanning 290 kb upstream to 359 kb downstream of the S.

Fig. 1.

Construction of BAC contigs, schematic genomic structures, and locations of (pseudo) F-box protein genes around S (A) and S (B) of the Japanese pear. Names, T7-, and RP-ends of BAC clones are shown in boxes. Black ends represent BAC-ends used for chromosome walking. Hatched boxes indicate BAC clones chosen for complete sequencing. Schematic genomic structures of S- and S-haplotypes are shown below the BAC contigs. The directions of transcription of S-RNase and PpSFBB genes are represented by arrows. Physical distances from S-RNase are indicated in parentheses. A double-headed arrow indicates the 236 kb deleted region in the S-haplotype.

Analysis using GENSCAN software predicted 89 open reading frames (ORFs) in the S BAC 649 kb contig sequence (Fig. 2A). Among the 89 ORFs, 34 (ORF33–ORF66) were included in the 236 kb deleted region of the S-haplotype that spans 48 kb upstream to 188 kb downstream of S (Fig. 2A; Okada ). The other 55 ORFs (ORF1–ORF32 and ORF67–ORF89) were located outside the deleted region. A BLASTX search of the 89 ORFs yielded 61 ORFs with significant similarity (E-value

Fig. 2.

ORF maps of the region around S (A) and S (B). Arrowheads indicate the location and transcriptional direction of genes predicted by GENSCAN software. Open arrowheads indicate genes showing no significant homology to proteins in databases. Grey arrowheads represent transposable elements. Black arrowheads indicate non-transposon-like genes. The 649 kb and 378 kb sequences around S and S have been deposited with the EMBL/GenBank Data Libraries under accession nos AB545981 and AB545982, respectively.

Table 1.

Open reading frames (ORFs) predicted by GENSCAN in the 649 kb region around S4-RNase

