Heterochromatic and genetic features are consistent with recombination suppression of the self-incompatibility locus in Antirrhinum

Authors

  • Qiuying Yang,

    1. Laboratory of Molecular and Developmental Biology, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences and National Center for Plant Gene Research, Beijing 100101, China,
    2. Graduate University of Chinese Academy of Sciences, Beijing 100049, China
    Search for more papers by this author
  • Dongfen Zhang,

    1. State Key Laboratory of Plant Genomics and Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China, and
    2. Graduate University of Chinese Academy of Sciences, Beijing 100049, China
    Search for more papers by this author
  • Qun Li,

    1. Laboratory of Molecular and Developmental Biology, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences and National Center for Plant Gene Research, Beijing 100101, China,
    Search for more papers by this author
  • Zhukuan Cheng,

    1. State Key Laboratory of Plant Genomics and Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China, and
    Search for more papers by this author
  • Yongbiao Xue

    Corresponding author
    1. Laboratory of Molecular and Developmental Biology, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences and National Center for Plant Gene Research, Beijing 100101, China,
    Search for more papers by this author

(fax +86 10 62537814; e-mail ybxue@genetics.ac.cn).

Summary

Self-incompatibility (SI) is a genetic mechanism to prevent self-fertilization that is found in many species of flowering plants. Molecular studies have demonstrated that the S-RNase and SLF/SFB genes encoded by the single polymorphic S locus, which control the pollen and pistil functions of SI in three distantly related families, the Solanaceae, Scrophulariaceae and Rosaceae, are organized in a haplotype-specific manner. Previous work suggested that the haplotype structure of the two genes is probably maintained by recombination suppression at the S locus. To examine features associated with this suppression, we first mapped the S locus of Antirrhinum hispanicum, a member of the Scrophulariaceae, to a highly heterochromatic region close to the distal end of the short arm of chromosome 8. Both leptotene chromosome and DNA fiber fluorescence in situ hybridization analyses showed an obvious haplotype specificity of the Antirrhinum S locus that is consistent with its haplotype structure. A chromosome inversion was also detected around this region between A. majus and A. hispanicum. These results revealed that DNA sequence polymorphism and a heterochromatic location are associated with the S locus. Possible roles of these features in maintenance of the haplotype specificity involved in both self and non-self recognition are discussed.

Introduction

Self-incompatibility (SI) systems are widespread genetic mechanisms that prevent self-fertilization and are thought to play an important role in the diversification and dominance of angiosperms (Whitehouse, 1950; de Nettancourt, 2001). There are three known major types of SI systems, the Papaveraceae, the Brassicaceae and the Solanaceae systems, which have been extensively studied. In these systems, the SI responses are all controlled by a single polymorphic locus, termed the S locus (Kao and Tsukamoto, 2004; Wheeler et al., 2003). In the Papaveraceae, a pistil-specific protein encoded by the S locus mediates pollen-tube growth inhibition through a Ca2+ signaling cascade involving programmed cell death (Franklin-Tong et al., 2002; Thomas and Franklin-Tong, 2004). In the Brassicaceae, the two S-locus proteins known as SRK (S-locus receptor-like kinase), which is expressed in the stigma, and SCR (S-locus cysteine-rich)/SP11(S-locus pollen 11), which is expressed in pollen, have been shown to control pollen recognition and rejection through a ligand–receptor binding mechanism (Chookajorn et al., 2004; Kachroo et al., 2001; Takayama et al., 2001). In the Solanaceae-type gametophytic self-incompatibility (GSI) system found in three plant families, Rosaceae, Solanaceae and Scrophulariaceae, the S-locus product in the pistil is a glycoprotein with ribonuclease activity called S-RNase, whereas the S-locus pollen product is one of a class of F-box proteins called SLF (S-locus F-box) or SFB (S-haplotype-specific F-box) (Kao and Tsukamoto, 2004). The SLF male determinant physically interacts with its haplotype-specific female determinant, S-RNase, probably forming an SCF complex to control pollen rejection (Qiao et al., 2004a,b; Sijacic et al., 2004; Huang et al., 2006). The tight linkage of the two S-locus genes is possibly maintained by recombination suppression, which could be involved in guaranteeing the haplotype-specific interaction of their products (Cui et al., 1999; Wheeler et al., 2003).

However, it is still unclear how recombination suppression occurs around the S locus in the species studied. Chromosome fluorescence in situ hybridization (FISH) analysis revealed that the S locus of Papaver rhoeas was probably located in the centromeric region of a chromosome (Wheeler et al., 2003). For the solanaceous species, the S locus was also found to be located in the centromeric region by genetic analysis and direct FISH localization in Petunia, Nicotiana and Lycopersicon (Bernacchi and Tanksley, 1997; Bernatzky, 1993; Brewbaker and Natarajan, 1960; Entani et al., 1999; Golz et al., 2001; ten Hoopen et al., 1998; Pandy, 1965). For these species, a centromeric location is considered as an important factor in S-locus recombination suppression (Entani et al., 1999; ten Hoopen et al., 1998; Wheeler et al., 2003). However, the S locus of the rosaceous almond was deduced to lie in a region outside the centromere that lacks repetitive sequences (Ushijima et al., 2001).