ORFs		Homologous protein	Species	Amino acid identity	Score (bits)	E-valuea	Accession no.
ORF1		None
ORF2		Transposon protein	Oryza sativa	64/122 (52%)	134	9e-30	DP000011
ORF3	(SFBB4-u4)	PpSFBB9-γ	Pyrus pyrifolia	291/378 (76%)	566	4e-159	AB297939
ORF4		Putative retroelement pol polyprotein	Arabidopsis thaliana	24/36 (66%)	50.8	4e-05	AC006920
ORF5		GAG-POL precursor	Vitis vinifera	34/124 (27%)	61.2	3e-08	AB111100
ORF6		None
ORF7		Retrotransposon protein	Oryza sativa	45/107 (42%)	79.7	9e-14	DP000009
ORF8		Retrotransposon protein	Oryza sativa	45/158 (28%)	82.4	1e-14	DP000009
ORF9	(SFBB4-u3)	MdSFBB9-β	Malus×domestica	350/390 (89%)	726	0.0	AB270792
ORF10		Retrotransposon protein	Oryza sativa	31/78 (39%)	54.3	6e-08	DP000011
ORF11		None
ORF12		Retrotransposon protein	Oryza sativa	36/82 (43%)	81.6	6e-14	DP000009
ORF13		Hypothetical protein	Vitis vinifera	59/181 (32%)	71.6	3e-11	AM472051
ORF14		Zinc knuckle family protein	Oryza sativa	60/182 (32%)	100	4e-19	DP000010
ORF15	(SFBB4-u2)	MdSFBB9-α	Malus×domestica	300/315 (95%)	590	4e-166	AB270792
ORF16		Retrotransposon protein	Oryza sativa	71/165 (43%)	124	7e-27	DP000009
ORF17		None
ORF18		Predicted protein	Populus trichocarpa	44/134 (32%)	69.7	2e-10	DS017968
ORF19		None
ORF20		Retrotransposon protein	Oryza sativa	31/81 (38%)	60.1	1e-09	DP000086
ORF21		Hypothetical protein	Vitis vinifera	43/132 (32%)	70.5	1e-10	AM426737
ORF22		Retrotransposon protein	Beta vulgaris	61/173 (35%)	82.0	3e-13	EF101866
ORF23		Retrotransposon gag protein	Asparagus officinalis	418/767 (54%)	772	0.0	AC183435
ORF24		Retrotransposon gag protein	Oryza sativa	487/1032 (47%)	921	0.0	AC120534
ORF25	(SFBB4-u1)	MdSFBB3-β	Malus×domestica	306/394 (77%)	628	4e-178	AB270796
ORF26		Hypothetical protein	Vitis vinifera	28/42 (66%)	64.3	4e-08	AM455744
ORF27		None
ORF28		Reverse transcriptase	Vigna radiata	23/37 (62%)	48.5	2e-04	AY684634
ORF29		Retrotransposon protein	Oryza sativa	521/962 (54%)	1028	0.0	DP000011
ORF30		Retrotransposon gag protein	Asparagus officinalis	48/129 (37%)	91.3	2e-16	AC183436
ORF31		Retrotransposon gag protein	Asparagus officinalis	827/1636 (50%)	1568	0.0	AC183435
ORF32		Integrase	Populus trichocarpa	48/149 (32%)	84.0	3e-14	DQ536160
ORF33		Predicted protein	Populus trichocarpa	34/92 (36%)	62.4	3e-08	EQ134071
ORF34		None
ORF35		Retrotransposon gag protein	Asparagus officinalis	95/194 (48%)	172	4e-41	AC183435
ORF36		Retrotransposon gag protein	Asparagus officinalis	117/405 (28%)	141	2e-31	AC183435
ORF37		None
ORF38		None
ORF39		None
ORF40		Transposon protein Pong subclass	Zea mays	31/120 (25%)	60.1	3e-07	EU964924
ORF41		Transposon protein Pong subclass	Zea mays	185/380 (48%)	355	8e-96	EU962682
ORF42	(S4-RNase)	S4-RNase	Pyrus pyrifolia	49/50 (98%)	115	2e-24	AB014072
ORF43		Zinc finger	Medicago truncatula	23/63 (36%)	47.8	6e-04	AC148290
ORF44		None
ORF45		Retrotransposon gag protein	Asparagus officinalis	65/291 (22%)	59.7	5e-07	AC183435
ORF46		Retroelement pol polyprotein-like	Arabidopsis thaliana	667/1331 (50%)	1272	0.0	AB024037
ORF47		Retrotransposon gag protein	Asparagus officinalis	793/1644 (48%)	1496	0.0	AC183435
ORF48		Retrotransposon protein	Oryza sativa	352/728 (48%)	673	0.0	DP000011
ORF49		None
ORF50		Chromosome-associated kinesin KIF4A	Ricinus communis	37/96 (38%)	60.5	2e-07	EQ974117
ORF51		None
ORF52		None
ORF53		Retrotransposon protein	Beta vulgaris	101/201 (50%)	191	7e-46	EF101866
ORF54	(SFBB4-d1)	S4F-box0	Pyrus pyrifolia	400/400 (100%)	834	0.0	AB308360
ORF55		None
ORF56		Hypothetical protein	Vitis vinifera	45/154 (29%)	72.0	2e-11	AM429787
ORF57		Hypothetical protein	Vitis vinifera	50/135 (37%)	81.3	2e-13	AM467140
ORF58		RNase H family protein	Asparagus officinalis	35/63 (55%)	78.6	2e-18	AC183436
ORF59		Retrotransposon protein	Oryza sativa	67/121 (55%)	137	1e-30	DP000009
ORF60		Retrotransposon protein	Oryza sativa	127/297 (42%)	234	2e-59	DP000009
ORF61		Retrotransposon gag protein	Asparagus officinalis	37/129 (28%)	74.7	1e-11	AC183435
ORF62		Unknown protein	Arabidopsis thaliana	51/74 (68%)	102	6e-20	AK117191
ORF63		Retrotransposon gag protein	Asparagus officinalis	119/299 (39%)	195	3e-47	AC183435
ORF64		None
ORF65		Retrotransposon protein	Beta vulgaris	35/44 (79%)	76.3	2e-18	EF101866
ORF66		Hypothetical protein	Vitis vinifera	39/110 (35%)	55.1	5e-06	AM489256
ORF67		RNase H family protein	Asparagus officinalis	58/121 (47%)	115	2e-24	AC183436
ORF68		Retrotransposon gag protein	Asparagus officinalis	229/692 (33%)	341	2e-91	AC183435
ORF69		Hypothetical protein	Vitis vinifera	68/227 (29%)	84.7	3e-14	AM451669
ORF70		None
ORF71		None
ORF72		Retrotransposon gag protein	Asparagus officinalis	714/1299 (54%)	1440	0.0	AC183435
ORF73		Retrotransposon gag protein	Asparagus officinalis	401/608 (65%)	821	0.0	AC183435
ORF74		None
ORF75		Integrase	Populus trichocarpa	212/535 (39%)	357	2e-96	DQ536178
ORF76		Retrotransposon gag protein	Asparagus officinalis	167/506 (33%)	223	1e-55	AC183435
ORF77		None
ORF78		None
ORF79	(SFBB4-d2)	MdSFBB3-β	Malus×domestica	293/393 (74%)	561	4e-158	AB270796
ORF80		Retrotransposon protein	Oryza sativa	175/487 (35%)	225	9e-57	DP000009
ORF81		None
ORF82		None
ORF83		Hypothetical protein	Vitis vinifera	27/46 (58%)	56.2	1e-06	AM482339
ORF84		None
ORF85		Retrotransposon protein	Oryza sativa	242/432 (56%)	471	4e-130	DP000009
ORF86		None
ORF87		None
ORF88		None
ORF89		None

a Significant similarity corresponds to an E-value

Fig. 3.

Alignment of deduced amino acid sequences of PpSFBB4-u1–u4, 4-d1–d2 and PpSFBB2-u1–u5, 2-d1–d5. Amino acid sequences were aligned using ClustalW. Conserved sites and relatively conserved sites are marked with asterisks and dots, respectively. F-box domains and FBA_1 domains of F-box proteins are coloured and underlined, respectively. Accession numbers for the F-box protein genes are as follows: PpSFBB4u1–u4, 4-d1–d2 (AB545981) and PpSFBB2-u1–u5, 2-d1–d5 (AB545982).

Table 3.