The S locus of the scrophulariaceous Antirrhinum was previously located in a peri-centromeric region using mitotic metaphase chromosome FISH analysis (Ma et al., 2003). FISH analysis using mitotic metaphase chromosomes has a very low resolution, far lower than for meiotic pachytene chromosomes or extended DNA fibers (Cheng et al., 2002). It is necessary to detect the precise chromosomal location of the Antirrhinum S locus using high-resolution FISH techniques to provide new cytological clues to the mechanism of recombination suppression in this organism. As a closer relative of the Solanaceae than the Rosaceae, the S-locus location of Antirrhinum might provide important information for recombination suppression and the evolutionary conservation of the S locus in the S-RNase-based SI systems.

In this study, several TAC (transformation-competent artificial chromosome) clones derived from the region of the S locus (Zhou et al., 2003) were used as FISH probes. Pachytene and leptotene chromosomes, as well as DNA fibers, were employed to detect the chromosomal location of the S locus, to examine S-locus haplotype specificity and to measure the distance between the two S-determinant genes, respectively. Our goal is to elucidate the possible cytological features involved in maintenance and evolution of the haplotype specificity of the S loci involved in both self and non-self recognition.

Results

Fine detection of the S-locus chromosomal location in Antirrhinum

The S locus was previously mapped on the smallest chromosome of A. hispanicum by mitotic metaphase chromosome FISH (Ma et al., 2003). For fine mapping of the S locus using a cytological approach, TAC clones (Zhou et al., 2003) corresponding to S-RNase and SLF genes were used for pachytene chromosome FISH. The clone’s names and their corresponding genes are listed in Table 1. We first examined the S2 haplotype. The two genes, S2-RNase and AhSLF-S2, are separated by 9 kbp (revealed by sequence analysis) (Lai et al., 2002), which was too close to distinguish by pachytene chromosome FISH. S2SLF-TAC and S2RNaseR-TAC were used to determine the location and orientation of S2-RNase and AhSLF-S2 on the S2S4 chromosomes (the correlation of S2-RNase and AhSLF-S2 with the TACs of the S2 haplotype is illustrated in Figure 3). A centromeric repeat sequence, CentA1, was used to mark the centromere position and also to help identify the individual chromosomes (Zhang et al., 2005). To separate the closely linked S2RNaseR-TAC and S2SLF-TAC, we probed early pachytene chromosomes (Figure 1). The FISH results showed that these two clones are indeed very close to each other, and located at a distal position on the short arm of chromosome 8, two-thirds of the arm length away from the centromere. It was also found that the S2RNaseR-TAC, representing the S2-RNase, was closer to the centromeric region based on analysis of more than 20 pachytene cells. Signals of the S-locus-derived TAC clones were not detected in the centromeric region defined by CentA1. These results show that the S locus is located away from the centromere. This is quite different from the centromeric localization of some solanaceous S loci observed in metaphase chromosome FISH analyses (Entani et al., 1999; Golz et al., 2001; ten Hoopen et al., 1998).

Table 1. S-locus TAC clones used in FISH analyses of A. hispanicum
TAC clonesS-RnaseSLFOrigin
  1. Absence (−) and presence (+) of the S-locus genes.

S2RNaseR-TACThis study
S4RNase-TAC+This study
S5RNase-TAC+Zhou et al. (2003)
S2SLF-TAC+This study
S4SLF-TAC+This study
S5SLF-TAC+This study
Figure 3.

 Estimation of the physical distances between S-RNase and SLF in three S haplotypes of A. hispanicum.
The top panel shows the representative results for FISH on the extended DNA fibers of the three S haplotypes of A. hispanicum using TAC probes as follows: (S2) S2RNaseR-TAC (green) and S2SLF-TAC (red); (S4) S4RNase-TAC (green) and S4SLF-TAC (red); (S5) S5RNase-TAC (green) and S5SLF-TAC (red). The bottom panel shows a schematic representation of the physical distances between the TAC clones in three S haplotypes. The S2-RNase (blue circle, not present in S2RNaseR-TAC) and AhSLF-S2 (blue rectangle, present in S2SLF-TAC) are shown. The 26 kbp distance between S2RNaseR-TAC and S2SLF-TAC was used as a control distance. The estimated maximum physical distances between S4RNase-TAC and S4SLF-TAC and between S5RNase-TAC and S5SLF-TAC were deduced from the fiber FISH result.

Figure 1.

 Sub-chromosomal localization of the S locus in A. hispanicum.
(a) Biotin-labeled S2SLF-TAC containing AhSLF-S2 (red signal) and digoxigenin-labeled S2RNaseR-TAC containing S2-RNase (green signal) were hybridized to early pachytene chromosomes of the S2S4 haplotype of Antirrhinum hispanicum. The gap between these two TAC clones was 26 kbp (Zhou et al., 2003). A centromeric tandem repeat sequence, CentA1, was biotin-labeled to assign the chromosomes, except for the signal indicating S2SLF-TAC (red signals, see also Zhang et al., 2005). The image of the short arm of chromosome 8 is enlarged and shown in the upper right corner. Scale bar = 5 μm.
(b) Schematic representation of the sub-chromosomal location of the TAC clones derived from the S locus.

To further examine the proposed centromeric localization of the solanaceous S locus, digoxigenin-labeled PhS3-RNase-TAC was hybridized to pachytene chromosomes of the S3S3 haplotype of Petunia hybrida . The nucleolus organizer region (NOR) sequence was biotin-labeled to assign this NOR region to chromosome 3 (Gerlach and Bedbrook, 1979). The major constriction of this chromosome is lightly stained with DAPI. The CP100-TAC containing CP100, which is closely linked to the S-RNase in the Solanaceae (Gebhardt et al., 1991; ten Hoopen et al., 1998; Harbord et al., 2000; McCubbin et al., 2000; Golz et al., 2001), was also selected and used as a biotin-labeled FISH marker (red signal,). The results show that the S locus of P. hybrida is localized on the short arm of chromosome 3, just on the border of major constriction. This is consistent with its centromeric localization proposed previously (Entani et al., 1999; ten Hoopen et al., 1998). However, although linked to the PhS3-RNase, the CP100-TAC appeared distant from the centromere, which is not consistent with the results found in Nicotiana alata (Golz et al., 2001), which is a relative but belongs to another genus of the Solanaceae. Unfortunately, the TAC containing PhS3-SLF showed too much non-specific hybridization signal when used as a FISH probe (data not shown), and the relative positions of the PhS3-RNase and PhS3-SLF could not be determined. Nevertheless, these results indicate that the relative positions of the S locus and the centromere appeared to be different between Antirrhinum and the solanaceaous species.

To further confirm these results in Antirrhinum, the S4RNase-TAC and S4SLF-TAC and the S5RNase-TAC and S5SLF-TAC from A. hispanicum were probed to the S4S5 or S1S5 chromosomes, respectively (data not shown). They produced two doublet signals with similar distance and orientation to those of the S2RNaseR-TAC and S2SLF-TAC on the S2S4 chromosomes. These results indicated that the S-RNase and SLF genes were relatively close to each other on the short arm of chromosome 8 in Antirrhinum.

Determination of the S-haplotype specificity of the TAC clones derived from the S locus

To examine the S-haplotype specificity of S2, S4 and S5 TAC clones, we probed them to the leptotene chromosomes before synapsis. The S2RNaseR-TAC and S2SLF-TAC were labeled with digoxigenin and biotin, respectively, and hybridized together to S2S4 and S2S5 leptotene chromosomes (Figure 2). On the S2S4 chromosomes, they showed two pairs of signals. But on the S2S5 chromosomes, only one pair of signals could be detected. The signals of the S2RNaseR-TAC and S2SLF-TAC on leptotene chromosomes could not be separated as clearly as those on the early pachytene chromosomes. The chromatin of this region was more compact at the leptotene stage than at the pachytene stage, indicating heterochromatic characteristics.

Figure 2.

 Determination of the haplotype specificity of the TAC clones from the S locus of A. hispanicum.
(a) Digoxigenin-labeled S2RNaseR-TAC (green signal) and biotin-labeled S2SLF-TAC (red signal) were used to probe S2S4 leptotene chromosomes. S2RNaseR-TAC and S2SLF-TAC show two pairs of signals here. Enlarged images of the hybridization results are shown on the right: (1) S2RNaseR-TAC (green); (2) S2SLF-TAC (red); (3) merged S2RNaseR-TAC (green) and S2SLF-TAC (red). Digoxigenin-labeled S4RNase-TAC containing SRNase-S4 (green signal) and biotin-labeled S4SLF-TAC containing AhSLF-S4 (red signal) were used to probe S2S4 leptotene chromosomes: (4) S4RNase-TAC (green); (5) S4SLF-TAC (red); (6) merged S4RNase-TAC (green) and S4SLF-TAC (red). These show two pairs of signals at the same positions as S2RNaseR-TAC and S2SLF-TAC (3). The red and green signals are overlapping, indicating the proximity of S4RNase-TAC and S4SLF-TAC.
(b) S2RNaseR-TAC and S2SLF-TAC were used to probe S2S5 leptotene chromosomes, and show only one pair of signals. Enlarged images of the hybridization results are shown on the right: (7) S2RNaseR-TAC (green); (8) S2SLF-TAC (red); (9) merged S2RNaseR-TAC (green) and S2SLF-TAC (red). Digoxigenin-labeled S5RNase-TAC containing SRNase-S5 (green signal) and biotin-labeled S5SLF-TAC containing AhSLF-S5 (red signal) were used to probe S2S5 leptotene chromosomes: (10) S5RNase-TAC (green); (11) S5SLF-TAC (red); (12) merged S5RNase-TAC (green) and S5SLF-TAC (red). They show only one pair of signals, which localizes to positions different from those of S2SLF-TAC and S2RNaseR-TAC (9). The distance between S5RNase-TAC and S5SLF-TAC was estimated to be larger than that of S2RNaseR-TAC and S2SLF-TAC. Scale bar = 5 μm in (a) and (b), and 1 μm in (1)–(12).