Pairwise amino acid sequence identities (%) of PpSFBB4 and PpSFBB2 genes

	PpSFBB4-u2	PpSFBB4-u3	PpSFBB4-u4	PpSFBB4-d1	PpSFBB4-d2	PpSFBB4-α	PpSFBB4-β	PpSFBB4-γ	PpSFBB2-u1	PpSFBB2-u2	PpSFBB2-u3	PpSFBB2-u4	PpSFBB2-u5	PpSFBB2-d1	PpSFBB2-d2	PpSFBB2-d3	PpSFBB2-d4	PpSFBB2-d5	PpSFBB2-γ
PpSFBB4-u1	71.4	70.6	72.7	73.1	71.8	69.6	71.0	66.3	72.3	73.6	73.3	71.2	74.0	72.8	76.3	72.6	76.3	71.5	66.3
PpSFBB4-u2	–	86.2	73.7	67.3	69.7	83.2	67.0	70.8	92.1	87.8	71.5	66.5	68.5	70.3	72.4	72.0	75.1	66.2	70.3
PpSFBB4-u3		–	70.3	67.2	69.5	80.5	67.2	67.4	84.9	93.1	70.1	66.9	68.8	70.8	70.6	72.4	73.7	67.1	66.9
PpSFBB4-u4			–	69.7	68.8	70.9	67.9	77.5	71.2	73.2	68.4	67.1	70.4	70.2	70.4	69.7	72.7	69.4	77.3
PpSFBB4-d1				–	71.2	70.7	69.6	63.6	68.7	68.7	68.8	68.6	72.5	73.9	72.3	82.8	74.3	73.1	63.6
PpSFBB4-d2					–	69.2	67.9	63.7	67.6	70.7	69.7	68.1	71.2	77.4	73.0	74.8	84.2	73.6	63.4
PpSFBB4-α						–	68.7	68.2	81.9	81.6	70.2	68.7	70.7	69.4	70.5	72.5	73.0	69.2	67.7
PpSFBB4-β							–	62.3	66.6	68.4	68.8	94.9	69.2	71.8	69.0	71.3	73.0	70.5	62.1
PpSFBB4-γ								–	68.2	70.0	64.9	62.5	65.5	65.6	65.1	65.1	69.2	65.5	99.0
PpSFBB2-u1									–	86.0	69.7	66.3	69.7	70.5	72.0	70.5	73.3	67.9	67.7
PpSFBB2-u2										–	70.5	67.9	70.7	72.8	72.8	74.1	75.1	70.2	69.5
PpSFBB2-u3											–	69.5	70.7	74.1	71.1	72.3	73.8	71.0	64.9
PpSFBB2-u4												–	68.4	70.8	69.0	70.9	72.5	70.0	63.0
PpSFBB2-u5													–	71.0	74.0	72.8	74.3	71.8	65.5
PpSFBB2-d1														–	72.8	75.4	81.2	79.0	65.4
PpSFBB2-d2															–	74.9	77.8	73.6	64.3
PpSFBB2-d3																–	77.1	73.6	65.1
PpSFBB2-d4																	–	78.7	69.4
PpSFBB2-d5																		–	65.3

Values >90% are shown in bold.

To examine expression of PpSFBB4-u1–u4 and PpSFBB4-d1–d2, total RNA was extracted from pollen, pistils, and leaves of the S homozygote. RT-PCR analyses were conducted using gene-specific primer pairs (Supplementary Table S2 at JXB online). PpSFBB4-u1–u4 and PpSFBB4-d1–d2 were all specifically expressed in pollen, but not in pistils or leaves (Supplementary Fig. S1A). The PpSFBB4-d2-specific primer pair yielded fragments of 1373 bp and 1142 bp, which both were derived from the PpSFBB4-d2 transcript, because the forward primer annealed to the 5' untranslated region (UTR) and the coding region of PpSFBB4-d2 (Supplementary Table S2, Fig. S1A). Thus, in the 649 kb sequence around S-RNase there were six F-box protein genes (PpSFBB4-u1–u4 and PpSFBB4-d1–d2) with pollen-specific expression. The three PpSFBB genes previously shown to be linked to the S, PpSFBB4-α–γ, were not within the sequenced region.

Construction of a BAC contig around the S gene

To analyse the sequence polymorphism of PpSFBB4-u1–u4 and PpSFBB4-d1–d2 in another haplotype, a BAC library was constructed from the Japanese pear cultivar ‘Choujuuro’ (S). The BAC library consisted of two sublibraries derived from two DNA size fractions. One sublibrary, which was derived from the 145–185 kb size fraction, consisted of 30 720 clones with an average insert size of 111 kb. The other sublibrary, which was derived from the 185–205 kb size fraction, consisted of 30 720 clones with an average insert size of 127 kb. The average insert size of the whole BAC library was ∼119 kb. The haploid genome size of pear is estimated to be 496–536 Mb (Arumuganathan and Earle, 1991). Therefore, the BAC library represented ∼14-fold genome coverage, giving a >99% theoretical probability of recovering any single-copy DNA sequences in the genome.