The S4RNase-TAC and S4SLF-TAC and the S5RNase-TAC and S5SLF-TAC of A. hispanicum were also probed to S2S4 and S2S5 leptotene chromosomes, respectively. On the S2S4 chromosomes, the S4RNase-TAC and S4SLF-TAC showed two pairs of signals at the same positions as those of the S2SLF-TAC and S2RNaseR-TAC, and the red and green signals always overlapped, showing the proximity of the S4RNase-TAC and S4SLF-TAC (Figure 2a). On S2S5 leptotene chromosomes, the S5RNase-TAC and S5SLF-TAC showed only one pair of signals located at positions different from those of S2SLF-TAC and S2RNaseR-TAC (Figure 2b), and there was a small distance between the S5RNase-TAC and the S5SLF-TAC. The physical distance between the S5RNase-TAC and S5SLF-TAC was estimated to be larger than that between the S2SLF-TAC and S2RNaseR-TAC. These results indicate that the leptotene FISH hybridization signals of S2 and S4 haplotypes cross-hybridized more with each other without obvious S-haplotype specificity, whereas those of the S2 and S5 haplotypes showed strong S-haplotype specificity with only faint cross-hybridization.

Estimation of the physical distance between S-RNase and AhSLF in the S4 and S5 haplotypes of A. hispanicum

To measure the distance between S4RNase-TAC and S4SLF-TAC, and between S5RNase-TAC and S5SLF-TAC, on the chromatin of their corresponding S haplotypes, we hybridized them together to the DNA fiber generated from S2S5 interphase cells (Figure 3). The S2RNaseR-TAC and S2SLF-TAC were hybridized to DNA fibers of the S2 haplotype as a control, where they are known to be separated by 26 kbp. The signals for the S2SLF-TAC and S2RNaseR-TAC unambiguously showed their actual size, and the gap between the two clones showed very low cross-hybridization. However, the signals between S4RNase-TAC and S4SLF-TAC or between S5RNase-TAC and S5SLF-TAC showed considerable cross-hybridization and could not be separated easily, indicating that they contained much more repetitive sequence than the TAC clones from the S2 haplotype. The S4 haplotype had mostly green dots at one end and red dots at the other. These represent the S4RNase-TAC and S4SLF-TAC, respectively. The dots between these TACs were probably caused by shared repetitive sequences located on both sides. The distance between the S4RNase-TAC and S4SLF-TAC could be deduced from the total length of the fiber FISH signals minus the length of the S4RNase-TAC and S4SLF-TAC, and was estimated to be approximately 50 kbp. As for the S5 haplotype, it showed scattered green and red dots, with one end biased toward green, the other biased toward red. The starts or ends of S5RNase-TAC and S5SLF-TAC could not be delimited. Thus, only a maximum distance between the S5RNase-TAC and S5SLF-TAC could be deduced from the total length of the fiber FISH signals minus the length of the S5RNase-TAC and S5SLF-TAC, and was estimated to be approximately 100 kbp. These results show that the S-RNase and AhSLF genes in the S haplotypes are closely linked to each other, with variable abundances of repetitive sequences, and their physical separation ranged from 9 to approximately 100 kbp.

An inversion was detected between self-compatible A. majus and self-incompatible A. hispanicum around the S locus

Antirrhinum majus is a self-compatible (SC) species in the Antirrhinum genus. To investigate the cytological features of the S locus in A. majus, we first cloned AmSLF-like 1, similar to AhSLF, from A. majus based on amino acid sequence homology. It diverged earlier than AhSLF-S1, S2, S4 or S5, and shared 95% identity with the AhSLFs at the amino acid level. The expression pattern of AmSLF-like 1 was determined by RT-PCR and Western blot analyses (data not shown) and was found to be similar to that of AhSLF (Lai et al., 2002; Zhou et al., 2003), suggesting that AmSLF-like 1 is specifically expressed in pollen. However, we were unable to obtain an S-RNase sequence from A. majus by a similar approach. Thus, it is unclear whether a functional S-RNase gene is present in A. majus.

To compare the S-locus regions in the SI and SC Antirrhinum, the TAC clones from A. hispanicum were probed to the chromosomes of A. majus. The S2RNaseR-TAC and S2SLF-TAC were selected to probe early pachytene chromosomes of A. majus. Surprisingly, although the two TAC clones were located in a similar region of chromosome 8, they had an inverted orientation compared with that on the chromosome of A. hispanicum (Figure 4a).

Figure 4.