To construct a BAC contig around S, chromosome walking was initiated from S. PCR screening of the BAC library of ‘Choujuuro’ with an S-RNase-specific primer pair yielded 10 BAC clones containing S: 2E10, 5B5, 13C10, 15E3, 21F7, 37F11, 41A10, 48F8, 53E3, and 57B1. These BAC clones were aligned by PCR analysis with primer pairs designed from each BAC-end sequence, and a first contig was constructed based on the insert size and restriction pattern of the BAC plasmids (Fig. 1B).

For chromosome walking, two non-repetitive and S-haplotype specific primer pairs, 13C10-RP and 2E10-T7, were selected from the BAC-end primer pairs located at the outer ends of the first contig (Supplementary Table S1 at JXB online). PCR screening of the BAC library with 13C10-RP yielded two BAC clones (43D6 and 55A9) upstream of S PCR screening of the BAC library with 2E10-T7 yielded five BAC clones (3A1, 7F3, 10H7, 27C4, and 52C1) downstream of S. Finally, chromosome walking from S yielded a total of 17 BAC clones. These were aligned to construct a BAC contig of ∼391 kb spanning 166 kb upstream to 225 kb downstream of S (Fig. 1B).

Sequence analysis of 378 kb around the S gene

To identify the genes around S, three overlapping BAC clones, 55A9, 48F8, and 10H7, were subcloned and completely sequenced (Fig. 1B). Sequence assembly of the three BAC clones yielded a 378 419 bp sequence. Analysis using GENSCAN software predicted 57 ORFs in the 378 kb region (Fig. 2B). A BLASTX search of these ORFs yielded 41 ORFs with significant similarity (E-value NBS-LRR-type disease resistance proteins, and ORF53 to a cyclin-like F-box. Ten ORFs (ORF3, ORF7, ORF8, ORF16, ORF19, ORF24, ORF31, ORF43, ORF46, and ORF49) showed high sequence similarity to MdSFBB genes, MdSLF genes, PpSFBB4-β, or S. Using GENETYX-MAC Ver. 13 software, the predicted ORFs were reanalysed to determine the precise ORFs from the start (ATG) to the stop codon. ORF3, ORF7, ORF8, ORF16, ORF19, ORF24, ORF31, ORF43, ORF46, and ORF49 encoded 393, 396, 394, 392, 392, 394, 395, 400, 393, and 390 amino acid residues, respectively. A Pfam motif search predicted that these proteins had an F-box domain at the N-terminus and an FBA_1 domain in the centre (Fig. 3). When compared with reported PpSFBB, MdSFBB, and MdSLF genes, these ORFs showed pairwise deduced amino acid sequence identities ranging from 62.0% to 94.9% (Supplementary Table S3 at JXB online). The F-box protein genes differed from PpSFBB2-γ, which was reported to be linked to S (Kakui ). Thus, these ORFs were assigned as new PpSFBB2 genes. ORF3, ORF7, ORF8, ORF16, and ORF19 were located ∼146, ∼108, ∼102, ∼53, and ∼22 kb, respectively, upstream of S. ORF24, ORF31, ORF43, ORF46, and ORF49 were located ∼10, ∼56, ∼119, ∼137, and ∼158 kb, respectively, downstream of S. These new PpSFBB2 genes upstream and downstream of S were named PpSFBB2-u and PpSFBB2-d5, respectively, and lower case numbers were assigned to the PpSFBB2-u and PpSFBB2-d located close to S. Therefore, ORF19, ORF16, ORF8, ORF7, and ORF3 were designated as PpSFBB2-u1, PpSFBB2-u2, PpSFBB2-u3, PpSFBB2-u4, and PpSFBB2-u5, and ORF24, ORF31, ORF43, ORF46, and ORF49 were designated as PpSFBB2-d1, PpSFBB2-d2, PpSFBB2-d3, PpSFBB2-d4, and PpSFBB2-d5, respectively. These PpSFBB2 genes around S shared variable transcriptional orientations (Fig. 1B). Using ATGC Ver. 4 software, the 378 kb S BAC contig sequence was searched for SFBB-like sequences. The analysis revealed two pseudogenes (ΨPpSFBB2-u1 and ΨPpSFBB2-u2) encoding truncated F-box proteins that were located ∼46 kb upstream and ∼159 kb upstream of S, respectively (Fig. 1B). PpSFBB2-u1–u5 and PpSFBB2-d1–d5 shared 66.3–86.0% amino acid sequence identity with each other, and showed 66.2–93.1% identity with PpSFBB4-u1 and PpSFBB4-d1–d2 (Table 3).

Table 2.