 The S-locus region is inverted in A. majus relative to A. hispanicum.
(a) In A. majus, digoxigenin-labeled S2RNaseR-TAC (green signal) and biotin-labeled S2SLF-TAC (red signal) probes localize to a similar S-locus region containing S2RNaseR-TAC and S2SLF-TAC in A. hispanicum, with S2SLF-TAC farther from the telomere.
(b) TACs containing CYC and RAD were used to determine the inversion boundary. Biotin-labeled RAD-TAC (red signal) and digoxigenin-labeled CYC-TAC (green signal) were used for A. majus. CYC-TAC showed two hybridization signals, but only the one at the S-locus region represents CYC, and is located closer to the telomere.
(c) Digoxigenin-labeled S2RNase-TAC, which is located between S2RNaseR-TAC and S2SLF-TAC, and contains SRNase-S2, was used to identify the S-locus region (green signal). In A. majus, biotin-labeled RAD-TAC (red signal) localized closer to the telomere.
(d) In A. hispanicum, digoxigenin-labeled S2RNase-TAC (green signal) localized closer to the telomere than biotin-labeled RAD-TAC (red signal).
(e) CYC-TAC was biotin-labeled (red signal) for A. hispanicum, and localized closer to the telomere than digoxigenin-labeled S2RNase-TAC (green signal).
(f) Schematic representation of the TAC clones around the S-locus region in SI A. hispanicum and SC A. majus. The arrows indicate the orientation and extension of the inverted fragments.
*These TACs isolated from SI Antirrhinum hybridized to the chromosomes of SC A. majus. The arrangements of the three signals on the chromosomes of SC (b, c) and SI (d, e) Antirrhinum indicate that the chromosome fragments containing RAD and the S locus were inverted relative to each other. The yellow arrowheads indicate the end of the short arm of chromosome 8. Scale bar = 5 μm in (a), and 1 μm in (b)–(e).

S-locus-linked TAC clones containing CYC and RAD markers (Schwarz-Sommer et al., 2003; Zhang et al., 2005) were used to determine the inversion boundary. The CYC-TAC showed two hybridization signals, one in the S-locus region and the other in the long arm of chromosome 8. Based on genetic linkage evidence (Schwarz-Sommer et al., 2003), the short-arm signal most likely contained the CYC gene, whereas the long-arm signal probably did not. The S2RNase-TAC located between the S2RNaseR-TAC and S2SLF-TAC, containing S2-RNase, was used to designate the S locus. The order of the three signals on the chromosomes of A. majus (Figure 4b,c) was different from that of A. hispanicum (Figure 4d,e). The fact that the fully sequenced S-locus TAC clones did not contain CYC and RAD (Zhou et al., 2003) demonstrated that CYC-TAC and RAD-TAC were outside the locus. The physical distance between the CYC-TAC and RAD-TAC was 0.41 ± 0.04% of the total length of chromosome 8 in A. majus, and 2.4 ± 0.2% in A. hispanicum, based on measurement of three different chromosome samples. These results show that the chromosome fragments containing RAD and the S locus are inverted in relation to each other in A. majus and A. hispanicum (Figure 4f).

The S locus is localized to a highly heterochromatic region

When pachytene chromosomes were stained with DAPI, the brightly stained regions correspond to the heterochromatic domains, and were highly consistent among different cells (Zhang et al., 2005). The DAPI staining pattern at pachytene stage showed that there were five heterochromatic domains on the short arm of chromosome 8 (Zhang et al., 2005). If numbered from first to fifth from the short-arm end, the S-locus signal was located in the 2nd heterochromatic domain (Figure 5a,b). This was apparent in the relatively lightly stained distal domains where the S-locus region could be easily identified on the DAPI-stained chromosomes even without the S-locus FISH markers.

Figure 5.

 The S locus is located in a highly heterochromatic region.
(a) At the pachytene stage, the S-locus signal (digoxigenin-labeled S2SLF-TAC, green) was located in the second heterochromatic domain of chromosome 8. The centromere was probed using the centromeric repeat sequence CentA1 (red signal). Scale bar = 5 μm.
(b) The DAPI-stained chromosome 8 shown in (a) was converted to a black-and-white image to enhance the visibility of the distribution of euchromatin and heterochromatin. An ideogram of the relationship of the S locus and DAPI-bright regions is shown on the right.
(c) Black-and-white image of the DAPI staining pattern corresponding to Figure 2(1–6). The arrow and arrowhead indicate the S-locus position for the two S haplotypes, S2 and S4, without discriminating the S haplotypes. Both the arrow and arrowhead show the border of the heavily and lightly stained regions.
(d) Black-and-white image of the DAPI staining pattern corresponding to Figure 2(7–12). The arrow and arrowhead indicate the S-locus position for the S2 and S5 haplotypes respectively. As in S2S4 chromosomes, both the arrow and arrowhead show the border of the heavily and lightly stained regions in the S2S5 chromosomes. In the S2, S4 and S5 haplotypes analyzed, the S locus is located at the border of the heterochromatin and euchromatin of the various leptotene S haplotypes.

On the more stretched leptotene chromosomes, the S-locus-derived TACs were found to occupy the borderline of the heterochromatin and euchromatin of both S2S4 and S2S5 chromosomes (Figure 5c,d), indicating that the S locus is located in a less condensed region of the second heterochromatic domain.