Open reading frames (ORFs) predicted by GENSCAN in the 378 kb region around S2-RNase

ORFs		Homologous protein	Species	Amino acid identity	Score (bits)	E-valuea	Accession no.
ORF1		Serine-threonine protein kinase	Ricinus communis	262/411 (63%)	506	2e-141	EQ974075
ORF2		Retrotransposon protein	Oryza sativa	189/479 (39%)	355	1e-95	DP000011
ORF3	(SFBB2-u5)	S2-locus F-box	Malus×domestica	280/312 (89%)	602	9e-170	DQ422811
ORF4		DNA glycosylase DEMETER	Arabidopsis thaliana	351/1051 (33%)	385	1e-110	DQ335243
ORF5		None
ORF6		DNA glycosylase	Populus trichocarpa	196/291 (67%)	365	4e-99	CM000346
ORF7	(SFBB2-u4)	PpSFBB4-β	Pyrus pyrifolia	376/396 (94%)	752	0.0	AB270798
ORF8	(SFBB2-u3)	S1-locus F-box	Malus×domestica	369/394 (93%)	747	0.0	DQ422810
ORF9		Retrotransposon protein	Oryza sativa	78/248 (31%)	104	2e-20	DP000010
ORF10		Retrotransposon protein	Oryza sativa	132/267 (49%)	249	8e-64	DP000010
ORF11		None
ORF12		None
ORF13		Retrotransposon protein	Oryza sativa	38/82 (46%)	83.6	3e-14	DP000086
ORF14		GAG-POL precursor	Vitis vinifera	65/208 (31%)	105	2e-20	AB111100
ORF15		None
ORF16	(SFBB2-u2)	MdSFBB9-β	Malus×domestica	369/391 (94%)	768	0.0	AB270792
ORF17		Retrotransposon protein	Oryza sativa	332/820 (40%)	570	4e-160	DP000011
ORF18		Retrotransposon protein	Oryza sativa	96/180 (53%)	184	5e-44	DP000011
ORF19	(SFBB2-u1)	MdSFBB9-α	Malus×domestica	361/392 (92%)	694	0.0	AB270792
ORF20		Retrotransposon protein	Oryza sativa	269/693 (38%)	466	8e-129	DP000009
ORF21		Retrotransposon protein	Beta vulgaris	41/57 (71%)	89.0	9e-16	EF101866
ORF22	(S2-RNase)	S2-RNase	Pyrus pyrifolia	191/191 (100%)	410	5e-112	AB014073
ORF23		None
ORF24	(SFBB2-d1)	MdSFBB3-α	Malus×domestica	366/394 (92%)	735	0.0	AB270795
ORF25		Hypothetical protein	Vitis vinifera	22/36 (61%)	46.2	3e-09	AM426737
ORF26		Retrotransposon protein	Oryza sativa	34/68 (50%)	80.9	4e-13	DP000009
ORF27		None
ORF28		Retrotransposon protein	Beta vulgaris	25/38 (65%)	54.3	4e-06	EF101866
ORF29		None
ORF30		None
ORF31	(SFBB2-d2)	MdSFBB3-β	Malus×domestica	304/394 (77%)	637	0.0	AB270796
ORF32		None
ORF33		Hypothetical protein	Vitis vinifera	69/195 (35%)	88.2	9e-16	AM483001
ORF34		None
ORF35		Hypothetical protein	Vitis vinifera	49/180 (27%)	60.8	2e-07	AM423348
ORF36		None
ORF37		None
ORF38		Retroelement pol polyprotein-like	Arabidopsis thaliana	151/243 (62%)	238	1e-60	AB024037
ORF39		Retrotransposon gag protein	Asparagus officinalis	72/152 (47%)	146	5e-33	AC183435
ORF40		TIR-NBS-LRR-type disease resistance protein	Populus trichocarpa	109/213 (51%)	204	3e-51	DQ513203
ORF41		LTR retrotransposon like protein	Arabidopsis thaliana	148/283 (52%)	283	4e-74	AL022140
ORF42		TIR-NBS-LRR-type disease resistance protein	Populus trichocarpa	66/90 (73%)	137	3e-31	DQ513203
ORF43	(SFBB2-d3)	S4F-box0	Pyrus pyrifolia	331/400 (82%)	681	0.0	AB308360
ORF44		None
ORF45		Retrotransposon protein	Oryza sativa	145/328 (44%)	249	2e-63	DP000011
ORF46	(SFBB2-d4)	MdSFBB3-β	Malus×domestica	366/392 (93%)	772	0.0	AB270796
ORF47		Putative retroelement polyprotein	Arabidopsis thaliana	388/919 (42%)	652	0.0	AC018460
ORF48		None
ORF49	(SFBB2-d5)	MdSFBB3-β	Malus×domestica	309/388 (79%)	647	0.0	AB270796
ORF50		Retrotransposon protein	Oryza sativa	178/526 (33%)	233	1e-58	DP000010
ORF51		Polyprotein 1	Petunia vein clearing virus	67/284 (23%)	66.2	7e-09	AY228106
ORF52		Retrotransposon protein	Oryza sativa	145/350 (41%)	255	2e-65	DP000011
ORF53		Cyclin-like F-box	Medicago truncatula	41/89 (46%)	88.2	2e-16	AC150889
ORF54		None
ORF55		Hypothetical protein	Prunus persica	42/91 (46%)	92.8	6e-17	DQ863257
ORF56		Retrotransposon protein	Beta vulgaris	64/86 (74%)	139	7e-31	EF101866
ORF57		None

a Significant similarity corresponds to an E-value

Total RNA was extracted from pollen, pistils, and leaves of the S homozygote to examine the expression of PpSFBB2-u1–u5 and SFBB2-d1–d5. RT-PCR analyses were conducted using gene-specific primer pairs (Supplementary Table S2 at JXB online). PpSFBB2-u1–u5 and PpSFBB2-d1–d5 were all specifically expressed in pollen, but not in pistils or leaves (Supplementary Fig. S1B). Thus, in the 378 kb sequence around S-RNase there were 10 F-box protein genes (PpSFBB2-u1–u5 and PpSFBB2-d1–d5) with pollen-specific expression. The PpSFBB2-γ gene, previously shown to be linked to the S, was not within the sequenced region.

Comparison of deduced amino acid sequences between the PpSFBB and PpSFBB genes

The pairwise deduced amino acid sequence identities of nine PpSFBB4 genes (PpSFBB4-u1–u4, PpSFBB4-d1–d2, and PpSFBB4-α–γ) and 11 PpSFBB2 genes (PpSFBB2-u1–u5, PpSFBB2-d1–d5, and PpSFBB2-γ) were compared within and between haplotypes (Table 3). Sequence identity among the PpSFBB4 genes ranged from 62.3% to 86.2%, and among PpSFBB2 genes ranged from 63.0% to 86.0%. The PpSFBB4 and PpSFBB2 genes showed 62.1–99.0% identity between haplotypes. Identities of >90% were found between PpSFBB4-u2 and PpSFBB2-u1 (92.1%), PpSFBB4-u3 and PpSFBB2-u2 (93.1%), PpSFBB4-β and PpSFBB2-u4 (94.9%), and PpSFBB4-γ and PpSFBB2-γ (99.0%). Identities ranging from 80% to 90% were found between PpSFBB4-u2 and PpSFBB2-u2 (87.8%), PpSFBB4-u3 and PpSFBB2-u1 (84.9%), PpSFBB4-d1 and PpSFBB2-d3 (82.8%), PpSFBB4-d2 and PpSFBB2-d4 (84.2%), PpSFBB2-u1 and PpSFBB4-α (81.9%), and PpSFBB2-u2 and PpSFBB4-α (81.6%). The other 89 pairwise comparisons showed identities of <80%.

Phylogenetic analysis of the F-box protein genes of Pyrus and Malus

Most PpSFBB4 and PpSFBB2 genes cloned in this study shared the highest amino acid sequence identities with the F-box protein genes of Malus (MdSFBB and MdSLF genes), although PpSFBB4-u4 and PpSFB2-u4 showed the highest identities with PpSFBB3, 4, 9-γ (77.5%) and PpSFBB4-β (94.9%) derived from the same species, respectively (Supplementary Table S3 at JXB online). The deduced amino acid sequences of the 36 F-box protein genes of Pyrus and Malus were aligned with PmSLFS7 of P. mume using ClustalW, and a rooted phylogenetic tree was constructed by the Neighbor–Joining method with PmSLFS7 as an outgroup (Fig. 4). F-box protein genes of Pyrus and Malus did not form taxa-independent clusters, and several PpSFBB genes were positioned closest to MdSFBB and MdSLF genes. The F-box protein genes of Pyrus and Malus were grouped into two major groups: group I (84% bootstrap value) and group II (91% bootstrap value). Group I included PpSFBB4-u1, 4-d1–d2, PpSFBB2-u3–u5, 2-d1–d5, PpSFBB-β genes, MdSFBB3 genes, and MdSLF genes, while group II included PpSFBB4-u2–u4, PpSFBB2-u1–u2, PpSFBB-α genes, PpSFBB-γ genes, and MdSFBB9 genes. Comparing group I with group II, amino acid sequences were conserved in F-box domains, but were divergent in the five regions designated as R1, R2, R3, R4, and R5. In these regions, sequences and/or insertions/deletions (indels) were relatively conserved within each group (Fig. 3).

Fig. 4.

Phylogenetic analysis of the F-box protein genes of Pyrus and Malus, and Japanese apricot PmSLFS7. The phylogenetic tree was constructed using the Neighbor–Joining method. PmSLFS7 was used as an outgroup. Numbers besides the branches are bootstrap values >50%. The bar under the tree represents the number of amino acid substitutions per site.

Discussion

The results of a previous study suggested that the pollen S allele is distal to the region from 48 kb upstream to 188 kb downstream of S (Okada ). In this study, the BAC contig around S was extended to 659 kb, and a 648 516 bp region spanning 290 kb upstream to 359 kb downstream of S was sequenced. Sequence analysis of the 649 kb region predicted five new pollen-specific F-box protein genes (PpSFBB4u1–u4 and PpSFBB4-d2). The 649 kb sequence around S included six PpSFBB4 genes including PpSFBB4-d1 (S), but not PpSFBB4-α–γ. In addition, a BAC library was constructed from ‘Choujuuro’ (S), and a BAC contig of 391 kb around S was assembled. Sequence analysis of a 378 419 bp region spanning 166 kb upstream to 212 kb downstream of S predicted 10 new pollen-specific F-box protein genes (PpSFBB2-u1–u5, 2-d1–d5). The 378 kb sequence around S included 10 PpSFBB2 genes, but not PpSFBB2-γ. The predicted products of PpSFBB4-u1–u4, 4-d1–d2, and PpSFBB2-u1–u5, 2-d1–d5 showed typical features of F-box proteins: an F-box domain at the N-terminus and an FBA_1 domain in the centre (Fig. 3). These results indicated that F-box protein genes with pollen-specific expression are clustered around the S-RNase of Japanese pear, and that PpSFBB4-α–γ and PpSFBB2-γ, which are linked to the S-RNase, were located outside the sequenced region.