As the S locus is located in a highly heterochromatic region in the different Antirrhinum species, the DNA elements might be in a highly methylated state around this region. To investigate this, we selected several repetitive DNA elements from the S-locus region and further checked their methylation status. Transposon Tam3-like and retrotransposon copia-like were identified in the S locus by comparing the S-locus DNA sequence with that in the NCBI database (Zhou et al., 2003). We examined the methylation status of these transposable elements (TEs) to investigate epigenetic modification in the S locus. Tam3-like and copia-like were used to probe genomic DNA digested by HapII/MspI (isoschizomers with different DNA methylation sensitivity). Most of the Tam3-like and copia-like TE copies were heavily methylated (Supplementary, consistent with the usual features of TE (Lippman et al., 2004).

To examine the methylation status of coding genes in the S locus, we used AhSLF as a representative. As all the 5′ UTR regions of AhSLF genes share the same restriction sites, the 5′ UTR region of AhSLF-S2 was chosen to probe genomic DNA digested by HapII/MspI and BstNI/PspGI (Supplementary). The size of resultant bands showed that all the restriction sites in the AhSLF sequence were fully digested, suggesting that AhSLF lacks DNA methylation. We also tested other S-locus genes, the S-RNase genes, but the probe used appeared to contain unknown repetitive sequences that generated smeared Southern hybridization signals that could not be analyzed (data not shown).

To examine whether small RNA is involved in epigenetic modification of the S locus, Northern blot hybridization was conducted. It showed that the Tam3-like and copia-like had their corresponding 25 nt small interfering RNAs without tissue specificity, and these small RNAs were probably involved in regulation of this region.

For AhSLF-S2, no small RNA corresponding to the AhSLF-S2 promoter was detected, and no microRNA or small interfering RNA could be detected corresponding to the known S-locus sequences in small RNA databases such as the NCBI microRNA registry (http://www.sanger.ac.uk/Software/Rfam/mirna) and the siRNA database (http://sirna.cgb.ki.se/). These results indicate that the S-locus coding genes are probably not regulated by small RNA.

To investigate whether histone modification is associated with the S-locus gene, we performed a ChIP (chromatin immuno-precipitation) experiment with anti-H3 K4me2 antibody using nuclei isolated from leaf tissues. Mock experiments using pre-immunized rabbit serum served as the non-specific binding control. Precipitated DNA was analyzed by semi-quantitative PCR. ChIP-PCR analyses of Actin and Tam3-like sequences were included as positive and negative controls, respectively. Primers for the promoter region of AhSLF, S2-RNase and S4-RNase were designed to produce fragments of 332, 179 and 332 bp, respectively. It was observed that the promoters of these genes had an H3 K4me2 positive modification. Because the H3 K4me2 modification is generally associated with euchromatin (Sims et al., 2003), this suggests that the chromatin regions of these genes retain euchromatic status in the heterochromatic region in the leaves despite being pollen- or stylar-specific genes.

Discussion

The S-loci are not always located in the centromeric region among different S-RNase-based SI families

Our results have shown that the S locus of A. hispanicum is located outside the centromeric region on the short arm of chromosome 8 despite the fact that S loci are believed to lie in the centromeric region in several other S-RNase-based SI species (Bernacchi and Tanksley, 1997; Bernatzky, 1993; Brewbaker and Natarajan, 1960; Entani et al., 1999; Golz et al., 2001; ten Hoopen et al., 1998; Pandy, 1965). The S locus of P. hybrida was confirmed to be located in the centromeric region using pachytene chromosome FISH (Entani et al., 1999; ten Hoopen et al., 1998). These findings suggest that the relative positions of the S locus and the centromere differ between the Solanaceae and Scrophulariaceae, although both possess the S-RNase-based SI system. Moreover, the relative positions of the pollen and stylar S components and the centromere appear to be different between N. alata and A. hispanicum, but similar between N. alata and self-compatible A. majus, despite its self-compatibility. In P. hybrida, the relative positions of the PhS3-RNase and PhS3-SLF were not resolved. In N. alata, the pollen S component was postulated to be close to the centromere based on frequent associations of the pollen-part mutations (PPMs) and centromeric fragments (Golz et al., 2001), which appears to be quite common for several other S-RNase-based self-incompatible species (Brewbaker and Natarajan, 1960).

The S locus of N. alata and P. hybrida resides in the centromeric region (Entani et al., 1999; Golz et al., 2001; ten Hoopen et al., 1998), indicating that it is prone to breaking and duplicating with the centromere to produce a centromeric fragment, in which the functional centromere guarantees the stable inheritance of the chromosome fragment. The PPMs of these species were mostly generated by duplication of a small centromeric fragment containing the pollen S gene (Brewbaker and Natarajan, 1960; Golz et al., 1999, 2001). In addition, the relatively short distance between the S locus and the centromere in Antirrhinum suggests that it would also be possible to generate PPM by an additional centromeric fragment in this species. Consistently, we have found a trisomic PPM in A. hispanicum (unpublished data). Thus, it is likely that the frequent generation of centromeric fragments in the PPMs is based on close linkage of the pollen S gene and the centromere.