Organization of the F-box protein gene cluster around the S-RNase gene of Japanese pear

Among PpSFBB4-u1–u4, 4-d1–d2 and PpSFBB2-u1–u5, 2-d1–d5, some genes may be located more distantly from S-locus regions. Entani conducted pattern matching analysis of homologies (Harr plot analysis) for the sequences around PmS and PmS of P. mume. Their results revealed that highly divergent S and Slocus regions are surrounded by co-linear flanking regions, and that S and Slocus regions are ∼27 kb and 15 kb long, respectively. Harr plot analysis of the 649 kb and 378 kb sequences around S and S was conducted, and no co-linearity was found between these sequences (data not shown). This result suggests that both the 649 kb and 378 kb sequences are a part of the S-locus region, or that either sequence could contain both the S-locus region and its flanking region. Sequence analysis of the 649 kb and 378 kb regions predicted 40 and 20 transposon-like sequences around S and S, respectively (Tables 1, 2). The S-locus, which controls S-RNase-based GSI, contains many transposon-like sequences. For example, transposon-like sequences were found in three out of 12 ORFs in 72 kb of the P. dulcis S-haplotype (Ushijima ), in four out of 11 ORFs in 64 kb of the A. hispanicum S-haplotype (Lai ), and in 31 out of 50 ORFs in 328 kb of the P. inflata S-haplotype (Wang ). These transposon-like sequences generate polymorphisms among S-haplotypes, and might contribute to suppression of recombination between S-RNase and SLF/SFB. In the sequenced regions around S and S-RNase, the non-co-linearity, the abundant (retro) transposon insertions, and the absence of PpSFBB4-α–γ and PpSFBB2-γ suggest that the 649 kb and 378 kb sequences around S and S are part of the S-locus region, and that the S-locus regions of the Japanese pear are probably larger than those of Prunus species.

The organization of the F-box protein gene clusters around the S and S was compared when S and S2-RNase were fixed in the same transcriptional orientation (Fig. 1). PpSFBB4-u1 and PpSFBB4-d1 are located ∼113 kb upstream and ∼127 kb downstream of S, whereas PpSFBB2-u1 and PpSFBB2-d1 are located close to S (∼22 kb upstream and ∼10 kb downstream of S, respectively). The average densities of F-box protein genes were one gene/108 kb around S and one gene/38 kb around S. Together, these results suggest that F-box protein genes are clustered in the region around S more tightly than in the region around S

F-box protein genes, SLF/SFB and SLF-like genes (SLFL), were identified in cosmid and fosmid contigs around the S-RNase of Prunus species. SLF/SFB genes are the pollen S genes, but SLFL genes are probably not involved in SI recognition (Entani ; Ushijima ). SLF/SFB and SLFL1–SLFL3 cloned from the same haplotypes show low amino acid sequence identity with each other. For example, PmSLFS7 is 11.7–16.9% identical to PmSLFL1S7, PmSLFL2S7, and PmSLFL3S7, which share 26.9–45.3% identity with each other (Entani ; Matsumoto ). PdSFB and PdSFB are 18.7 and 20.2% identical to PdSLF and PdSLF (orthologuess of PmSLFL1 of P. mume), respectively (Ushijima ). In contrast to Prunus species, PpSFBB4-u1–u4, 4–d1–d2 and PpSFBB2-u1–u5, 2-d1–d5 shared 67.2–86.2% and 66.3–86.0% identity within each haplotype, respectively (Table 3). This indicates that the region around an S-RNase of the Japanese pear comprises related F-box protein genes, which is different from the F-box protein gene organization around the S-RNases in Prunus, in which there are clusters of F-box protein genes that show low levels of identity to each other. The amino acid sequence identities between PpSFBB4-u1–u4, 4-d1–d2 and PpSFBB2-u1–u5, 2-d1–d5 ranged from 66.2% to 93.1% (Table 3), and were higher than those within each haplotype (66.3–86.2%). These similarities between haplotypes indicated that related polymorphic F-box protein genes between haplotypes were clustered in the regions around S and S.

Classification of PpSFBB genes based on phylogenetic analysis and sequence polymorphism

In Prunus species, F-box protein genes around S-RNase genes were grouped into two major classes, the SLF/SFB clade and the SLFL clade, by a phylogenetic analysis (Matsumoto ). SLF/SFB genes show lower levels of allelic sequence identity (77.8–81.3% for PmSLF genes, 68.4–76.4% for PdSFB genes, and 75.1–81.1% for PavSFB genes, respectively) than SLFL genes (88.5–92.0% for PmSLFL1, 95.8–98.6% for PmSLFL2, and 95.1% for PdSLF) (Entani ; Ushijima ; Ikeda ; Matsumoto ). The sequence differences of the F-box protein genes among haplotypes implied that SLF/SFB genes with lower levels of identity were pollen S candidates, and that SLFL genes with high levels of identity were not (Entani ; Ushijima ).