Rosaceae apparently have the same SI mechanism as members of the Solanaceae and Scrophulariaceae (Kao and Tsukamoto, 2004). However, recently, some differences have been noted (Sonneveld et al., 2005; Ushijima et al., 2004; Yamane et al., 2003). Although the precise chromosomal localization and chromatin structure of the locus is not clear in the Rosacecae, it is rich in repetitive sequences (Ushijima et al., 2003) and was recently reported to be of simple genomic structure and suggested to reside outside the centromeric region (Ushijima et al., 2001). As no PPM containing the centromeric fragment has ever been described in the Rosaceae, its S locus might be located distant from the centromere, and not prone to being duplicated on a stably inherited centromeric fragment. Thus, it could be deduced that the chromosomal localization of the S locus appears to be different in the Solanaceae, Scrophulariaceae and Rosaceae, even though they share similar S-RNase-based SI systems.

Interestingly, the chromosome regions containing the S locus are inverted between A. majus and A. hispanicum. This inversion is seen for all three S haplotypes of A. hispanicum studied here. In general, SI is thought to be an ancestral state (Igic et al., 2003). Although the genus Antirrhinum contains both SI and SC species, it is not clear how the SC type evolved from the SI type. Nevertheless, several scenarios including deletion/point mutations in the S gene and mutations outside the S gene (modifiers) have been described to account for the origin of SC in SI species (Hancock et al., 2005; Huang et al., 1994; Kao and Tsukamoto, 2004; Kondo et al., 2002; Kowyama et al., 1994; Royo et al., 1994). It is not clear whether the inversion could be associated with self-compatibility of A. majus. Further examination of more SC and SI species in Antirrhinum would help to clarify this issue.

The S locus is located in a highly heterochromatic region in Antirrhinum

Our findings have revealed that the S locus of Antirrhinum is located in a pronounced heterochromatic region. The heterochromatic localization of the Antirrhinum S locus is consistent with the fact that it contains abundant repetitive sequences, a feature shared with other S-RNase-based self-incompatible species in the Solanaceae (Coleman and Kao, 1992; Entani et al., 1999; Matton et al., 1995). In Antirrhinum, many retro-elements or transposons were identified in the sequenced S-locus-derived TACs, and account for most of the predicted genes of the locus (Zhou et al., 2003).

The possible centromeric localization and the presence of repetitive sequences of the solanaceous S locus suggest that it probably resides in heterochromatin. Although not localized to the centromeric region, the S locus of Antirrhinum has a similar position in a condensed chromatin environment. The S loci of other SI systems also contain various transposable elements, such as those found in the Brassicaceae and Papaveraceae (Cui et al., 1999; Wheeler et al., 2003), implying that they are also likely to be heterochromatic. Localization in a heterochromatic environment of low recombination is common to the S locus of most, if not all, SI species, providing a genetic mechanism for recombination suppression.

Haplotype polymorphisms of the S-locus

The differences in length and gene arrangements between S haplotypes in Antirrhinum are still unclear. In estimating the sizes of S haplotypes, we found that, although tightly genetically linked, the physical distances between the S-RNase and AhSLF of respective S haplotypes vary widely. The distance between the Antirrhinum S2-RNase and AhSLF-S2 is only approximately 9 kbp, with only one transposon-like sequence predicted in this region (Lai et al., 2002). By contrast, several large insertions were predicted to have occurred between the S-RNase and AhSLF in the S4 and S5 haplotypes compared with the S2 haplotype (Zhou et al., 2003), and the gaps between the S-RNase and AhSLF for these two haplotypes are much longer and estimated to be approximately 50 and 100 kbp, respectively (Figure 3). In addition, these regions are not as simple as that between the S2-RNase and AhSLF-S2, and are rich in retroelements and transposons, as revealed by both fiber FISH analyses in this study and DNA sequence analyses by Zhou et al. (2003). The length of this region of the S5 haplotype was almost double that of the S4 haplotype, and was rich in repetitive sequences as deduced from the more mixed signals beyond the TACs in our fiber FISH analyses. Consistent with this, the leptotene chromosome FISH result clearly displayed distinct hybridization signals for the different haplotypes, indicating different sequence compositions (Figure 2). The regions between the S-RNase and AhSLF genes have not been fully sequenced in the S4 and S5 haplotypes, so the exact repetitive element organization in these regions is still unknown. Although it was noted that the density of repetitive elements is low in the 40 kbp region containing PhS3-RNase compared with other regions of the S locus in P. hybrida (our unpublished data), it is not clear whether the distribution of repetitive elements is unique to P. hybrida. However, owing to the duplications, insertions and deletions (Zhou et al., 2003), repetitive element compositions and arrangements appear not to be uniform among different S haplotypes, despite their similarities in Antirrhinum. This feature appears to be present widely in other SI species (Cui et al., 1999; Suzuki et al., 1999; Ushijima et al., 1998; Wheeler et al., 2003).

Haplotype polymorphisms of the S locus have been identified previously in Brassica (Cui et al., 1999; Shiba et al., 2003) and Prunus (Entani et al., 2003; Ushijima et al., 2001, 2003). Despite a rather smaller and simpler S-locus region, the Rosaceae displayed a similar variation in S-haplotype genomic structure. The rosaceous S-locus region not only showed S-haplotype sequence diversity, but also varied greatly in the extent of its S-locus region between different S haplotypes (Entani et al., 2003; Ushijima et al., 2001, 2003). Aside from the species possessing GSI systems, the S locus of Brassica, which encodes a sporophytic SI (SSI) system, also displayed S-haplotype polymorphisms (Cui et al., 1999; Shiba et al., 2003). Taken together, genomic haplotype polymorphisms are characteristic for the S loci.

Possible mechanisms for maintaining S-haplotype structural diversity

The S loci known in eukaryotic organisms might share some similar mechanisms to maintain their haplotype structures during evolution. It has been thoroughly investigated that large genomic structural differences, including repetitive sequences, haplotype-specific intergenic sequences and gene arrangements between homologous chromosomes, may contribute to the recombination suppression that is intrinsic to recognition loci (May and Matzke, 1995; O’hUigin, 1995; Nasrallah, 2002).

The well-studied plant disease resistance (R) locus shares several features with S loci, such as clustered duplicated genes and repetitive sequences (Meyers et al., 1998; Wei et al., 2002). The Mi gene in tomato is localized at the border of the heterochromatin (Zhong et al., 1999). Furthermore, the mating-type locus in yeast is heterochromatic (reviewed by Haber, 1998), and the major histocompatibility complex (MHC) locus in mammals is probably heterochromatic, with repetitive sequences and low recombination levels (Singer et al., 1983; Walsh et al., 2003), although no direct evidence exists as yet. The large heterochromatic region could block recombination structurally within the centromere, despite the coding regions that are present within it (Saffery et al., 2003; Yan et al., 2005).

In conclusion, sequence polymorphisms and the highly condensed and extensive heterochromatic region are always associated with the S locus in Antirrhinum. These features appear to be shared by the S loci of other SI systems that possess two S determinants, and perhaps are common to recognition loci in yeast, animal and plant organisms. They could also contribute to the regional recombination suppression, retention of point mutations and accumulation of retro-element insertions that are required for maintaining recognition specificity while promoting diversification.

Experimental procedures

Plant materials

A. majus (stock 75), A. hispanicum and P. hybrida plants were grown in a greenhouse environment as previously described (Lai et al., 2002; Robbins et al., 2000; Xue et al., 1996). Young buds were collected for meiotic chromosome preparation. Leaf tissue was harvested for nuclei preparation, genomic DNA and small RNA isolation, and flower tissue was also used for small RNA isolation.

Screening of TAC library

The TAC libraries from A. hispanicum and P. hybrida have been described previously (Qiao et al., 2004a; Zhou et al., 2003). For the S1S5 library, clones of 384-well plates were imprinted onto a 15 cm plate using a VP384 pin replicator (V&P Scientific, http://www.vp-scientific.com) and inoculated onto LB agar medium containing kanamycin (25 mg l−1). After incubation at 37°C overnight, bacteria were collected for plasmid preparation. Plasmid DNA from ten 384-well plates was mixed as a pool for PCR screening. The TAC library was screened with primers specific for marker genes. When a specific PCR product was detected in one or more pools, the ten 384-well plates of the positive pool were individually re-screened with the primer pair, and positive 384-well plates were identified. Finally, the positive clone was identified by PCR screening in a row and column combination.

Chromosome preparation

Immature Antirrhinum flower buds (length 1.5–3.0 mm) were harvested and fixed in Carnoy’s solution (ethanol/glacial acetic acid 3:1). Microsporocytes at meiosis stage were squashed in an acetocarmine solution according to the method described by Wu (1967). Slides were frozen in liquid nitrogen. After coverslip removal, slides were dehydrated through an ethanol series (70%, 90%, and 100%) prior to use in FISH.

Chromosome fluorescence in situ hybridization

Chromosome FISH and fiber FISH were performed according to published protocols (Jackson et al., 1998; Jiang et al., 1995). BAC/TAC DNA was isolated using a standard alkaline extraction procedure (Sambrook et al., 1989) and labeled with either biotin-11-dUTP or digoxigenin-16-dUTP (Roche, http://www.roche.com) by nick translation. Chromosomes were counterstained with 4’,6-diamidino-phenylindole (DAPI) in an anti-fade solution (Vector Laboratories, http://www.vectorlabs.com). Chromosomes and FISH signal images were captured with an Olympus BX61 fluorescence microscope (http://www.olympus-global.com/) coupled to an Apogee KX85 CCD camera. Grey-scale images were captured for each color channel, and then merged using Image-Pro Plus (IPP) software (Media Cybernetics, http://www.mediacy.com). Pachytene chromosome lengths were measured using IPP software.

Acknowledgements

We thank Drs Enrico Coen and Rosemary Carpenter of John Innes Center, UK, for providing Antirrhinum plants and Dr Tim Robbins of University of Nottingham, UK, for P. hybrida. We are also grateful to Drs Andy McCubbin of Washington State University, USA, and Tim Robbins as well as the two anonymous reviewers for their careful reading and critical comments. This work was supported by the Chinese Academy of Sciences and the National Natural Science Foundation of China (30221002).

Ancillary