The phylogenetic relationships and sequence differences of F-box protein genes of Pyrus and Malus would be useful for delineating pollen S candidates from PpSFBB4-u1–u4, 4-d1–d2 and PpSFBB2-u1–u5, 2-d1–d5. Phylogenetic analysis based on the deduced amino acid sequences of 36 F-box protein genes of Pyrus and Malus allowed them to be classified into two major groups, I and II (Fig. 4). PpSFBB4-u2–u4, PpSFBB2-u1–u2, PpSFBB genes, PpSFBB-γ genes and MdSFBB9 genes were classified into group II, and the other PpSFBB, MdSFBB3, and MdSLF genes were in group I. The phylogenetic analysis also generated a PpSFBBγ subgroup and 10 gene pairs of PpSFBB genes and PpSFBB/Malus F-box protein genes. Sequence identities between the paired genes ranged from 76.3% to 96.4% (Table 3, Supplementary Table S3 at JXB online), which were higher than those among PpS, PpS, PpS, MdS, MdS, MdS, and MdS (60.9–71.1%). Group I consisted of gene pairs with high levels of identity (>91%): MdSLF2/PpSFBB2-u5 (91.3% identity), PpSFBB4-β/PpSFBB2-u4 (94.9% identity), MdSLF1/PpSFBB2-u3 (93.9% identity), MdSFBB3-β/PpSFBB2-d4 (93.1% identity), MdSFBB3-α/PpSFBB2-d1 (92.9% identity); and gene pairs with low levels of identity: PpSFBB4-u1/PpSFBB2-d2 (76.3% identity) and PpSFBB4-d1/PpSFBB2-d3 (82.8% identity). Group II consisted of the PpSFBB-γ subgroup sharing 97.5–99.7% identity among 10 haplotypes (Kakui ), and gene pairs with high levels of identities (>92%): PpSFBB5-α/PpSFBB4-α (96.4% identity, Sassa ), PpSFBB4-u3/PpSFBB2-u2 (93.1% identity), and PpSFBB2-u1/PpSFBB4-u2 (92.1% identity). The gene pairs with low levels of identity were included in group I, not in group II, suggesting that pollen S candidates were included in the group I.

Therefore, the group I F-box protein genes from the region around S-RNase with low levels of sequence identity, PpSFBB4-u1/PpSFBB2-d2 and PpSFBB4-d1/PpSFBB2-d3, are expected to be pollen S candidates of Japanese pear. In a previous study, PpSFBB4-d1 (S) was thought unlikely to be the pollen S allele, because it is found in the deleted region of the S-haplotype (Okada ). Interestingly, Saito observed that S pollen is rejected by pistils harbouring not only the S-haplotype, but also the S-haplotype. It seems, therefore, that S pollen has a dual specificity for S4-RNase and S1-RNase, which have high amino acid identity (90.0%) (Ishimizu ). This dual specificity is probably due to the loss of PpSFBB4-d1, and S pollen might come to recognize S1-RNase. Therefore, PpSFBB4-d1 might also be a pollen S candidate. However, SLF genes of A. hispanicum and P. inflata share high levels of amino acid identity among haplotypes (97–99% and 88.4–89.7%, respectively; Zhou ; Sijacic ). There is no evidence for a co-evolutionary relationship between SLF/SFB and S-RNase in A. hispanicum and P. inflata, or in Prunus species, which implies that sequence polymorphism between haplotypes can no longer be considered a reliable diagnostic feature of pollen S genes, and functional analysis must be used to identify pollen S genes (Newbigin ). Therefore, all PpSFBB4 genes and PpSFBB2 genes should be considered as pollen S gene candidates.

However, it is not a reasonable interpretation that all PpSFBB genes act in concert as the pollen S genes. Among several F-box protein genes around the S-RNases of Prunus species, A. hispanicum, and P. inflata, one F-box protein gene, SLF/SFB, functions as the pollen S gene; the other F-box protein genes, SLFL genes, are non-S pollen genes (Entani ; Ushijima ; Zhou ; Wang ; Hua ). Therefore, several PpSFBB genes in a particular haplotype are probably not pollen S genes. The non-pollen S proteins of P. inflata, PiSLFLs, either fail to interact with S-RNase or interact much more weakly than PiSLF. When the deduced amino acid sequences of PiSLF and PiSLFLs were compared, three PiSLF-specific regions (SR1, SR2, and SR3) that confer on PiSLF its unique function in SI were revealed (Hua ). Although the interactions of PpSFBBs with S-RNase have not yet been analysed, five regions (R1, R2, R3, R4, and R5) were identified where amino acid sequences were variable between the group I and II F-box proteins (Fig. 3). The sequence differences in these regions might account for different interactions with S-RNase between the group I and II F-box proteins. Therefore, there remains the possibility that the less polymorphic group II F-box protein genes are non-pollen S genes.

Supplementary data

Supplementary data are available at JXB online.

Expression of PpSFBB genes located around S (A) and S (B).

Primer pairs used to construct BAC contigs around S-RNase.

Gene-specific RT-PCR primer pairs.

Pairwise amino acid sequence identities (%) of PpSFBB genes with previously reported PpSFBB, MdSFBB, and MdSLF genes.